PRILOGE PRILOGA A Pregled raziskav na področju 3D katastra / A review of research on 3D real property cadastre Tekavec, J., Ferlan, M., Lisec, A. (2018) Geodetski vestnik, 62(2): 249-278 DOI: 10.15292/geodetski-vestnik.2018.02.249-278 Faktor vpliva za leto 2019: 0,469 (Q4) G 8012 GEODETSKI VESTNIK | letn. / Vol. 62 | št. / No. 2 | | 62/2 | V PREGLED RAZISKAV NA A REVIEW Of RESEARCH TICLES PODROČJU 3D-KATASTRA ON 3D REAL PROPERTY WED ARVIE NEPREMIČNIN CADASTRE ANI ČLANKI | PEER-RE RECENZIR Jernej Tekavec, Miran Ferlan, Anka Lisec UDK: 347.2:528.44 DOI: 10.15292/geodetski-vestnik.2018.02.249-278 Klasifikacija prispevka po COBISS.SI: 1.02 REVIEw ARTICLE Prispelo: 3. 3. 2018 Received: 3. 3. 2018 Sprejeto: 7. 5. 2018 Accepted: 7. 5. 2018 IZVLEČEK ABSTRACT SI | EN 3D-kataster nepremičnin (krajše 3D-kataster) je eno izmed The 3D real estate cadastre (‘3D cadastre’ in short) is an pomembnih interdisciplinarnih raziskovalnih področij, na important interdisciplinary research topic at both the evropski in širši mednarodni ravni. Prve teoretične razprave European and international levels. Initial theoretical segajo v 90. leta preteklega stoletja in se v začetku novega scientific discussions on the 3D cadastre began in the 1990s tisočletja še krepijo. Takrat so se oblikovali prvi mednarodni and gained momentum at the turn of the millennium, forumi za spodbujanje raziskav in oblikovanje teoretičnih when the first international forums were organised. Their zasnov 3D-katastra nepremičnin, ki jih je mogoče uveljaviti principal aim was to develop the theoretical concepts for the v praksi. Za spodbujanje sodelovanja med različnimi 3D cadastre that would foster the research activities and their raziskovalnimi pobudami na širokem področju 3D-katastra implementation. At the time, the FIG Working Group on je bila pri mednarodnem združenju FIG oblikovana delovna 3D Cadastres was formed to connect the research activities in skupina za 3D-katastre. Pod okriljem FIG-a se je do danes the field. To date, five international thematic workshops have zvrstilo pet tematskih delavnic oziroma mednarodnih forumov. been organised. This article aims to provide a chronological Namen prispevka je podati kronološki pregled raziskav na overview of research activities by highlighting publications področju 3D-katastrov s predstavitvijo objav, ki so pomembno that have had a significant impact on 3D cadastre research. usmerjale razvoj tega področja. Glavni vir so bile objave na Our main sources have been publications at the FIG mednarodnih forumih organizacije FIG, doktorske disertacije international thematic forums, doctoral dissertations, in znanstvene objave v mednarodnih revijah, katerih članki so and papers published in scientific journals (included in opremljeni z digitalnim identifikatorjem DOI. Z raziskavami the DOI system). Many issues and challenges have been so bila v dveh desetletjih rešena številna pereča vprašanja, a resolved, and major progress has been seen in the past two obenem se pojavljajo nova, kompleksnejša vprašanja, predvsem decades. Nevertheless, numerous new complex issues have povezana z uveljavitvijo 3D-konceptov v katastrskih sistemih arisen, particularly regarding the realisation of 3D concepts posameznih držav, z večnamensko uporabo vsebinsko bogatih within cadastral systems in the various countries, the idea of in kakovostnih 3D-katastrskih podatkovnih zbirk ter z a multipurpose 3D cadastre, and the integration of various združevanjem različnih prostorskih podatkovnih nizov v spatial datasets within a 3D cadastre. okviru 3D-katastra. KLJUČNE BESEDE KEY WORDS kataster, nepremičnine, 3D-kataster, zemljiška administracija, cadastre, real property, 3D cadastre, land administration, LADM LADM Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 249 | | 62/2 | GEODETSKI VESTNIK 1 inTroduCTion TICLES Urban development coupled with increasingly complex cases of spatial delineation in terms of ownership WED AR and other property rights requires a new approach in land administration, which allows for registering and VIE changing of property units, and associated information, in three spatial dimensions. The increasing physical and legal complexity of the built as well as natural environment necessitate an upgrading of the two-dimensional spatial modelling approach, which is conventionally used in national land administration systems. ANI ČLANKI | PEER-RE The land administration domain has always been highly demanding in research terms at the international RECENZIR level, as countries developed their own systems underpinned by their historical background concerning land administration, their legal system, social setting as well as social needs associated with spatial development (Zupan et al., 2014). The requirement for international comparability and thus structured treatment of land, and the rights, restrictions, and responsibilities associated to it, stemmed from the growing needs to develop state-of-the-art solutions in land administration. A result of many international discussions in the field was the international ISO standard 19152:2012: Land Administration Domain Model (LADM), adopted in 2012. The purpose of this paper is to provide a broad review of internationally recognised publications and thus present the evolution of the 3D property cadastre over recent decades. Based on these publications we analysed the topics that are currently the focus of international research undertaken in this field. ENSI | 2 MeTHodoloGy and reSourCeS aPPlied Researchers and developers in various fields are concerned with developing 3D cadastre concepts as well as technical and legal solutions for its implementation. They all focus on a very specific domain, i.e. real property cadastre, which thus brings together the latest progress achieved worldwide. The main resources for this study were the available resources and records of the Thematic Working Group on 3D Cadastres, established by the International Federation of Surveyors (FIG)1, which back in 2001 organised its first international forum with a view to help to develop solutions in the 3D property cadastre. An important resource for our work was publications in two special issues of the international scientific journal Computers, Environment and Urban Planning from 2003 and 2013, respectively, where also an overview of discussions under FIG until 2012 (Oosterom, 2013) was published, and publications in ISPRS International Journal of Geo-Information with a special issue Research and Development Progress in 3D cadastral systems of 2017. Additionally, we reviewed relevant PhD researches and English papers that appeared in other international journals published with the well-established Digital Object Identifier (DOI). The CrossRef 2reference linking service was used, which is one of the solutions that publishers use to create DOI and include journals into an extremely large international community of electronic scientific and professional publications (see also Koler Povh and Lisec, 2015). 3 reSulTS – overview oF 3d ProPerTy CadaSTre develoPMenT The beginnings of intensified efforts to develop the 3D real property cadastre date back to 1994 when FIG Working Group 7.1 was initiated, which in 1998 published the vision of developing future cadastral 1 Fédération Internationale des Géomètres: www.fig.net. 2 CrossRef: www.crossref.org, last r eviewed 10 January 2018. | 250 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | systems entitled Cadastre 2014 (Kaufmann and Steudtler, 1998). This document underlined the role TICLES of the cadastre as an important stakeholder to support sustainable development and decision-making concerning spatial decisions. The document provides important definitions and, inter alia, instead of a WED ARVIE parcel (parcel-based cadastre), introduces the more general term “(land) object” as the basic real property element to which rights, restrictions, and responsibilities apply. The cadastre should give enough information to provide a complete picture of the situation of land, legal security, and transparency regarding ANI ČLANKI | PEER-RE the rights, restrictions, and responsibilities associated to cadastral objects. The end of separating between descriptive and graphic representations in the cadastre and the introduction of computer modelling, RECENZIR replacing analogue cadastral mapping, were projected, both because of the rapidly developing information technology. These new definitions and orientations encouraged, among other things, the discussion on introducing the third spatial dimension into real property records. 3.1 Studies in 3d property cadastre between 2000 and 2010 The first results of the studies in the research domain of 3D real property cadastres, which strongly affected further international research efforts, were published at the turn of the millennium (Stoter, 2000; Stoter and Zevenbergen, 2001). The authors find that the 2D system to register the legal status of real estate objects in many cases does not provide enough legal security regarding the rights and restrictions on real EN property objects and it can also no longer satisfy other functionalities of the land administration system. SI | 3.1.1 early internationally recognised studies, discussions, and publications The previously mentioned research work at the Delft University of Technology, the Netherlands, was an introduction to the first workshop on the topic of real property cadastres in 2001, organised by FIG, which sparked interest and strengthened the topic of 3D cadastres in the research sphere. The workshop introduced the classification of the research field into legal, technical, and organisational aspects, which have been preserved, almost without modification, to this day. The workshop featured presentations on land administration systems in individual countries, existing ways of solving complex cases of real property registration where there is a need for vertically stratified allocation of rights, and on possibilities for further development (Grinstein, 2001; Huml, 2001; Menda, 2001; Onsrud, 2001; Ossko, 2001; Rokos, 2001; Viitanen, 2001). Among them was the presentation of the then introduced Building Cadastre in Slovenia (Pogorelčnik and Korošec, 2001). The conclusions were drawn by Lemmen and Oosterom (2003) as an introduction to the special issue of the international journal Computers, Environment and Urban Planning, which published selected papers from this workshop. In his work, Molen (2003) argues that changes in complex systems, such as that of the cadastre, always require organisational development of institutional conditions and that they need to follow technological progress. The dilemma, i.e. the difference between legal objects and objects representing physical structures in space, is particularly underlined. Onsrud (2003) presented a new regulation in Norway, which was being adopted at the time, allowing for registering 3D properties to settle rights and restrictions on “construction properties”. The possibility of combining this regulation with the existing one, based on the condominium concept, was presented. Back then, the real property registration system in Norway was still completely based on 2D parcels, partially even in analogue form. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 251 | | 62/2 | GEODETSKI VESTNIK The author saw no real possibility in the near future to technically register legal 3D property units as 3D TICLES objects. A major challenge in 3D cadastres, which remains pertinent today, was how to efficiently model WED AR cadastral data in information systems, as reflected in several papers published in the aforementioned special VIE issue (Billen and Zlatanova, 2003; Stoter and Ploeger, 2003; Tse and Gold, 2003). Tse and Gold (2003) propose using a Triangulated Irregular Network (TIN) to model geometry and topology of 3D cadastral objects, which they justify by the feasibility of the proposed solution. Billen and Zlatanova (2003) also ANI ČLANKI | PEER-RE study how to model spatial objects, with an emphasis on spatial relationships. Stoter and Ploeger (2003) RECENZIR present the ways of developing conventional systems towards a 3D cadastre. The doctoral thesis by Stoter (2004) is the first extensive research work on the 3D cadastre, which comprehensively discusses its technical aspect, while also delving into its legal and organisational aspects. It presents several practical cases in the Netherlands, where the two-dimensional approach in land administration is no longer meeting the demands for transparent real property registration. By analysing the state of land administration internationally and a detailed analysis of selected countries, she found that up to the point no country had developed a system for 3D registration of property units; moreover, it was not actively developed anywhere. She underlined and made clear the rationale to introduce 3D cadastres in selected study cases. She particularly addressed modelling, administration, and presentation of data on real properties in a 3D environment. Furthermore, she discussed the capacity of information ENSI | technology for establishing databases, solutions for 3D geometry storage, procedures for validating the accuracy of the information recorded, and data administration functions of the time. In her research for her doctoral thesis, she developed three cadastral data models. The first one is an upgrading of existing cadastral systems, by storing links to 3D data, which are stored separately. The hybrid model preserves the existing role of the traditional 2D cadastre as the basis onto which rights, restrictions, and responsibilities are bound, but it allows for registration of 3D objects to show more clearly the situation regarding rights and restrictions in space in special cases. The real 3D cadastre allows for registration of volume parcels. 3D parcels assume the role of cadastral objects – the parcels in the cadastre are no longer defined as 2D polygons, but rather as 3D bodies, while in the case of traditional parcels lacking the vertical division they are defined as upright towers, which are not vertically limited. Hence, this is a volumetric division of the entire space with 3D property units. Here, the activities of the United Nations Economic Commission for Europe should be mentioned, which in 2004, to support the development of efficient land administration systems, published guidelines focusing on real property units and object identifiers (United Nations Economic Commission for Europe, 2004), which importantly underlines the role of modern cadastre from the economic viewpoint. This document should help to align terminology and understand the differences between the systems of individual countries, facilitating international collaboration and data exchange in the field. It also touches upon the problem of the third spatial dimension of real property units, the condominium or strata title, and mineral extraction sites. 3.1.2 3d cadastre and data models Rather than introducing the storage of the additional spatial dimension, the introduction of the 3D cadastre entails radical changes of the entire cadastral data model. The start of 3D cadastre development | 252 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | in the late 1990s coincides with the intensive period of introducing computer modelling in land admin-TICLES istration as well. In 2003 the first proposal for a cadastral data model3 was proposed to unify the concepts and data models of national systems (Oosterom and Lemmen, 2003). 3D cadastre is highlighted as a WED ARVIE special case, defined as a possible upgrade or extension of the CCDM, along with the temporal aspect (Oosterom, Lemmen and Molen, 2004). In the latter case, the authors refer to the study by Stoter (2004). 3D models of physical objects (buildings and infrastructure) are not included into the Core Cadastral ANI ČLANKI | PEER-RE Domain Model (CCDM), but they are included into the set of relevant and related topics. The primary RECENZIR guidelines in developing CCDM are the inclusion of a maximum range of common characteristics of cadastral systems worldwide, as set out in Cadastre 2014 (Kaufmann and Steudtler, 1998) and specified in international standards (Oosterom et al., 2006). Great attention was given to the determination of its thematic scope; the authors developed CCDM in a very narrow manner but at the same time predicted the option of various thematic extensions. Such design facilitates the adjustment of the model to various systems around the world and at the same time preserves the basic level of comparability, i.e. common characteristics of cadastral systems. Further development of the model made it possible to include 3D parcels using the concept of bound surfaces, but with the limitation, i.e. that the area in question is recorded exclusively either in 2D format or 3D format. CCDM was the predecessor of the LADM model4. This new name of the model was first mentioned in 2008 (Groothedde et al. , 2008). FIG proposed the model to become an ISO standard (Lemmen, Oosterom and Uitermark, 2009) and since 2012 it has ENSI | been officially published as ISO 19152:2012 standard (LADM, 2012). In the initial period of research activities concerned with 3D property cadastre, studies and analyses of 3D geo-objects to be used in the 3D cadastre were carried out (Billen and Zlatanova, 2003; Tse and Gold, 2003). Stoter and Oosterom (2002) present the possibilities of modelling geo-objects in DBMS, providing the basis for managing cadastral systems from the perspective of information technology. Studies discuss management of 3D geo-objects in DBMS5 in terms of modelling, functionality, and visualisation (Zlatanova, 2006; Khuan, Abdul-Rahman and Zlatanova, 2008). 3D objects can be represented as tetrahedrons, polyhedrons, and multipolygons. In these papers, authors argue that 3D geo-objects can, indeed, be stored, as DBMSs support the storage of spatial features, such as points, lines, and polygons in 3D space, but difficulties arise in terms of their administration, analysis, transmission, and visualisation. These problems stemmed from the fact that, at the time, DBMS did not yet support the data type of volumetric 3D objects and thus did not allow for management of such data and analyses in a 3D environment. The checking of compliance of data with their formal definition is one of the most important aspects of data management; this also refers to 3D spatial data (Kazar et al. , 2008; Ledoux, Verbree and Si, 2009). This is highly significant in reference to 3D cadastres as well (Karki, Thompson and McDougall, 2010). Rather than focusing on the internal validity of individual objects, the authors addressed them in the context of a 3D cadastral system, where interrelationships between 3D objects and 2D parcels are important. Time, i.e. the temporal aspect, as the fourth dimension of reality, is among the key cadastral data components. The first explorations in this field (Oosterom et al. , 2006) deal with time-sensitive cases and focus 3 Core Cadastral Domain Model (CCDM) 4 Land Administration Domain Model 5 Database Management System (DBMS) Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 253 | | 62/2 | GEODETSKI VESTNIK on the meaning of adding the temporal component to the cadastral data model, regardless of it being a TICLES 2D or a 3D cadastre. The temporal aspect is also addressed in the framework of the LADM standard that WED AR was proposed at the time (Döner et al. , 2008) and investigated on the case of registering underground VIE infrastructures in various countries. The authors establish the relationship between the legal object and the physical object of underground utility networks by buffering, and at the same time distance themselves from registering the geometric description of the physical object in the cadastre. Notably, the authors ANI ČLANKI | PEER-RE argue that land administration systems have from the very beginning dealt with three dimensions as well RECENZIR as the temporal dimension within the current technical structure, i.e. as attributes. The temporal part of the data model is thus based on registering situations in time, i.e. “snapshots”, which is basically done in most cadastral systems, or registering the initial situation and all ensuing events. 3.1.3 legal aspects of the 3d cadastre In early investigations on the 3D property cadastre, the legal topic was given significant less consideration than studies focusing on information technology. The first comprehensive and extensive study concerned with the legal aspects of the 3D property cadastre in the broader international context was done in Sweden (Paulsson, 2007), where the author dealt with the basic problem of defining a 3D property unit. With the purpose of universality, she defines it as a spatial unit that is delimited both EN horizontally and vertically. She divides rights, restrictions, and responsibilities, which are distributed in SI | space, into condominium ownership, i.e. strata title, and independent 3D areas of rights. The strata title is treated as an established means of settling rights, which are delimited both horizontally and vertically, and therefore it is thoroughly examined in this work. The author comprehensively and systematically examines, and compares, four selected legal systems, which have different traditions and use different property right registration procedures in terms of their horizontal and vertical division: Germany with its traditional system of strata-title ownership and codified law, Sweden with a detailed legal system and new legislation allowing for registration of independent 3D property units, and Australian federal states New South Wales and Victoria, with ordinary law and legislation allowing for establishment of both strata title and independent 3D property units. 3.1.4 organisational aspect – 3d cadastre situation and perspectives Many publications from the first decade of intensified research on 3D property cadastres represent the situations and perspectives of introducing the 3D property cadastre in the individual countries. Most of them analyse the existing cadastral systems from their legal and technical aspects, complex cases where property registration should be tackled in three dimensions, and the options of 3D cadastre introduction in individual countries. These include specific conceptual designs, nevertheless, in all cases the technical solutions are, for the time being, practically not yet directly applicable. The possibilities of 3D cadastre establishment were studied in Israel (Benhamu and Doytsher, 2003), where solutions were sought on the principle of multi-layered cadastres, which could contain, along with the data layer for traditional parcels, the data layers for structures above and below the surface (Benhamu and Doytsher, 2001; Benhamu, 2006). Technical challenges, related to Israel and beyond, were discussed by Peres and Benhamu (2009), when the efforts towards operational6 implementation of 6 In this paper, the term “operational” refers to the actual use/implementation of something in the cadastral system of a certain country. | 254 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | 3D cadastre had already strengthened. It is also worth mentioning China, where private property can TICLES be established on built structures only. Tang and Yang (2009) developed a conceptual model to enable registration of 3D property units, while recognising that, at the time, it was not yet feasible due to the WED ARVIE lack of data and non-existent information and technological solutions for storing and managing 3D data. The literature reveals that Australian federal states New South Wales, Victoria (Paulsson, 2007), and Queensland (Stoter, 2004) have a tradition of a particular statutory scheme and registration of 3D ANI ČLANKI | PEER-RE property units. These can also apply to non-built-up areas above or below the Earth’s surface. From the perspective of storing data on property unit geometry, the property registration system in all the federal RECENZIR states mentioned was entirely based on 2D concepts. In this period Norway adopted legislation allowing for establishment of independent 3D property units (Valstad, 2010). The basic characteristics of this legislation were previously addressed by Onsrud (2003). Registration of independent 3D real property units is possible only for the purpose of registering engineering objects (Valstad, 2006), similarly as in Sweden (Eriksson, 2005; Paulsson, 2007). The cadastral system in both countries technically did not allow for digital registration of 3D geometry of property units. Registration of a single apartment as an independent 3D unit is not possible neither in Sweden nor in Norway. For this, the condominium registration has to be used, which regulates relationships between individual property units of a building. It should be underlined that both countries have a single land administration system, which was years ago established by combining the former dual system (legal and ENSI | technical). This fact is stressed because of the organisational and institutional aspects of introducing the 3D cadastre and other major changes into land administration. Contrary to the previously mentioned countries, which are in this period addressed more often due to their way of managing and registering 3D property units, the Netherlands kept the traditional organisation of its cadastral system. For several decades, the Netherlands has had a single land administration system inside one organisation (previously it had a dual system consisting of a land cadastre and a land register). The fact that the Netherlands frequently comes up in studies is the result of Dutch researchers’ efforts and collaboration of the academic sphere and the surveying administration, which at the beginning of the decade greatly accelerated studies into 3D cadastres (Stoter and Ploeger, 2003; Stoter and Salzmann, 2003; Stoter, 2004). In the first decade, by introducing the Building Cadastre, Slovenia set the basis for developing the 3D cadastre (Pogorelčnik and Korošec, 2001; Rijavec, 2009), but to date there have not been any major steps taken in this direction, while a major problem is also the poor link between land parcels and buildings, deficiencies in cadastral recording of engineering structures and infrastructural works (that are not buildings), and the insufficient data model of the Building Cadastre (see also Drobež, 2016; Drobež et al., 2017). In terms of operational implementation of the 3D real property cadastre in practice, the literature at the end of the decade often highlights that further development in all relevant fields is necessary, with the exception of the legal field in some countries, where there are practically no legal constraints. Interestingly, the study by Çağdaş and Stubkjær (2009), analysing methodological approaches used in doctoral researches concerned with cadastral system development, does not recognise the 3D cadastre nor the aforementioned doctoral studies (Stoter, 2004; Paulsson, 2007) as an important part of modern cadastre development. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 255 | | 62/2 | GEODETSKI VESTNIK 3.2 Studies concerned with the 3d cadastre after 2010 TICLES In 2011 a survey was conducted among the members of FIG Working Group on 3D Cadastres, with WED AR 36 member states of this group taking part (Oosterom et al. , 2011). The survey’s content focused on VIE inventorying the situation by countries in 2010 and their expectations for 2014. The results of the survey importantly contribute to studies on 3D cadastres, as they allow free access to the extensive set of data on cadastral systems in many countries. The authors of the survey find that countries have ANI ČLANKI | PEER-RE different cadastral systems, where the incongruent perception of the 3D cadastre stems from. The dif-RECENZIR ferences are mostly regarding the understanding of the connection of traditional 2D parcels and 3D property units with physical structures. By the time no country had developed the system for storing and managing 3D data on property units in cadastres. Most of them were highly restrained in their plans and expectations for 2014. In 2014 representatives of 31 replied to the second, updated survey on the condition and expectations in 3D cadastre for 2018 (Oosterom, Stoter and Ploeger, 2014). In all countries, where the legal system allowed for registration of 3D property units, the ways of data registration, storing, and management were still based on 2D cadastre. What the countries had in common was that digital cadastral databases were mostly “incongruent” with the standard scheme ISO 19152:2012 (LADM, 2012). China stands out in terms of storing 3D data in digital format, stating in the replies to the survey that their database allows for storing, validating, and managing the 3D geometry of property units. Nevertheless, In the Chinese case we find an extremely small ENSI | total number of parcels given the size of the country, so we assume that this situation is valid only for limited (urban) areas of China. Later studies (Guo et al. , 2013; Ho et al. , 2013; Dimopoulou, Karki and Roič, 2016; Stoter et al. , 2017) also confirm that at the beginning of the decade China did not have a fully operational 3D cadastral system. In 2011, the second workshop on 3D cadastre took place in the Netherlands, 10 years after the first one. Interestingly, the next, third, workshop was planned to take place in two or three years, but it was held the very next year, which shows the growing international interest in research and knowledge exchange in this field. The report from the 2011 workshop (Banut, 2011) breaks down the situation in individual development fields of the 3D cadastre, divided into legal aspects, first registrations of 3D property units, administration of 3D spatial data and visualisation, transmission of data, and accessibility of data on 3D property units: – The problem of terminological incongruency and various definitions of 3D cadastre was highlighted in the legal field. – More than two thirds of the papers at the 2011 workshop describe land administration systems and different regulations for vertical stratification of property units from the perspective of current studies and data models. – The field of managing digital 3D spatial data, particularly the fields of analyses and operations in DBMS and GIS7, has been strongly lagging behind the field of 3D visualisation of spatial data, which made strong progress in the first decade of this millennium. The importance of electronic accessibility of 3D property cadastral data and 3D web-based visualisation techniques is underlined. The thematic focus of the third workshop in 2012 was development and best practices in 3D cadastre (Oosterom, 2012). The need for more studies and comparative analyses of legal schemes in various coun-7 Geographical Information System | 256 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | tries and the requirement to use existing standards, both in terms of modelling property units (LADM) TICLES and modelling physical spatial structures in land administration information systems, were expressed. Inter alia, the meaning of visualisation of 3D property objects was stressed, as the needs and challenges WED ARVIE are different than with visualisation of more commonly used 3D models of cities and landscapes. Based on the papers from FIG workshops and conferences in 2011 and 2012, Oosterom (2013) provides an overview of development in 3D cadastres and presents the most important topics for further studies. ANI ČLANKI | PEER-RE We particularly underline the topics that remain topical at a global level today: RECENZIR – As mentioned earlier, the lack of studies concerned with legal aspects was discussed by Paulsson and Paasch (2013) after reviewing 156 publications in English between 2001 and 2011. They identified the lack of terminological and comparative studies that would cover several countries and several 3D cadastre legal schemes. – Heights and vertical systems in 3D cadastres were addressed in detail for the first time by Navratil and Unger (2013). They represented the general problems of vertical reference systems, restrictions, and demands of 3D cadastres, also on practical cases. The greatest attention is given to the analysis of strengths and weaknesses of using absolute and relative heights in the 3D cadastre. – An important element, which greatly influences the dynamics of establishing the 3D cadastre, is its cost–benefit relationships. One of such studies, with reference to Trinidad and Tobago, found EN a positive cost/benefit ration in urban, densely populated areas and the oil mining areas (Griffith SI | -Charles and Sutherland, 2013). The authors conclude that it is reasonable to explore the possibility of introducing the 3D cadastre in selected areas only, where benefits exceed costs. – Operational implementation of 3D cadastres, inter alia, requires the specification of clear rules regarding division of space into 3D property units, their modelling in the 3D environment, and, at the same time, coupled with validation procedures as to their compliance with the rules set. This is much more difficult in three spatial dimensions, as the set of rules is more extensive and the procedures of compliance verification are more complex than in the conventional 2D cadastre. Karki, Thompson and McDougall (2013) thoroughly studied this field and developed specific solutions and a set of challenges and questions that remain to be solved. They conclude that the development of land administration systems towards the 3D cadastre is not possible in a short period of time. One of the solutions is the gradual adaptation of existing systems, as argued by Guo et al. (2013). On the case of the Chinese cadastre, authors stem from the existing legal system and the 2D cadastral data model, into which they include elements of the 3D cadastre. This paper is also interesting when compared to the results of the previously mentioned research on 3D cadastre development by Oosterom, Stoter and Ploeger (2014), where we could have made the wrong assumption the China had had a fully operational 3D cadastral system before 2014. – The comparison of cadastral systems of various countries is difficult due to their diversity. Pouliot, Vasseur and Boubehrezh (2013) compared cadastral system models in France and Canada with a focus on the third spatial dimension based on the LADM standard (2012). They identified, as the most demanding part, the transformation of the data model of the individual system into the standard LADM scheme, which then allows for direct comparison between various cadastral systems, their classes, and attributes. They propose and justify the inclusion of volumetric geometry in the standard, which would increase the applicability of the LADM standard in the 3D cadastre as well. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 257 | | 62/2 | GEODETSKI VESTNIK The papers in the fourth FIG workshop on the topic of 3D cadastre in 2014 were mostly technically oriented, TICLES regardless of the voiced concerns about the lack of consideration of the legal problem (Paulsson and Paasch, 2013): WED AR – At the workshop, the legal aspect was addressed as a main topic only in one presentation, on the VIE case of the national study for Poland (Karabin, 2014). – There was a growing consideration of the ISO standard LADM (2012) and the open standardised data model CityGML by OGC8 (CityGML, 2012), which were found, in a combination or separately, in ANI ČLANKI | PEER-RE eight publications. Compared to the previous workshops, the number of studies and presentations RECENZIR of national cadastral systems dropped. Two publications were particularly interesting (Almeida et al., 2014; El-mekawy, Paasch and Paulsson, 2014), as they discussed voluntary geographical information and linked BIM9 solutions with the 3D cadastre. These two topics were presented as a challenge to the 3D cadastre, as both areas are intensively studied in the wider area of geosciences. Building Information Modeling (BIM) provides a potential for developing a 3D cadastre, and together they provide an important area for future research (Rajabifard, 2014). – The content of publications shows the growing interest in studies on 3D visualisation (Navratil and Fogliaroni, 2014; Pouliot, Wang and Hubert, 2014; Ribeiro, de Almeida and Ellul, 2014) which is confirmed by two extensive doctoral dissertations from the period (Shojaei, 2014; Wang, 2015). Wang (2015) focused on evaluating the suitability of 3D model visualisations for the case of strata title, while Shoaei (2014) mostly analysed user requirements and needs. ENSI | 3.2.1 Challenges related to the 3d cadastre from a legal perspective Even though the legal aspect of the 3D cadastre was not given significant consideration at the 2014 FIG workshop, this research domain remains topical at the international level. In land administration and property records the concepts of physical and abstract space meet, where rights, restrictions and responsibilities are associated to “abstract” spatial units. With incomplete knowledge of the field they can be equated based on coincidence of boundaries of physical structures and rights, i.e. restrictions in some cases. The division of space from the legal aspect is fundamentally abstract, while its link with physical space is established in various forms and from various reasons, while it varies from one legal system to another. The 3D cadastre domain is mostly directed towards treating partitioning of buildings into property units, as a relationship between space of legal significance and physical space and its structures (Aien, 2013; Aien et al. , 2013, 2015) therefore physical boundaries are often equated with legal boundaries. In these studies, the authors developed the 3DCDM data model10, which combines the physical and legal aspects of dividing space for the needs of the 3D cadastre. The evolution of the 3D cadastre in the first decade led to discrepancies regarding legal definitions of a property in three dimensions (Paasch and Paulsson, 2012). The authors find that the latter causes problems also in research, where inconsistent definitions of basic terminology limit the possibilities of comparative analyses and studies. They emphasise that the legal definition of a property in three dimensions must be broad enough to be acceptable in most legal systems. Secondly, the definition must provide a clear and unique definition of the property and delimit it from the traditional property in two dimensions. The authors propose using a universal definition of 3D properties, as previously proposed by Paulsson (2007). 8 Open Geospatial Consortium: http://www.opengeospatial.org/ 9 Building Information Modelling 10 3D Cadastral Data Model | 258 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | Among interesting legally-oriented studies is that by Ho et al. (2013) where the authors argue that the TICLES significant barriers to 3D cadastre implementation lie not in technology, which is developed well enough, nor in legal systems, as in many countries, thay allow for registration of 3D property units – instead, there WED ARVIE must be a limiting factor, inertia, preventing faster 3D cadastre implementation. Authors explain and break down this inertia by introducing the institutional theory and justifying that the reasons for slow changes lie in the slow adaptability within organisations responsible and the strongly rooted 2D concept ANI ČLANKI | PEER-RE in land administration systems. Oosterom and Lemmen (2015) stress that 3D (and 4D) administration are among the most significant development trends in land administration and thus also the LADM RECENZIR standard. They present the studies arising from the first thematic workshop after the publication of the LADM standard, among which two studies treat the topic of 3D cadastre development in Korea and Malesia (Lee et al. , 2015; Zulkifli et al. , 2015). 3.2.2 Challenges of introducing the 3d cadastre A special topic in developing the 3D cadastre is its operational implementation. The first studies concerned with the topic were done under the Dutch project intended to help the transition to the 3D cadastre, which is based on two implementation phases (Stoter, Oosterom and Ploeger, 2012; Stoter, Ploeger and Oosterom, 2013). The first phase was to gain experience, adjusting its solutions to existing legal and technical frameworks. Its implementation part relates to the possibility of property registration based ENSI | on a PDF document, which contains 3D geometry and is connected with other data about the property through a link in the database. The assessment of registration and system maintenance costs is interesting as, for new buildings, they should not be higher than the existing registration costs. Today, the first phase of implementing 3D cadastre allows for solving some complex situations, particularly to unambiguously show the division into property units, while the existing 2D land administration system basically remains the same. The second phase is far more ambitious, as it provides for 3D cadastre establishment, allowing for a comprehensive digital registration of property units, including geometry, in the form of 3D objects (volumetric bodies) directly in the cadastral database. Many issues arise in the second phase related to validation of 3D data geometries, required positioning and geometrical accuracy, data formats, inclusion of curved surfaces, partially open elements, etc. (Stoter et al., 2017). The project of 3D cadastral modelling in Russia began in collaboration with Dutch researchers. They developed a prototype that mostly focuses on the manner of modelling and representing 3D property units. In their designs, they defined a pilot project in a small area, where they would approach the real implementation of registration. Vandysheva et al. (2012) underline the meaning of automated control during the entry of new property units in the sense of compliance with previously set rules. Despite the intensified efforts for 3D cadastre establishment and land administration system upgrade in recent years, the Russian cadastral system is still based on two spatial dimensions (Ilyushina, Noszczyk and Hernik, 2017). 3.3 Current topics and studies on the 3d cadastre The last FIG workshop on 3D cadastre took place in 2016. The fact that this research domain is active is also proven by the increased number of papers (31) compared to the previous workshop (25). Extended papers were collected in a special issue of the international journal ISPRS International Journal of GeoInformation entitled Research in Development Progress in 3D cadastral systems 2017. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 259 | | 62/2 | GEODETSKI VESTNIK Most notable is the major body of research around the analysis of situations in individual countries and TICLES the different possibilities for developing land administration systems, particularly publications focus-WED AR ing on technical solutions of modelling 3D property units in land administration information systems VIE (Araújo and Oliveira, 2016; Dimas, 2016; Soon, Tan and Khoo, 2016; Gulliver, Haanen and Goodin, 2017). Other publications in internationally renowned journals also touch upon this field. Aien et al. (2017) underline six of the most established data models in the cadastre, where, due to the differences ANI ČLANKI | PEER-RE in land administration systems, many data models have emerged. The authors particularly highlight RECENZIR three of them (LADM, ePlan, ArcGIS Parcel Data Model) and analyse them in detail in terms of their usability for 3D cadastres. Data model 3DCDM, which was as part of his doctoral research developed by the lead author, is not included nor mentioned in the study. The authors conclude that some data models discussed allow for modelling 3D property units, but each of them has important limitations for 3D cadastre development. At the research level, the legal domain of studies is strongly represented (Kitsakis and Dimopoulou, 2017; Vučić et al. , 2017), headed by an extensive comparative study of selected countries (Kitsakis, Paasch and Paulsson, 2016; Paasch et al., 2016). Kitsakis, Paasch and Paulsson (2016) present the legal definition of 3D property units in various countries (Austria, Brazil, Croatia, Greece, Poland, and Sweden) and the plans for future development. Of these countries, Sweden is the only one that does not restrict registra-ENSI | tion of 3D property units in its legal system; however, data management in Sweden, and elsewhere, is still based on 2D concepts. Another interesting study, by Janečka and Souček (2017), is concerned with data modelling in management in 3D cadastres. The authors discuss the current situation in the wider area of 3D geoinformatics, which covers concepts, data models, standards, and operations related to 3D spatial data. The emphasis is on the current capacities of spatial databases in view of modelling and managing 3D spatial data. The connection or integration of BIM data with the 3D cadastre data model is extremely topical. In relation to BIM data, a growing number of studies is focusing on modelling indoor spaces of buildings for the needs of registering property units (Oldfield et al. , 2016; Atazadeh, 2017). Atazadeh et al. (2017) treat BIM as the basis for managing rights and restrictions associated to buildings. They propose the extension of the data model so that it could support the input of data on the rights and restrictions inside buildings and their management. Among other, the authors address the topical questions of relationship of 3D units of legal significance and a building’s physical model. Along with strengths, the authors discuss the limitations of the proposed approach, which include institutional barriers, the too extensive data structure, and the discrepancy between the planned structure and the structure actually built. The research by Zlatanova et al. (2016) is oriented towards modelling indoor spaces of buildings, where in 2014 the standard OGC – IndoorGML (2014) was used for the first time as part of the studies into the 3D cadastre. In this paper the authors discuss the options for linking the aforementioned standard with the LADM standard. Further research in this area was done by Alattas et al. (2017). The IndoorGML standard is based on a multi-layered space-event model, which was originally intended for indoor navigation, as proposed in 2009 (Becker, Nagel and Kolbe, 2009). This group of authors also led the development of CityGML, the previously developed standard for modelling cities and landscapes in the 3D environment. The IndoorGML standard introduces a cellular approach to modelling indoor | 260 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | spaces of buildings, using the duality principle (Munkres, 1984), coupled with mathematical graphs. It TICLES allows for the operation of the optimal path searching algorithms to support navigation as well as other algorithms based on topological relationships among the objects. The standard allows for extensions of the WED ARVIE data model in the form of extension modules for various fields. Kang and Li (2017) particularly stressed the possibility of implementing the extension module of the IndoorGML standard to be used in the 3D cadastre. Linking outdoor city models and models of buildings’ indoor spaces has been recognised as a ANI ČLANKI | PEER-RE research problem also by the United Nations Committee of Experts on Global Geospatial Information RECENZIR Management – UN-GGIM (2015). 4 ConCluSionS The beginnings of introducing the term 3D cadastre date back to the publishing of the document Cadastre 2014 (Kaufmann and Steudtler, 1998). To facilitate the materialisation of these goals, in 2001 FIG held the first workshop, which encouraged research into the 3D cadastre. The various aspects of developing 3D cadastres were set more clearly: legal, technical, and administrative, of which the first two are more strongly represented in studies. The first decade was characterised by many analyses of land administration systems in individual countries and proposals for their upgrading. They collectively concluded that additional development in all the mentioned research fields are needed to establish 3D registration. ENSI | Doctoral dissertation by Stoter (2004) left an indelible mark on the technical aspect of studying 3D cadastres. Most studies thereon related to her findings, definitions, and proposals. Two countries stand out in the legal field: Sweden, which in 2004 introduced the option of registering 3D property units in its legal system, and Australia with a longstanding tradition of possibilities to register independent 3D property units. The doctoral dissertation by Paulsson (2007) is among the most acclaimed studies into legal systems related to the 3D cadastre, providing a comprehensive review and insight into the legal aspect of the 3D cadastre. Cadastre 2014 set off initial designs of CCDM to unify the key components of land administration systems, with open possibilities for including specificities, and characteristics of individual countries. This is the direct predecessor of the international standard LADM (2012), which basically does not restrict the evolution of the traditional 2D cadastre into the 3D cadastre. The last decade of studies concerned with the 3D cadastre has been characterised by the publication of standards LADM (2012), CityGML (2012), and IndoorGML (2014). Most studies in this period study the possibility of using the standards, analyse the strengths and weaknesses of the individual standards and compare them, while fewer studies tackle the legal aspect, as previously found by Paulsson and Pasch (2013). During this time, Dutch researchers importantly contributed to implementing the concepts of the 3D cadastre into practice; in the future it will be interesting to see how a growing number of countries will decide to include the third dimension in the cadastre. Research challenges in 3D cadastres also relate to the integration of data from other domains, particularly research regarding the use or inclusion of BIM data in the 3D data model, and vice versa. The treatment or modelling of indoor structure of buildings is also topical; it is complex both from the aspect of data structure and complexity of data models as well as from the aspect of data acquisition and integration of models of indoor spaces and outdoor models of cities and landscapes (UN-GGIM, 2015). Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 261 | | 62/2 | GEODETSKI VESTNIK acknowledgements TICLES This review article is the result of work undertaken under research programme No. P2-0227 entitled WED AR “Geoinformation Infrastructure and Sustainable Spatial Development of Slovenia”, co-financed by the VIE Slovenian Research Agency (ARRS) from the national budget. references ANI ČLANKI | PEER-RE Aien, A. (2013). 3D Cadastral Data Modelling. Doctoral thesis. Melbourne: University Results. 1st International Workshop on 3D Cadastres, 28–30 November 2001 RECENZIR of Melbourne. (pp. 35–51). Delft, the Netherlands. Aien, A., Kalantari, M., Rajabifard, A., Wil iamson, I., Wal ace, J. (2013). Towards Benhamu, M., Doytsher, y. (2003). Toward a spatial 3D cadastre in Israel. Computers, integration of 3D legal and physical objects in cadastral data models, Land Use Environment and Urban Systems, 27 (4), 359–374. policy, 35, 140–154. DOI: http://dx.doi.org/10.1016/j.landusepol.2013.05.014 DOI: http://dx.doi.org/10.1016/S0198-9715(02)00036-4 Aien, A., Rajabifard, A., Kalantari, M., Shojaei, D. (2015). Integrating Legal and Billen, R., Zlatanova, S. (2003). 3D spatial relationships model: a useful concept for physical Dimensions of Urban Environments. ISpRS International Journal of Geo-3D cadastre? Computers, Environment and Urban Systems, 27 (4), 411–425. Information, 4 (3), 1442–1479. DOI: http://dx.doi.org/10.3390/ijgi4031442 DOI: http://dx.doi.org/10.1016/S0198-9715(02)00040-6 Aien, A., Rajabifard, A., Kalantari, M., Williamson, I. (2017). Review and Assessment CityGML (2012). OGC CityGML 2.0, Open Geospatial Consortium, OGC 12-019 of Current Cadastral Data Models for 3D Cadastral Applications. In Rahman, A. Çağdaş, V., Stubkjær, E. (2009). Doctoral research on cadastral development. Land A. (Ed.) Advances in 3D Geoinformation, Lecture Notes in Geoinformation and Use policy, 26 (4), 869–889. Cartography (pp. 423–442). Berlin: Springer-Verlag. DOI: http://dx.doi.org/10.1016/j.landusepol.2008.10.012. DOI: http://dx.doi.org/10.1007/978-3-319-25691-7_24 Dimas, E. (2016). 3D Cadastral Objects of the hellenic Cadastre. 5th International fIG ENSI | Alattas, A., Zlatanova, S., van Oosterom, p., Chatzinikolaou, E., Lemmen, C., Li, K. J. 3D Cadastre Workshop, 18–20 October 2016 (pp. 331–332). Athens, Greece. (2017). Supporting Indoor Navigation Using Access Rights to Spaces Based on Dimopoulou, E., Karki, S., Roič, M. (2016). Initial Registration of 3D parcels. 5th Combined Use of IndoorGML and LADM Models. ISpRS International Journal of International fIG 3D Cadastre Workshop, 18–20 October 2016 (pp. 105–132). Geo-Information, 6 (12), 1–32. DOI: http://dx.doi.org/10.3390/ijgi6120384 Athens, Greece. Almeida, J. D. De, haklay, M., Ellul, C., Carvalho, M.-M. (2014). The Role of Volunteered Döner, f., Thompson, R., Stoter, J. E., Lemmen, C., ploeger, h., van Oosterom, p. (2008). Geographic Information towards 3D property Cadastral Systems. 4th 4D Land Administration Solutions in the Context of the Spatial Information International Workshop on 3D Cadastres, 9–11 November 2014 (pp. 219–238). Infrastructure. fIG Working Week, 14–19 June 2008 (pp. 1–25). Stockholm, Dubai, United Arab Emirates. Sweden. Araújo, A. L. De, Oliveira, f. h. De (2016). Overlapping Characterization of Spatial parcels Drobež, p. (2016). Analiza možnosti vzpostavitve 3D katastra ob uporabi virov in Brazil: Case in florianopolis. 5th International fIG 3D Cadastre Workshop, daljinskega zaznavanja. Doctoral thesis. Ljubljana: University of Ljubljana, 18–20 October 2016 (pp. 405–418). Athens, Greece. faculty of Civil and Geodetic Engineering. Atazadeh, B. (2017). Assessing performance of Three BIM-Based Views of Buildings for Drobež, p., Kosmatin fras, M., ferlan, M., Lisec, A. (2017). Transition from 2D to 3D real Communication and Management of Vertically Stratified Legal Interests. ISpRS property cadastre: The case of the Slovenian cadastre, Computers, Environment International Journal of Geo-Information, 6 (7), 1–22. and Urban Systems, 62, 125–135. DOI: http://dx.doi.org/10.3390/ijgi6070198 DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2016.11.002 Atazadeh, B., Kalantari, M., Rajabifard, A., ho, S., Ngo, T. (2017). Building Information El-mekawy, M., paasch, J., paulsson, J. (2014). Integration of 3D Cadastre, 3D property Modelling for high-rise Land Administration. Transactions in GIS, 21 (1), 91–113. formation and BIM in Sweden. 4th International Workshop on 3D Cadastres, DOI: http://dx.doi.org/10.1111/tgis.12199 9–11 November 2014 (pp. 17–34). Dubai, United Arab Emirates. Banut, R. (2011). Overview working sessions. 2nd International Workshop on 3D Eriksson, G. (2005). A New Multi-Dimensional Information System Introduced in Cadastres, 16–18 November 2011 (pp. 1–10). Delft, the Netherlands. Sweden. fIG Working Week 2005 (pp. 1–15). Cairo. Becker, T., Nagel, C., Kolbe, T. h. (2009). A multilayered space-event model for Griffith-Charles, C., Sutherland, M. (2013). Analysing the costs and benefits of 3D navigation in indoor spaces. In Lee, J., Zlatanova, S. (Eds.) 3D Geo-Information cadastres with reference to Trinidad and Tobago. Computers, Environment and Sciences (pp. 61–77). Berlin: Springer-Verlag. Urban Systems, 40, 24–33. DOI: http://dx.doi.org/10.1007/978-3-540-87395-2 DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.07.002. Benhamu, M. (2006). A GIS-Related Multi Layers 3D Cadastre in Israel. XXIII Grinstein, A. (2001). Aspects of a 3D Cadastre in the New City of Modi’in, Israel. 1st International fIG Congress, 8–13 October 2006 (pp. 1–12). Munich, Germany. International Workshop on 3D Cadastres, 28–30 November 2001 (pp. 25–33). Benhamu, M., Doytsher, y. (2001). Research Toward A Multilayer 3-D Cadastre: Interim Delft, the Netherlands. | 262 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | Groothedde, A., van der Molen, p., Lemmen, C., van Oosterom, p. (2008). A DOI: http://dx.doi.org/10.1007/978-3-540-72135-2_16. Standardized Land Administration Domain Model As part Of The (spatial) Kitsakis, D., Dimopoulou, E. (2017). Addressing public Law Restrictions within a 3D TICLES Information Infrastructure. In van Oosterom, p., Zlatanova, S. (Eds.) Creating Cadastral Context. ISpRS International Journal of Geo-Information, 6 (7), 1–14. WED AR Spatial Information Infrastructures (pp. 129–150). CRC press. DOI: http://dx.doi.org/10.3390/ijgi6070182. VIE DOI: http://dx.doi.org/10.1201/9781420070729.ch9. Kitsakis, D., paasch, J. M., paulsson, J. (2016). 3D Real property Legal Concepts and Gulliver, T., haanen, A., Goodin, M. (2017). A 3D Digital Cadastre for New Zealand and Cadastre : A Comparative Study of Selected Countries to propose a Way forward. the International Opportunity. ISpRS International Journal of Geo-Information, 5th International fIG 3D Cadastre Workshop, 18-20 October 2016 (pp. 1–24). 6 (12), 1–375. DOI: http://dx.doi.org/10.3390/ijgi6110375. ANI ČLANKI | PEER-RE Athens, Greece. Guo, R., Li, L., ying, S., Luo, p., he, B., Jiang, R. (2013). Developing a 3D cadastre for the Koler-povh T., Lisec A. (2015). Geodetski vestnik na poti boljše mednarodne RECENZIR administration of urban land use: A case study of Shenzhen, China. Computers, prepoznavnosti. Geodetski vestnik, 59 (2), 289–319. Environment and Urban Systems, 40, 46–55. DOI: http://dx.doi.org/10.15292/geodetski-vestnik.2015.02.289-319 DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.07.006. LADM. (2012). ISO 19152:2012. Geographic information––Land ho, S., Rajabifard, A., Stoter, J., Kalantari, M. (2013). Legal barriers to 3D cadastre AdministrationDomain Model (LADM). International Organization for implementation: What is the issue? Land Use policy, 35, 379–387. Standardization (ISO),Geneva, Switzerland. DOI: http://dx.doi.org/10.1016/j.landusepol.2013.06.010. Ledoux, h., Verbree, E., Si, h. (2009). Geometric Validation of GML Solids with the huml, M. (2001). how To Define Real Estate: 2D or 3D? Legal View, Conditions and Constrained Delaunay Tetrahedralization. 4th International Workshop on 3D Experiences in the Czech Republic. 1st International Workshop on 3D Cadastres, Cadastres, 9–11 November 2014 (pp. 143–148). Dubai, United Arab Emirates. 28–30 November 2001 (pp. 293–299). Delft, the Netherlands. Lee, B.-M., Kim, T.-J., Kwak, B.-y., Lee, y., Choi, J. (2015). Improvement of the Korean Ilyushina, T. V, Noszczyk, T., hernik, J. (2017). Cadastral system in the Russian federation LADM country profile to build a 3D cadastre model. Land Use policy, 49, after the modern transformation. Survey Review. Taylor & francis, 0(0), 1–10. 660–667. DOI: http://dx.doi.org/10.1016/j.landusepol.2015.10.012. DOI: http://dx.doi.org/10.1080/00396265.2017.1308700. EN Lemmen, C., van Oosterom, p. (2003). 3D Cadastres. Computers, Environment and SI | IndoorGML. (2014). OGC IndoorGML, Open Geospatial Consortium, OGC 14-005r4 Urban Systems, 27 (4), 337–343. Janečka, K., Souček, p. (2017). A Country profile of the Czech Republic Based on an DOI: http://dx.doi.org/10.1016/S0198-9715(02)00034-0. LADM for the Development of a 3D Cadastre. ISpRS International Journal of Lemmen, C., van Oosterom, p., Uitermark, h. (2009). Transforming the Land Geo-Information, 6 (5), 1–19. DOI: http://dx.doi.org/10.3390/ijgi6050143. Administration Domain Model (LADM) into an ISO Standard (ISO19152). fIG Kang, h.-K., Li, K. J. (2017). A Standard Indoor Spatial Data Model—OGC Working Week, 3–8 May 2009 (pp. 1–24). Eilat, Israel. IndoorGML and Implementation Approaches. ISpRS International Journal of Geo-Information, 6 (4), 1–25. DOI: http://dx.doi.org/10.3390/ijgi6040116. Menda, J. N. (2001). Registration of properties in strata in Kenya. 1st International Workshop on 3D Cadastres, 28–30 November 2001 (pp. 67–77). Delft, the Karabin, M. (2014). A Concept of a Model Approach to the 3D Cadastre in poland : Netherlands. Technical and Legal Aspects. 4th International Workshop on 3D Cadastres, 9–11 November 2014 (pp. 281–298). Dubai, United Arab Emirates. van der Molen, p. (2003). Institutional aspects of 3D cadastres. Computers, Environment and Urban Systems, 27(4), 383–394. Karki, S., Thompson, R., McDougall, K. (2010). Data validation in 3D cadastre. In T. DOI: http://dx.doi.org/10.1016/S0198-9715(02)00038-8. Neutens, Maeyer, p. D (Eds.) Developments in 3D Geo-Information Sciences, Lecture Notes in Geoinformation and Cartography (pp. 92–122). Berlin: Munkres, J. R. (1984). Elements of Algebraic Topology. Edited by R. W. Mixter and C. Springer-Verlag. DOI: http://dx.doi.org/10.1007/978-3-642-04791-6_6. W. harris. Addison-Wesley. Karki, S., Thompson, R., McDougall, K. (2013). Development of validation rules to Navratil, G., fogliaroni, p. (2014). Visibility Analysis in a 3D Cadastre. 4th International support digital lodgement of 3D cadastral plans. Computers, Environment and Workshop on 3D Cadastres, 9–11 November 2014 (pp. 183–196). Dubai, Urban Systems, 40, 34–45. United Arab Emirates. DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.10.007. Navratil, G., Unger, E. M. (2013). Reprint of: Requirements of 3D cadastres for height Kaufmann, J., Steudtler, D. (1998). Cadastre 2014: A Vision for future Cadastral Systems. systems. Computers, Environment and Urban Systems, 40, 14–23. Working Group 7.1, Commission 7, fIG, July 1998. DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2013.04.001. Kazar, B. M., Kothuri, R., van Oosterom, p., Ravada, S. (2008). On Valid and Invalid Oldfield, J., van Oosterom, p., Quak, W., van der Veen, J. (2016). Can Data from BIMs Three-Dimensional Geometries. In van Oosterom, p., Zlatanova, S., penninga, be Used as Input for a 3D Cadastre?. 5th International fIG 3D Cadastre Workshop, f., fendel, E. (Eds.) Advances in 3D Geoinformation Systems (pp. 19–46). Berlin: 18–20 October 2016 (pp. 199–214). Athens, Greece. Springer-Verlag. DOI: http://dx.doi.org/10.1007/978-3-540-72135-2_2. Onsrud, h. (2001). Making Laws for 3D Cadastre in Norway Making Laws for 3D Khuan, C. T., Abdul-Rahman, A., Zlatanova, S. (2008). 3D solids and their management Cadastre in Norway. 1st International Workshop on 3D Cadastres, 28–30 in DBMS. In van Oosterom, p., Zlatanova, S., penninga, f., fendel, E. (Eds.) November 2001 (pp. 191–199). Delft, the Netherlands. Advances in 3D Geoinformation Systems (pp. 279–311). Berlin: Springer-Verlag. Onsrud, h. (2003). Making a Cadastre law for 3D properties in Norway. Computers, Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 263 | | 62/2 | GEODETSKI VESTNIK Environment and Urban Systems, 27 (4), 375–382. Technical Considerations. fIG Working Week, 3–8 May 2009 (pp. 1–14). TICLES DOI: http://dx.doi.org/10.1016/S0198-9715(02)00037-6. Eilat, Israel. van Oosterom, p. (2012). Third International fIG Workshop on 3D Cadastres – pogorelčnik, E., Korošec, M. (2001). Land Cadastre and Building Cadastre in Slovenia: WED ARVIE Developments and practices, 3rd International Workshop on 3D Cadastres, Current Situation and potential of 3D Data. 1st International Workshop on 3D 25–26 October 2012 (pp. 1–6). Shenzhen, China. Cadastres, 28–30 November 2001 (pp. 79–90). Delft, the Netherlands. van Oosterom, p. (2013). Research and development in 3D cadastres. Computers, pouliot, J., Vasseur, M., Boubehrezh, A. (2013). how the ISO 19152 Land Administration Environment and Urban Systems, 40, 1–6. Domain Model performs in the comparison of cadastral systems: A case study of ANI ČLANKI | PEER-RE DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2013.01.002. condominium/co-ownership in Quebec (Canada) and Alsace Moselle (france). Computers, Environment and Urban Systems, 40, 68–78. RECENZIR van Oosterom, p., Lemmen, C. (2003). Towards a Standard for the Cadastral Domain: proposal to establish a Core Cadastral Data Model. Journal of Geospatial DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.08.006. Engineering, 5 (1), 11–27. pouliot, J., Wang, C., hubert, f. (2014). Transparency performance in the 3D van Oosterom, p., Lemmen, C. (2015). The Land Administration Domain Model (LADM): Visualization of Bounding Legal and physical Objects: preliminary Results Motivation, standardisation, application and further development. Land Use of a Survey Transparency performance in the 3D Visualization of Bounding policy, 49, 527–534. DOI: http://dx.doi.org/10.1016/j.landusepol.2015.09.032. Legal and physical Objects: preliminary Results of a Survey. 4th International van Oosterom, p., Lemmen, C., Ingvarsson, T., van der Molen, p., ploeger, h., Quak, W., Workshop on 3D Cadastres, 9–11 November 2014 (pp. 173–182). Dubai, Stoter, J., Zevenbergen, J. (2006). The core cadastral domain model. Computers, United Arab Emirates. Environment and Urban Systems, 30 (5), 627–660. Rajabifard, A. (2014). 3D Cadastres and Beyond. 4th International Workshop on 3D DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2005.12.002. Cadastres, 9–11 November 2014 (pp. 1–16). Dubai, United Arab Emirates. van Oosterom, p., Lemmen, C., van der Molen, p. (2004). Remarks and Observations Ribeiro, A., de Almeida, J.-p., Ellul, C. (2014). Exploring CityEngine as a Visualisation Tool related to the further development of the Core Cadastral Domain Model. for 3D Cadastre. 4th International Workshop on 3D Cadastres, 9–11 November 2014 (pp. 197–218). Dubai, United Arab Emirates. EN proceedings of the Workshop Standardisation of the Cadastral Domain, 9–10 SI | December 2004 (pp. 1–19). Bamberg, Germany. Rijavec, U. (2009). 3D-Katatser stavb - pogoj za vstop v tretje tisočletje. Geodetski van Oosterom, p., ploeger, h., Stoter, J., Thompson, R. (2006). Aspects of a 4D cadastre: Vestnik, 53 (4), 731–737. a first exploration. XXIII International fIG Congress, 8–13 October 2006 (pp. Rokos, D. (2001). Conceptual modeling of real property objects for the hellenic 1–23). Munich, Germany. cadastre. 1st International Workshop on 3D Cadastres, 28–30 November 2001 van Oosterom, p., Stoter, J., ploeger, h. (2014). Initial Analysis of the Second fIG (pp. 137–154). Delft, the Netherlands. 3D Cadastres Questionnaire: Status in 2014 and Expectations for 2018. 4th Shojaei, D. (2014). 3D Cadastral Visualisation: Understanding User’s Requirements. International Workshop on 3D Cadastres, 9–11 November 2014 (pp. 55–74). Doctoral thesis. University of Melbourne, Victoria, Australia. Dubai, United Arab Emirates. Soon, K. h., Tan, D., Khoo, V. (2016). Initial Design to Develop a Cadastral System that van Oosterom, p., Stoter, J., ploeger, h., Thompson, R., Karki, S. (2011). World-wide Supports Digital Cadastre, 3D and provenance for Singapore. 5th International Inventory of the Status of 3D Cadastres in 2010 and Expectations for 2014. fIG fIG 3D Cadastre Workshop, 18-20 October 2016 (pp. 419–432). Athens, Greece. Working Week, 18–22 May 2011 (pp. 1–21). Marrakech, Morocco. Stoter, J. (2000). Needs, possibilities and constraints to develop a 3D cadastral Ossko, A. (2001). problems in Registration in the Third Vertical Dimension in the Unified registration system. UDMS 2000 Annual Congress (pp. 43–58). Savannah, USA. Land Registry in hungary and possible Solution. 1st International Workshop on Stoter, J. (2004). 3D Cadastre. Doctoral thesis. TU Delft, Delft. 3D Cadastres, 28–30 November 2001 (pp. 305–314). Delft, the Netherlands. paasch, J., paulsson, J. (2012). Terminological Aspects on Three-dimensional Real Stoter, J. E., ploeger, h. D. (2003). property in 3D-registration of multiple use of property. Nordic journal of surveying and real estate research, 8 (1), 65–80. space: Current practice in holland and the need for a 3D cadastre. Computers, paasch J. M., paulsson J., Navratil G., Vučić N., Kitsakis D., Karabin M., El-Mekawy M. Environment and Urban Systems, 27, 553–570. (2016). Building a modern cadastre: legal issues in describing real property in DOI: http://dx.doi.org/10.1016/S0198-9715(03)00014-0. 3D. Geodetski vestnik, 60 (2), 256–68. Stoter, J., van Oosterom, p. (2002). Incorporating 3D geo-objects into a 2D geo-DBMS. DOI: http://dx.doi.org/10.15292/geodetski-vestnik.2016.02.256-268 proceedings fIG, ACSM/ASpRS (pp. 1–12). Washington D.C. paulsson, J. (2007). 3D property rights - An analysis of key factors based on Stoter, J., van Oosterom, p., ploeger, h. (2012). The phased 3D Cadastre Implementation international experience. Doctoral thesis. Stockholm: Royal Institute of in the Netherlands. 3rd International Workshop on 3D Cadastres, 25–26 October Technology, Sweden. 2012 (pp. 203–218). Shenzhen, China. paulsson, J., paasch, J. M. (2013). 3D property research from a legal perspective. Stoter, J., ploeger, h., van Oosterom, p. (2013). 3D cadastre in the Netherlands: Computers, Environment and Urban Systems, Special issue 3D Cadastres II, 40, Developments and international applicability. Computers, Environment and 7–13. DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.11.004 Urban Systems, 40, 56–67. peres, N., Benhamu, M. (2009). 3D Cadastre GIS – Geometry, Topology and Other DOI: http://dx.doi.org/10.1016/j.compenvurbsys.2012.08.008 | 264 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | Stoter, J., ploeger, h., Roes, R., van der Riet, E., Biljecki, f., Ledoux, h., Kok, D., Kim, S. Workshop on 3D Cadastres, 28–30 November 2001 (pp. 91–97). Delft, the (2017). Registration of Multi-Level property Rights in 3D in The Netherlands: Netherlands. TICLES Two Cases and Next Steps in further Implementation. ISpRS International Journal Vučić, N., Roić, M., Mađer, M., Vranić, S., van Oosterom, p. (2017). Overview of the WED AR of Geo-Information, 6 (6), 1–18. DOI: http://dx.doi.org/10.3390/ijgi6060158. Croatian Land Administration System and the possibilities for its upgrade to 3D VIE Stoter, J., Salzmann, M. (2003). Towards a 3D cadastre: Where do cadastral needs and by existing data. ISpRS International Journal of Geo-Information, 6, 1–20. DOI: technical possibilities meet?. Computers, Environment and Urban Systems, 27, http://dx.doi.org/10.3390/ijgi6070223. 395–410. DOI: http://dx.doi.org/10.1016/S0198-9715(02)00039-X. Wang, C. (2015). 3D Visualization of Cadastre: Assessing the Suitability of Visual ANI ČLANKI | PEER-RE Stoter, J., Zevenbergen, J. (2001). Changes in the definition of property: a consideration Variables and Enhancement Techniques in the 3D Model of Condominium for a 3D cadastral registration system. fIG Working Week 2001 (pp. 1–12). Seoul. property Units. Doctoral thesis. Quebec: Université Laval, Canada. RECENZIR Tang, K., yang, M. (2009). Registration for 3d cadastre in China. 2009 International Zlatanova, S. (2006). 3D Geometries in Spatial DBMS. In Abdul-Rahman, A., Zlatanova, forum on Computer Science-Technology and Applications, pp. 79–82. DOI: S., Coors, V. (Eds.) Innovations in 3D Geo Information Systems (pp. 1–14). Berlin: http://dx.doi.org/10.1109/IfCSTA.2009.140. Springer - heidelberg. DOI: http://dx.doi.org/10.1007/978-3-540-36998-1_1 Tse, R. O. C., Gold, C. (2003). A proposed connectivity-based model for a 3-D cadastre. Zlatanova, S., van Oosterom, p., Lee, J., Li, K. J., Lemmen, C. h. J. (2016). Ladm Computers, Environment and Urban Systems, 27, 427–445. and Indoorgml for Support of Indoor Space Identification. ISpRS Annals of DOI: http://dx.doi.org/10.1016/S0198-9715(02)00041-8. photogrammetry, Remote Sensing and Spatial Information Sciences, IV-2/ UN-GGIM (2015). future Trends in geospatial information management: the five to W1 (October), 257–263. ten year vision. Second edition. United Nations Committee of Experts on Global DOI: http://dx.doi.org/10.5194/isprs-annals-IV-2-W1-257-2016. Geospatial Information Management. Zulkifli, N. A., Abdul Rahman, A., van Oosterom, p., Tan, L. C., Jamil, h., Teng, C. h., United Nations Economic Commission for Europe (2004). Guidelines on Real property Looi, K. S., Chan, K. L. (2015). The importance of Malaysian Land Administration Units and Identifiers. Geneva, pp. 1–68. Domain Model country profile in land policy. Land Use policy, 49, 649–659. DOI: http://dx.doi.org/10.1016/j.landusepol.2015.07.015. Valstad, T. (2006). Developments of the 3D Cadastre in Norway. XXIII International fIG EN Zupan, M., Lisec, A., ferlan, M., čeh, M. (2014). Razvojne usmeritve na področju SI | Congress, 8-13 October 2006 (pp. 1–12). Munich, Germany. zemljiškega katastra in zemljiške administracije (= Development guidelines in Valstad, T. (2010). At Last! After 10 years in Waiting Norway has a New Cadastral Law the field of land cadastre and land administration). Geodetski vestnik, 58 (4), that also Includes 3D Objects. fIG Congress 2010, 11–16 April. Sydney, Australia. 710–723. DOI: http://dx.doi.org/10.15292/geodetski-vestnik.2014.04.710-723 Vandysheva, N., Sapelnikov, S., federation, R., van Oosterom, p., Vries, M. de, Spiering, B., Wouters, R. (2012). The 3D Cadastre prototype and pilot in the Russian federation. fIG Working Week 2012, 6-10 May (pp. 1–10). Rome, Italy. Viitanen, K. (2001). 3D property Ownership 3D Land Use in finland. 1st International Tekavec J., ferlan M., Lisec A. (2018). A review of research on 3D real property cadastre. Geodetski vestnik, 62 (2), 249-265. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 265 | | 62/2 | GEODETSKI VESTNIK PREGLED RAZISKAV NA PODROČJU 3D-KATASTRA TICLES NEPREMIČNIN WED ARVIE OSNOVNE INfORMACIJE O čLANKU: ANI ČLANKI | PEER-RE GLEJ STRAN 249 RECENZIR 1 uvod Urbani razvoj in z njim vse bolj zapleteni primeri razmejevanja prostora z vidika lastninske ali drugih pravic na nepremičninah zahtevajo nov pristop v zemljiški administraciji, ki omogoča evidentiranje in spreminjanje nepremičninskih enot ter z njimi povezanih podatkov v treh prostorskih razsežnostih. Vse večja fizična in pravna kompleksnost grajenega, pa tudi naravnega okolja narekuje nadgradnjo modeliranja prostora v dveh razsežnostih, ki je tradicionalno navzoče v nacionalnih sistemih zemljiške administracije. Področje zemljiške administracije je bilo od nekdaj zelo zahtevno za raziskave na mednarodni ravni, saj so države razvile svojevrstne sisteme, ki so med drugim zelo pogojeni z zgodovinskim ozadjem zemljiške ENSI| administracije, s pravnim sistemom in družbeno ureditvijo, pa tudi s potrebami družbe na področju upravljanja prostora (Zupan et al., 2014). Zaradi vse večjih potreb po razvoju sodobnih rešitev na področju zemljiške administracije se je pojavila zahteva po mednarodni primerljivosti in s tem po strukturirani obravnavi zemljišč ter pravic, omejitev in odgovornosti (angl. rights, restrictions and responsibilities) na njih. Rezultat številnih mednarodnih razprav na tem področju je v letu 2012 sprejeti mednarodni standard ISO 19152:2012: Land Administration Domain Model (LADM). Namen prispevka je podati širok pregled mednarodno prepoznanih objav in s tem predstaviti razvoj področja 3D-katastra nepremičnin v preteklih desetletjih. Na podlagi obravnavanih objav smo analizirali aktualne raziskovalne teme, ki na mednarodni ravni oblikujejo osrčje raziskovalnega dela na tem področju. 2 uPoraBlJena MeTodoloGiJa in viri S področjem razvoja konceptov 3D-katastra ter tehničnih in pravnih rešitev za uvedbo 3D-katastra se ukvarjajo raziskovalci in razvijalci z različnih področij. Skupno vsem je, da se osredotočajo na precej specifično domeno, to je kataster nepremičnin, ki tako združuje dognanja na svetovni ravni. Glavni viri za našo raziskavo so bili dosegljivi viri in zapisi tematske delovne skupine za 3D-kataster mednarodnega združenja FIG1, ki je že leta 2001 organizirala prvi mednarodni forum, s čimer je želela prispevati k razvojnim rešitvam na področju 3D-katastra nepremičnin. Pomemben vir za naše delo so bile objave v dveh tematskih številkah mednarodne znanstvene revije Computers, Environemnt and Urban Planning iz let 2003 in 2013, kjer je objavljen tudi pregled vsebin razprav v okviru organizacije FIG do leta 2012 (Oosterom, 2013), ter objave v reviji ISPRS International Journal of Geo-Information s tematsko številko Research and Development Progress in 3D cadastral systems iz leta 2017. Dodatno smo pregledali doktorske raziskave s tega področja in angleške članke v drugih mednarodnih revijah, ki so objavljeni s 1 Fédération Internationale des Géomètres: www.fig.net. | 266 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | široko uveljavljenim označevalcem elektronske lokacije DOI (angl. digital object identifier). V ta namen TICLES smo uporabili iskalnik CrossRef 2, ki je ena izmed rešitev za založnike in omogoča kreiranje navedenega označevalca ter uvrščanje revije v izjemno veliko mednarodno družino elektronskih znanstvenih in stro-WED ARVIE kovnih objav (glej tudi Koler Povh in Lisec, 2015). 3 reZulTaTi – PreGled raZvoJa PodroČJa 3d-KaTaSTra nePreMiČnin ANI ČLANKI | PEER-RE Začetki izrazitejšega oblikovanja zamisli o 3D-katastru nepremičnin segajo v leto 1994 z začetkom RECENZIR delovne skupine 7.1 mednarodnega združenja FIG, ki je leta 1998 objavilo vizijo razvoja katastra pod naslovom Cadastre 2014 (Kaufmann in Steudtler, 1998). V dokumentu je izpostavljena vloga katastra kot pomembnega deležnika pri trajnostnem razvoju in odločitvah v prostoru. V dokumentu so podane pomembne opredelitve, ki med drugim kot osnovni nepremičninski element v katastru namesto parcele (parcelno orientiran kataster) uvajajo splošnejši izraz, to je ‚objekt‘, na katerega so vezane pravice, omejitve in odgovornosti. Kataster naj bi tako celostno izkazoval stanje prostora ter zagotavljal pravno varnost in transparentnost glede pravic, omejitev in odgovornosti, ki se nanašajo na katastrske objekte. Napovedana sta bila konec ločevanja opisnih in grafičnih podatkov v katastru ter uvedba računalniškega modeliranja, ki nadomešča analogno katastrsko kartiranje, oboje kot posledica hitrega razvoja informacijske tehnologije. Navedene nove opredelitve in usmeritve so med drugim spodbudile razpravo o uvajanju tretje prostorske razsežnosti v nepremičninske evidence. ENSI | 3.1 raziskave na področju 3d-katastra nepremičnin v obdobju 2000–2010 Prvi rezultati raziskav, ki odpirajo raziskovalno področje 3D-katastrov nepremičnin in močno vplivajo na nadaljnjo mednarodno raziskovalno dejavnost, so bili objavljeni že na prelomu tisočletja (Stoter, 2000; Stoter in Zevenbergen, 2001). Avtorja ugotavljata, da registracija nepremičnin v dveh razsežnostih v številnih primerih ne prinaša zadostne pravne varnosti glede pravic in omejitev na nepremičninah, prav tako ne zadošča drugim funkcionalnostim sistema zemljiške administracije. 3.1.1 Prve mednarodno prepoznavne raziskave, razprave in objave Že navedeno raziskovalno delo na Tehniški univerzi v Delftu na Nizozemskem pomeni uvod v prvo delavnico na temo 3D-katastrov nepremičnin v letu 2001, ki je bila organizirana pod okriljem zveze FIG, s tem pa se je razširilo zanimanje in utrdil položaj tematike 3D-katastrov v raziskovalni sferi. Uvedena je bila klasifikacija področja raziskav na pravne, tehnične in organizacijske vidike, ki se je v skoraj nespremenjeni obliki ohranila vse do danes. Na delavnici so prevladovale predstavitve sistemov zemljiške administracije posameznih držav, obstoječih načinov reševanja zapletenih primerov registracije nepremičnin, kjer obstaja potreba po višinski razdelitvi pravic, in možnosti za nadaljnji razvoj (Grinstein, 2001; Huml, 2001; Menda, 2001; Onsrud, 2001; Ossko, 2001; Rokos, 2001; Viitanen, 2001). Med njimi najdemo tudi predstavitev takrat uvedenega Katastra stavb v Sloveniji (Pogorelčnik in Korošec, 2001). Ugotovitve z navedene delavnice sta predstavila Lemmen in Oosterom (2003) kot uvod v tematsko številko mednarodne revije Computers, Environemnt and Urban Planning, kjer so objavljeni izbrani 2 CrossRef: www.crossref.org, nazadnje pr egledano 10. 1. 2018. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 267 | | 62/2 | GEODETSKI VESTNIK prispevki z delavnice. Molen (2003) v svojem delu izpostavi, da spremembe kompleksnih sistemov, TICLES kot je kataster, vedno zahtevajo tudi organizacijske spremembe v institucijah in da morajo le-te slediti WED AR tehnološkemu napredku. Posebej je izpostavljena dilema oziroma razlika med objekti pravnega pomena VIE in objekti, ki predstavljajo fizične strukture v prostoru. Onsrud (2003) predstavi novo pravno ureditev na Norveškem, ki je bila v procesu sprejemanja in omogoča registracijo 3D-nepremičninskih enot za potrebe ureditve pravic in omejitev na grajenih objektih. Predstavljena je tudi možnost kombinacije ANI ČLANKI | PEER-RE navedene ureditve z obstoječo, ki temelji na konceptu etažne lastnine. Takratni sistem registracije ne-RECENZIR premičnin na Norveškem je še v celoti temeljil na 2D-parcelah, deloma celo še v analogni obliki. Avtor v bližnji prihodnosti ni videl realnih možnosti, da bi pravno veljavne 3D-nepremičninske enote tudi tehnično lahko registrirali kot 3D-objekte. Velik izziv, ki je še vedno aktualen na področju 3D-katastra, je bil način učinkovitega modeliranja katastrskih podatkov v informacijskih sistemih, kar se odraža v več objavljenih prispevkih navedene tematske številke (Billen in Zlatanova, 2003; Stoter in Ploeger, 2003; Tse in Gold, 2003). Tse in Gold (2003) predlagata za modeliranje geometrije in topologije 3D-katastrskih objektov mrežo nepravilnih trikotnikov, kar utemeljita z izvedljivostjo predlagane rešitve. Tudi Billen in Zlatanova (2003) se ukvarjata z načinom modeliranja prostorskih objektov, s poudarkom na njihovih medsebojnih povezavah. Stoter in Ploeger (2003) predstavita možne načine razvoja tradicionalnih sistemov v smeri 3D-katastra. EN Doktorska disertacija Stoterjeve (2004) je prvo obširno raziskovalno delo na področju 3D-katastra, v SI| katerem je celovito obravnavan njegov tehnični vidik, dotika pa se tudi pravnega in deloma organiza-cijskega vidika. Predstavi več primerov iz prakse na Nizozemskem, kjer dvorazsežni pristop v zemljiški administraciji ne zadovoljuje zahtev po pregledni registraciji nepremičnin. Z analizo stanja zemljiške administracije v mednarodnem okolju in podrobnejšo analizo izbranih držav je ugotovila, da takrat nobena država ni imela razvitega sistema za 3D-registracijo nepremičninskih enot, prav tako ta ni bil nikjer predmet aktivnega razvoja. Avtorica izpostavi in jasno utemelji potrebo po uvedbi 3D-katastra na izbranih študijskih primerih. Obravnava predvsem področje modeliranja, upravljanja in predstavitve podatkov o nepremičninah v 3D-okolju. Predstavljene so takratne zmogljivosti informacijske tehnologije za vzpostavitev podatkovnih baz, rešitve za shranjevanje 3D-geometrije, postopki za preverjanje pravil-nosti zapisanih podatkov in funkcije za upravljanje podatkov. V okviru doktorskega raziskovalnega dela je avtorica razvila tri različne katastrske podatkovne modele. Pri prvem gre za nadgradnjo obstoječih katastrskih sistemov s shranjevanjem povezav do 3D-podatkov, ki so shranjeni ločeno. Hibridni model ohranja obstoječo vlogo tradicionalnega 2D-katastra kot podlago, na katero se vežejo pravice, omejitve in obveznosti, a dovoljuje registracijo 3D-objektov za namene jasnejšega prikaza stanja pravic in omejitev v prostoru v posebnih primerih. Pravi 3D-kataster omogoča registracijo prostorninskih parcel. 3D-parcele prevzamejo vlogo katastrskih objektov – parcele v katastru niso več opredeljene kot 2D-poligon, ampak kot 3D-telo – pri tradicionalnih parcelah brez vertikalne razmejitve pa kot pokončni stolpi, ki vertikalno niso omejeni. Gre torej za prostorninsko razdelitev celotnega prostora s 3D-nepremičninskimi enotami. Tu velja omeniti še dejavnosti Ekonomske komisije Združenih narodov za Evropo, ki je leta 2004 z namenom podpore razvoju učinkovitih sistemov zemljiške administracije objavila dokument z usmerit-vami, s poudarkom na nepremičninskih enotah in objektnih identifikatorjih (United Nations Economic Commission for Europe, 2004), ki pomembno izpostavlja vlogo sodobnega katastra z ekonomskega vidika. Dokument naj bi prispeval k uskladitvi terminologije in razumevanju razlik med sistemi posa- | 268 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | meznih držav ter tako omočil lažje mednarodno sodelovanje in izmenjavo podatkov na tem področju. TICLES Deloma se dotika tudi problematike tretje prostorske razsežnosti nepremičninskih enot, etažne lastnine in območij mineralnih surovin. WED ARVIE 3.1.2 3d-kataster in podatkovni modeli Uvedba 3D-katastra ne pomeni le uvedbe shranjevanja dodatne prostorske razsežnosti, ampak prinaša ANI ČLANKI | PEER-RE korenite spremembe celotnega katastrskega podatkovnega modela. Začetek razvoja na področju 3D-ka-RECENZIR tastra konec devetdesetih sovpada z intenzivnim obdobjem uvajanja računalniškega modeliranja tudi na področju zemljiške administracije. V letu 2003 je bil z željo po poenotenju konceptov in podatkovnih modelov nacionalnih sistemov objavljen prvi predlog katastrskega podatkovnega modela CCDM3 (Oosterom in Lemmen, 2003). 3D-kataster je izpostavljen kot poseben primer, ki je opredeljen kot možnost za nadgradnjo oziroma razširitev osnovnega katastrskega podatkovnega modela skupaj s časovnim vidikom (Oosterom, Lemmen in Molen, 2004). Pri slednjem se avtorji sklicujejo na raziskavo Stoterjeve (2004). 3D-modeli fizičnih objektov (stavbe in infrastruktura) niso vključeni v osnovni podatkovni model CCDM, so pa uvrščeni v sklop ustreznih povezanih vsebin. Primarna vodila pri razvoju CCDM so vključitev čim širšega obsega skupnih značilnosti katastrskih sistemov po svetu, upoštevanje izhodišč dokumenta Cadastre 2014 (Kaufmann in Steudtler, 1998) in mednarodnih standardov (Oosterom et EN al. , 2006). Velika pozornost je bila namenjena določitvi njegovega tematskega obsega; avtorji so osnovni SI | podatkovni model zasnovali zelo ozko in obenem predvideli možnost različnih tematskih razširitev. Takšna zasnova omogoča lažje prilagajanje modela različnim sistemom po svetu in obenem ohranja osnovno raven primerljivosti oziroma skupnih značilnosti katastrskih sistemov. Nadaljnji razvoj modela je prinesel možnost vključitve 3D-parcel ob uporabi koncepta povezanih površin, a z omejitvijo, da je posamezno območje evidentirano izključno v 2D- ali izključno v 3D-obliki. CCDM je neposredni predhodnik modela LADM4. Novo ime modela se pojavi v letu 2008 (Groothedde et al., 2008). Zveza FIG ga je predlagala za ISO-standard (Lemmen, Oosterom in Uitermark, 2009) in je od leta 2012 uradno objavljen kot standard ISO 19152:2012 (LADM, 2012). Že v začetnem obdobju raziskovalne dejavnosti na področju 3D-katastra nepremičnin se pojavljajo raziskave in analize 3D-grafičnih gradnikov, ki naj bi se uporabljali v 3D-katastru (Billen in Zlatanova, 2003; Tse in Gold, 2003). Stoter in Oosterom (2002) predstavita možnosti modeliranja grafičnih gradnikov v okviru SUPB5, ki je osnova za upravljanje katastrskih sistemov z informacijsko-tehnološkega vidika. V raziskavah je obravnavano upravljanje 3D-grafičnih gradnikov v SUPB z vidika modeliranja, funkcionalnosti in vizualizacije (Zlatanova, 2006; Khuan, Abdul-Rahman in Zlatanova, 2008). Kot 3D-gradniki se omenjajo tetraedri, poliedri in poligoni. V navedenih delih avtorji ugotavljajo, da je 3D-grafične gradnike sicer mogoče shranjevati, saj SUPB podpirajo shranjevanje točk, linij in poligonov v 3D-prostoru, težave pa vidijo pri njihovem upravljanju, analizah, posredovanju in vizualizaciji. Navedene težave so izvirale iz dejstva, da takrat SUPB še niso podpirali podatkovnega tipa za prostorninske 3D-grafične gradnike in tako tudi niso omogočali upravljanja takšnih podatkov ter analiz v 3D-okolju. 3 Angl. core cadastral domain model. 4 Angl. land administration domain model. 5 Sistem za upravljanje s podatkovnimi bazami. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 269 | | 62/2 | GEODETSKI VESTNIK Med najpomembnejšimi vidiki upravljanja podatkov je preverjanje njihove skladnosti z njihovo formalno TICLES definicijo, kar velja tudi za 3D-prostorske podatke (Kazar et al., 2008; Ledoux, Verbree in Si, 2009). WED AR Slednje ima velik pomen tudi na področju 3D-katastrov (Karki, Thompson in McDougall, 2010). V VIE navedeni raziskavi se avtorji niso omejili zgolj na notranjo pravilnost posameznih grafičnih gradnikov, ampak jih obravnavajo v kontekstu 3D-katastrskega sistema, kjer so pomembni medsebojni odnosi 3D-grafičnih gradnikov in odnosi do 2D-parcel. ANI ČLANKI | PEER-RE Med ključnimi podatkovnimi komponentami katastra je poleg prostorskih razsežnosti tudi čas, ki ga RECENZIR lahko pojmujemo kot četrto razsežnost stvarnosti. Prve raziskave na tem področju (Oosterom et al., 2006) obravnavajo časovno problematične primere in se osredotočajo predvsem na pomen dodajanja časovne komponente v katastrski podatkovni model, ne glede na to, ali gre za 2D- ali 3D-kataster. Časovni vidik je obravnavan tudi v okviru takrat predlaganega standarda LADM (Döner et al., 2008) na primeru registracije podzemne infrastrukture v različnih državah. Avtorji vzpostavijo povezavo pravnega objekta in fizičnega objekta podzemne infrastrukture prek uporabe operacije območij (angl. buffer) in se hkrati oddaljijo od registracije geometrijskih podatkov o fizičnem objektu v katastru. Pomenljiva je teza avtorjev, da zemljiška administracija že od samega začetka upravlja tako s tremi prostorskimi kot tudi s časovno razsežnostjo v skladu s tehničnimi možnostmi, torej v obliki atributnih podatkov. Časovni del podatkovnega modela lahko temelji na registraciji stanj v času, ki se v osnovi uporablja v večini katastrskih ENSI| sistemov, ali registraciji začetnega stanja in vseh nadaljnjih dogodkov. 3.1.3 Pravni vidik 3d-katastra V prvem obdobju raziskav na področju 3D-katastra nepremičnin je bila pravna tematika zastopana šibkeje od informacijsko-tehnološko usmerjenih raziskav. Prva celovita in obširna raziskava pravnih vidikov 3D-katastra nepremičnin v širšem mednarodnem okolju prihaja iz Švedske (Paulsson, 2007), kjer se avtorica ukvarja z osnovnim problemom definicije 3D-nepremičninske enote (angl. 3D property unit). Zaradi univerzalnosti jo opredeli kot prostorsko enoto, ki je horizontalno in vertikalno razmejena. Pravice, omejitve in odgovornosti, ki so tako razmejene v prostoru, razdeli na etažno lastnino in samo-stojna 3D-območja pravic. Etažna lastnina je obravnavana kot uveljavljen način urejanja pravic, ki so v prostoru horizontalno in vertikalno razmejene, zato je v navedenem delu obravnavana najpodrobneje. Avtorica podrobno in sistematično obravnava in primerja izbrane štiri pravne ureditve, ki imajo različno tradicijo in različne ureditve registracije pravic na nepremičninah: Nemčijo s tradicionalnim sistemom etažne lastnine in kodificiranim pravom, Švedsko s podobnim pravnim sistemom in novo zakonodajo, ki omogoča registracijo neodvisne 3D-nepremičninske enote, ter avstralski državi Novi Južni Wales in Viktorija, z običajnim pravom in zakonodajo, ki dovoljuje vzpostavitev tako etažne lastnine kot tudi neodvisne 3D-nepremičninske enote. 3.1.4 organizacijski vidik – stanje in perspektive 3d-katastra V mnogih objavah iz prvega desetletja intenzivnega raziskovanja na področju 3D-katatsrov nepremičnin je predstavljeno stanje in perspektive uvedbe 3D-katastra nepremičnin v posameznih državah. V večini so analizirani obstoječi katastrski sistemi s pravnega in tehničnega vidika, zapleteni primeri, kjer je treba registracijo nepremičnine urejati v treh razsežnostih, ter možnosti uvedbe 3D-katastra v posamezni | 270 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | državi. Med njimi najdemo tudi posamezne konceptualne zasnove, vsem pa je skupno, da obravnavane TICLES tehnične rešitve še niso neposredno izvedljive v praksi. WED AR V Izraelu so raziskovali možnosti vzpostavitve 3D-katastra (Benhamu in Doytsher, 2003), kjer so VIE rešitve iskali po načelu večnivojskega katastra, ki bi poleg podatkovnega sloja za tradicionalne parcele lahko vseboval tudi podatkovna sloja za objekte pod in nad površjem (Benhamu in Doytsher, 2001; Benhamu, 2006). Tehnične izzive, vezane na Izrael, pa tudi širše, sta predstavila Peres in Benhamu ANI ČLANKI | PEER-RE (2009), ko so se že okrepila prizadevanja za operativno6 realizacijo 3D-katastra. Med državami velja RECENZIR izpostaviti še Kitajsko, kjer je zasebna lastnina lahko vzpostavljena le na grajenih strukturah. Tang in Yang (2009) sta razvila konceptualni model, ki bi omogočal registracijo 3D-nepremičninskih enot, a obenem priznavata, da takrat ni bil izvedljiv zaradi pomanjkanja podatkov ter neobstoječih informacijskih in tehnoloških rešitev za shranjevanje in upravljanje 3D-podatkov. V literaturi zasledimo tudi avstralske zvezne države Novi Južni Wales, Viktorija (Paulsson, 2007) in Queensland (Stoter, 2004) s tradicijo pravne ureditve in registracije 3D-nepremičninskih enot. Te so lahko formirane tudi za območja nad ali pod površjem Zemlje, neodvisno od fizičnih objektov. Sistem registracije nepremičnin z vidika shranjevanja podatkov o geometriji nepremičninske enote pa je v vseh navedenih zveznih državah v celoti temeljil na 2D-konceptih. Norveška je v tem obdobju uveljavila zakonodajo, ki omogoča vzpostavitev samostojnih 3D-nepre-EN mičninskih enot (Valstad, 2010). Osnovne značilnosti te zakonodaje je obravnaval že Onsrud (2003). SI | Registracija samostojnih 3D-nepremičninskih enot je mogoča le za namene registracije grajenih objektov (Valstad, 2006), podobno kot na Švedskem (Eriksson, 2005; Paulsson, 2007). V obeh državah katastrski sistem tehnično ni dovoljeval digitalne registracije 3D-geometrije nepremičninskih enot. Registracija posameznega stanovanja kot samostojne 3D-enote ni mogoča niti na Švedskem niti na Norveškem. Slednje je ostalo v domeni etažne lastnine. Poudariti velja, da imata obe državi enoten sistem zemljiške administracije, ki je pred leti nastal z združitvijo nekdanjega dualnega sistema (pravnega in tehničnega). Dejstvo izpostavljamo zaradi organizacijsko-institucionalnega vidika uvajanja 3D-katastra in drugih obsežnejših sprememb v zemljiško administracijo. V nasprotju z navedenimi državami, ki so v obravnavanem obdobju pogosteje obravnavane zaradi nji-hovega načina urejanja in registracije 3D-nepremičninskih enot, je Nizozemska ohranjala tradicionalno ureditev katastrskega sistema. Država ima sicer že več desetletij enoten sistem zemljiške administracije znotraj ene organizacije (prej je imela tudi dualni sistem z delitvijo na zemljiški kataster in zemljiško knjigo). Pogosta pojavnost Nizozemske v raziskavah je predvsem posledica dejavnosti raziskovalcev ter sodelovanja akademske sfere z njihovo geodetsko upravo, kar je v začetku desetletja zelo pospešilo raziskave na področju 3D-katastrov (Stoter in Ploeger, 2003; Stoter in Salzmann, 2003; Stoter, 2004). Slovenija je v prvem desetletju z uvedbo katastra stavb vzpostavila podlago za razvoj 3D-katastra (Pogorelčnik in Korošec, 2001; Rijavec, 2009), a vse do danes ni bilo intenzivnejšega razvoja v tej smeri,. Velika težava pri tem so slaba povezava zemljiških parcel in stavb, nedorečenost katastrskega evidentiranja infrastrukturnih objektov, ki niso stavbe, ter pomanjkljivi podatkovni model katastra stavb (glej tudi Drobež, 2016; Drobež et al., 2017). 6 Z besedo 'operativno' v članku označujemo dejansko uporabo/izvedbo nečesa v katastrskem sistemu posamezne države. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 271 | | 62/2 | GEODETSKI VESTNIK Glede operativne izvedbe 3D-katastra nepremičnin v praksi tudi ob koncu desetletja v literaturi večkrat TICLES zasledimo, da je potreben nadaljnji razvoj na vseh ustreznih področjih. Izstopa le pravno področje v tistih WED AR državah, kjer pravnih omejitev za registracijo 3D-nepremičninskih enot ni. Zanimivo je, da Çağdaş in VIE Stubkjær (2009) v raziskavi, v kateri sta analizirala metodološke pristope doktorskih raziskav na področju razvoja katastrskih sistemov, ne izpostavljata 3D-katastra in navedenih doktorskih raziskav (Stoter, 2004; Paulsson, 2007) kot pomembnega dela razvoja sodobnih katastrov. ANI ČLANKI | PEER-RE RECENZIR 3.2 raziskave na področju 3d-katastra po letu 2010 V letu 2011 je bila opravljena raziskava med člani FIG-ove delovne skupine o 3D-katastrih nepremičnin, v kateri je sodelovalo 36 držav članic navedene skupine (Oosterom et al., 2011). Vsebina vprašalnika je bila osredotočena na inventarizacijo stanja po državah v letu 2010 in njihova pričakovanja za leto 2014. Rezultati zelo pomembno prispevajo k raziskavam o 3D-katastrih, saj omogočajo prost dostop do obširnega nabora podatkov o katastrskih sistemih številnih držav. Avtorji raziskave ugotavljajo, da imajo države različne katastrske sisteme, iz česar izhaja tudi neenotno pojmovanje 3D-katastra. Razlike so predvsem glede razu-mevanja povezave tradicionalne 2D-parcele in 3D-nepremičninskih enot s fizičnimi strukturami. Nobena od držav takrat še ni imela razvite možnosti shranjevanja in upravljanja 3D-podatkov o nepremičninskih enotah v katastrih. Glede načrtov in pričakovanj za leto 2014 je bila večina zelo zadržanih. Leta 2014 so ENSI| predstavniki 31 držav podali odgovore na drugi, dopolnjen vprašalnik o stanju in pričakovanjih na področju 3D-katastra za leto 2018 (Oosterom, Stoter in Ploeger, 2014). V vseh državah, v katerih je pravni sistem omogočal registracijo 3D-nepremičninskih enot, so načini registracije, shranjevanja in upravljanja podatkov še vedno temeljili na 2D-zasnovanem katastru. Državam je bila skupna večinska »neskladnost« digitalnih katastrskih podatkovnih baz s shemo standarda ISO 19152:2012 (LADM, 2012). Glede shranjevanja 3D-podatkov v digitalni obliki močno izstopa Kitajska, ki v odgovorih na vprašalnik navaja, da njihova podatkovna baza omogoča shranjevanje, preverjanje in upravljanje 3D-geometrije nepremičninskih enot. A pri kitajskem primeru hkrati zasledimo izjemno majhno skupno število parcel glede na velikost države, zato gre sklepati, da opisano stanje velja le za omejena (urbana) območja Kitajske. Tudi kasnejše raziskave (Guo et al., 2013; Ho et al., 2013; Dimopoulou, Karki in Roič, 2016; Stoter et al., 2017) potrjujejo, da Kitajska v začetku desetletja ni imela polno operativnega sistema 3D-katastra. Leta 2011 je na Nizozemskem potekala druga delavnica o 3D-katastru, kar je bilo torej deset let po prvi. Zanimivo je, da je bila takrat napovedana naslednja, tretja delavnica v dveh ali treh letih, a je bila izvedena že v naslednjem letu, kar kaže na povečano zanimanje za raziskave in izmenjavo znanja na mednarodni ravni na tem področju. V poročilu o delavnici leta 2011 (Banut, 2011) je predstavljeno stanje na posameznih razvojnih področjih 3D-katastra, ki so deljena na pravne vidike, prve registracije 3D-nepremičninskih enot, upravljanje 3D-prosotrskih podatkov in vizualizacijo, posredovanje in dostopnost podatkov o 3D-nepremičninskih enotah: – Na pravnem področju je izpostavljena težava neusklajenosti terminologije ter različnih opredelitev 3D-katastra. – Več kot dve tretjini prispevkov opisujeta sisteme zemljiške administracije posameznih držav in načine urejanja višinske delitve nepremičninskih enot v luči aktualnih raziskav in podatkovnih modelov. – Na področju upravljanja digitalnih 3D-prostorskih podatkov, predvsem področja analiz in izvajanja | 272 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | operacij v SUPB in GIS7, je prepoznan močan zaostanek za področjem 3D-vizualizacije prostorskih podatkov, ki je v prvem desetletju tega tisočletja močno napredovalo. Izpostavljena je pomembnost TICLES spletne dostopnosti 3D-podatkov katastrov nepremičnin in tehnik 3D-vizualizacije na spletu. WED ARVIE Tematski poudarek tretje delavnice leta 2012 je bil na razvoju in dobrih praksah na področju 3D-katastra (Oosterom, 2012). Izražena je bila predvsem potreba po večjem številu raziskav in primerjalnih analiz pravnih ureditev v različnih državah in zahteva po uporabi obstoječih standardov, tako s področja mo-ANI ČLANKI | PEER-RE deliranja nepremičninskih enot (LADM) kot tudi s področja modeliranja fizičnih prostorskih struktur v informacijskih sistemih zemljiške administracije. Med drugim je bil poudarjen pomen vizualizacije RECENZIR 3D-nepremičninskih objektov, saj so potrebe in izzivi drugačni kot pri vizualizaciji bolj razširjenih 3D-modelov mest in pokrajin. Izhajajoč iz prispevkov z delavnic in konferenc združenja FIG leta 2011 in 2012, Oosterom (2013) podaja oris razvoja na področju 3D-katastrov, kjer predstavi tudi najpomembnejše teme za prihodnje raziskave. Izpostavljamo predvsem tematike, ki so še vedno aktualne na mednarodni ravni: – Že navedeno pomanjkanje raziskav na pravnem področju sta Paulsson in Paasch (2013) utemeljila na podlagi študije 156 objav angleških člankov med letoma 2001 in 2011. Ugotovila sta pomanjkanje predvsem terminoloških in primerjalnih raziskav, ki bi zajemale več držav in pravnih ureditev na področju 3D-katastra. EN – Višine in višinske sisteme 3D-katastrov sta prvič izčrpneje obravnavala Navratil in Unger (2013). SI | Predstavila sta splošno problematiko višinskih referenčnih sistemov in omejitve ter zahteve 3D-katastrov, tudi na primerih iz prakse. Največ pozornosti je namenjene analizi prednosti in slabosti uporabe absolutnih in relativnih višin v 3D-katastru. – Zelo pomemben element, ki ima velik vpliv na dinamiko uveljavitve 3D-katastra, je razmerje med stroški in koristmi, ki jih prinaša. Ena od redkih takšnih raziskav na primeru Trinidada in Tobaga ugotavlja pozitivno razmerje stroškov in koristi na urbanih in gosto poseljenih območjih ter naha-jališčih naravnih surovin (Griffith-Charles in Sutherland, 2013). Avtorja ugotavljata, da je smiselno proučiti možnosti vpeljave 3D-katastra le na nekaterih območjih, kjer koristi odtehtajo stroške. – Operativna uvedba 3D-katastra med drugim zahteva določitev jasnih pravil glede razmejitev prostora na 3D-nepremičninske enote ter modeliranja teh enot v 3D-okolju, skupaj s postopki preverjanja njihove skladnosti s postavljenimi pravili. Slednje je v treh prostorskih razsežnostih veliko težavnejše, saj je nabor pravil obširnejši, pa tudi postopki preverjanja skladnosti z njimi so zahtevnejši kot v tra-dicionalnem 2D-katastru. Karki, Thompson in McDougall (2013) so navedeno področje podrobno proučili, oblikovali nekatere rešitve ter nabor izzivov in nerešenih vprašanj. Avtorji sklenejo, da razvoj sistemov zemljiške administracije v smeri 3D-katastra ni mogoč v kratkem časovnem obdobju. Ena od rešitev je postopno prilagajanje obstoječih sistemov, kar zagovarjajo Guo et al. (2013). Avtorji na primeru kitajskega katastra izhajajo iz obstoječega pravnega sistema in 2D-podatkovnega modela, v katerega vključijo elemente 3D-katastra. Članek je zanimiv tudi za primerjavo z rezultati že navedene raziskave o razvoju 3D-katastra (Oosterom, Stoter in Ploeger, 2014), kjer bi lahko za Kitajsko napačno sklepali, da je že pred letom 2014 imela polno delujoč sistem 3D-katastra. – Primerjava katastrskih sistemov med posameznimi državami je zaradi njihove različnosti zelo te- žavna. Pouliot, Vasseur in Boubehrezh (2013) so primerjali modele katastrskih sistemov Francije 7 Geografski informacijski sistem. Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 273 | | 62/2 | GEODETSKI VESTNIK in Kanade, s poudarkom na tretji prostorski razsežnosti, na podlagi standarda LADM (2012). Kot TICLES najzahtevnejši del izpostavijo transformacijo podatkovnega modela posameznega sistema v standardno WED AR shemo LADM, kar v nadaljevanju omogoča neposredno medsebojno primerjavo katastrskih sistemov, VIE njihovih razredov in atributov. Predlagajo in utemeljijo tudi vključitev prostorninske geometrije v standard, s čimer bi povečali uporabnost standarda LADM tudi na področju 3D-katastra. Prispevki četrte delavnice združenja FIG na temo 3D-katastra iz leta 2014 so vsebinsko večinoma tehnično ANI ČLANKI | PEER-RE usmerjeni, ne glede na pozive o pomanjkanju obravnave pravne problematike (Paulsson in Paasch, 2013): RECENZIR – Pravni vidik je kot glavna tematika na delavnici obravnavan le v eni objavi v obliki nacionalne študije za Poljsko (Karabin, 2014). – Izrazito se je povečala obravnava ISO-standarda LADM (2012) in odprtega standarda CityGML združenja OGC8 (CityGML, 2012), ki ju v kombinaciji ali samostojno zasledimo v osmih objavah. Število raziskav in predstavitev sistemov po posameznih državah je glede na prejšnje delavnice upa-dlo. Med objavami sta zanimivi dve (Almeida et al., 2014; El-mekawy, Paasch in Paulsson, 2014), ki med prvimi obravnavata področje prostovoljnega zbiranja prostorskih podatkov in povezavo informacijskega modeliranja stavb (BIM9) s 3D-katastrom. Izpostavljeni tematiki sta predstavljeni kot izziv za 3D-kataster, saj sta obe področji danes raziskovalno zelo intenzivni na širšem področju geo-znanosti. BIM predstavlja potencial za razvoj večnamenskega 3D-katastra in pomembno področje EN prihodnjih raziskav (Rajabifard, 2014). SI| – Vsebina objav kaže na večje zanimanje za raziskave na področju 3D-vizualizacije (Navratil in Fogliaroni, 2014; Pouliot, Wang in Hubert, 2014; Ribeiro, de Almeida in Ellul, 2014), kar potrjujeta tudi obsežni doktorski disertaciji iz tega obdobja (Shojaei, 2014; Wang, 2015). Wang (2015) se je osredotočil na ocenjevanje primernosti načinov vizualizacije 3D-modelov za primer prikaza etažne lastnine, med tem ko se je Shoaei (2014) ukvarjal predvsem z analizami uporabniških zahtev in njihovih potreb. 3.2.1 izzivi na področju 3d-katastra s pravnega vidika Kljub relativno skromni obravnavi pravnega vidika 3D-katastra na navedeni FIG-ovi delavnici leta 2014 je področje bilo in je še vedno aktualno na mednarodni ravni. Na področju zemljiške administracije in nepremičninskih evidenc se namreč srečujeta koncepta fizičnega prostora in abstraktnega prostora, kjer se na »abstraktne« prostorske enote nanašajo pravice, omejitve in odgovornosti. Ob nepopolnem poznavanju področja ju lahko enačijo na podlagi pogostega sovpadanja meja fizičnih struktur in meja pravic oziroma omejitev. Razdelitev prostora s pravnega vidika je v osnovi abstraktna, njena povezava s fizičnim prostorom pa je vzpostavljena v različnih oblikah in iz različnih vzrokov, razlikuje pa se med pravnimi sistemi. Področje 3D-katastra je pogosto usmerjeno v obravnavo razdelitve stavb na nepremičninske enote, kjer gre za povezavo med prostorom pravnega pomena in fizičnim prostorom ter nje-govimi strukturami (Aien, 2013; Aien et al., 2013, 2015), iz česar izhaja tudi pogosto enačenje fizičnih meja prostora s pravnimi (abstraktnimi). V navedenih raziskavah so avtorji razvili model 3DCDM10, ki združuje fizične in pravne vidike razdelitve prostora za potrebe 3D-katastra. 8 Open Geospatial Consortium: http://www.opengeospatial.org/. 9 Angl. building information modelling. 10 Angl. 3D-cadastral domain model. | 274 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | Razvoj področja 3D-katastra v prvem desetletju je privedel do razhajanj glede pravnih opredelitev TICLES nepremičnine v treh razsežnostih (Paasch in Paulsson, 2012). Avtorja ugotavljata, da slednje povzroča težave tudi na raziskovalnem področju, saj so zaradi neenotnih opredelitev temeljnih pojmov omejene WED ARVIE možnosti primerjalnih analiz in študij. Kot poudarita, mora biti pravna opredelitev nepremičnine v treh razsežnostih dovolj široka, da je lahko sprejemljiva v večini pravnih sistemov. Poleg tega mora opredelitev takšno nepremičnino jasno in enolično opredeliti ter jo razmejiti od tradicionalne nepremičnine v dveh ANI ČLANKI | PEER-RE razsežnostih. Avtorja predlagata univerzalno opredelitev 3D-nepremičnine, kot jo je že predlagana Paul-ssonova (2007). Z vidika pravno usmerjenih raziskav je zanimiva raziskava Ho et al. (2013), kjer avtorji RECENZIR izhajajo iz teze, da je tehnologija dovolj razvita, pa tudi pravni sistemi v številnih državah dovoljujejo registracijo 3D-nepremičninskih enot, zato mora obstajati zaviralna sila, ki preprečuje hitrejše uveljavljanje 3D-katastra. To silo avtorji pojasnijo in razčlenijo z institucionalno teorijo in utemeljijo, da so razlogi za počasne spremembe v počasnem prilagajanju odgovornih organizacij in močno uveljavljenem 2D-konceptu v sistemih zemljiške administracije. Oosterom in Lemmen (2015) izpostavita 3D- (in 4D-) administracijo kot enega od pomembnejših razvojnih trendov zemljiške administracije in s tem tudi standarda LADM. Predstavita raziskave, ki izhajajo iz prve tematske delavnice po objavi standarda LADM, med katerimi sta tudi dve na temo razvoja 3D-katastra v Koreji in Maleziji (Lee et al., 2015; Zulkifli et al., 2015). 3.2.2 izzivi pri uveljavljanju 3d-katastra ENSI | Posebno poglavje v razvoju 3D-katastra je njegova operativna uvedba. Prve raziskave na to temati-ko so potekale v okviru nizozemskega projekta, namenjenega podpori prehoda na 3D-kataster, ki temelji na dveh izvedbenih fazah (Stoter, Oosterom in Ploeger, 2012; Stoter, Ploeger in Oosterom, 2013). Prva faza je bila namenjana pridobivanju izkušenj in se v rešitvah prilagaja obstoječim pravnim in tehničnim okvirom. Njen izvedbeni del se nanaša na možnosti registracije nepremičnine na podlagi dokumenta v obliki zapisa PDF, ki vsebuje 3D-geometrijo in je povezan z drugimi podatki o nepremičnini prek povezave v podatkovni bazi. Zanimiva je ocena stroškov registracije in vzdrževanja sistema, ki naj za nove stavbe ne bi bili višji od obstoječih stroškov registracije. Prva faza uvedbe 3D-katastra danes omogoča reševanje nekaterih kompleksnih situacij, predvsem z vidika nedvoumnega prikaza razdelitve na nepremičninske enote, še vedno pa obstoječ 2D-sistem zemlji- ške administracije v osnovi ostaja nespremenjen. Druga faza je zasnovana veliko ambiciozneje, saj predvideva uvedbo 3D-katastra, ki omogoča celovito digitalno registracijo nepremičninskih enot, vključno z geometrijo, v obliki 3D-objekotv (prostorninskih teles) neposredno v podatkovni bazi katastra. V zvezi z drugo fazo je nerešenih veliko vprašanj glede kontrole geometrije, zahtevane položajne in geometrijske natančnosti, podatkovnih formatov, vključitve ukrivljenih ploskev, delno odprtih gradnikov ipd. (Stoter et al., 2017). V Rusiji so začeli projekt 3D-katastrskega modeliranja v sodelovanju z nizozemskimi raziskovalci. Razvili so prototip, ki se osredotoča predvsem na način modeliranja in prikazovanja 3D-nepremičninskih enot. V načrtih so opredelili pilotni projekt na manjšem območju, kjer bi se približali realni izvedbi registracije. Vandysheva et al. (2012) poudarjajo pomen samodejne kontrole ob vpisu novih nepremičninskih enot v smislu skladnosti s predhodno postavljenimi pravili. Kljub intenzivnim naporom za vzpostavitev 3D-katastra in modernizaciji sistema zemljiške administracije v zadnjih letih, ta v Rusiji še vedno temelji na dveh prostorskih razsežnostih (Ilyushina, Noszczyk in Hernik, 2017). Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 275 | | 62/2 | GEODETSKI VESTNIK 3.3 aktualne teme in raziskave na področju 3d-katastra TICLES Zadnja delavnica na področju 3D-katastra pod okriljem FIG-a je potekala leta 2016. Da je področje razi-WED AR skovalno aktivno, med drugim kaže povečano število prispevkov (31) glede na predhodno delavnico (25). VIE Izbrani prispevki so v razširjeni različici zbrani tudi v posebni izdaji mednarodne revije ISPRS International Journal of Geo-Information z naslovom R esearch and Development Progress in 3D Cadastral Systems 2017. ANI ČLANKI | PEER-RE Najbolj opazno je ponovno veliko število objav, vezanih na analizo stanja v posameznih državah in različnih možnosti za razvoj sistemov zemljiške administracije, med katerimi prevladujejo objave, ki se RECENZIR osredotočajo na tehnične rešitve modeliranja 3D-neprmičninskih enot v informacijskih sistemih zemljiške administracije (Araújo in Oliveira, 2016; Dimas, 2016; Soon, Tan in Khoo, 2016; Gulliver, Haanen in Goodin, 2017). Na to področje se nanašajo tudi druge objave v mednarodno odmevnih revijah. Tako Aien et al. (2017) izpostavijo šest najbolj uveljavljenih podatkovnih modelov v katastru, kjer je so se zaradi različnosti sistemov zemljiške administracije pojavili številni podatkovni modeli. Med njimi avtorji izpostavljajo tri (LADM, ePlan, ArcGIS Parcel Data Model) in jih podrobneje analizirajo z vidika uporabnosti za 3D-kataster. Podatkovni model 3DCDM, ki ga je v okviru doktorske raziskave razvil vodilni avtor, v raziskavi ni vključen, niti ni v njej omenjen. Avtorji ugotavljajo, da nekateri obravnavani podatkovni modeli omogočajo modeliranje 3D-nepremičninskih enot, a pri vsakem razkrijejo pomembne omejitve za razvoj 3D-katastra. ENSI| Na raziskovalni ravni je spet močneje zastopano pravno področje raziskav (Kitsakis in Dimopoulou, 2017; Vučić et al., 2017), na čelu z obširno primerjalno študijo med izbranimi državami (Kitsakis, Paasch in Paulsson, 2016; Paasch et al., 2016). Kitsakis, Paasch in Paulsson (2016) predstavijo pravno opredelitev 3D-nepremičninskih enot v različnih državah (Avstrija, Brazilija, Hrvaška, Grčija, Poljska in Švedska) in načrte za prihodnji razvoj. Švedska, kot edina od naštetih držav, v pravnem sistemu ne omejuje registracije 3D-nepremičninskih enot, a upravljanje podatkov, tako kot v drugih državah, tudi na Švedskem še vedno temelji na 2D-konceptih. Na področju modeliranja in upravljanja podatkov v 3D-katastrih je zanimiva raziskava Janečka in Součka (2017). Avtorja predstavljata aktualno stanje na širšem področju 3D-geoinformatike, ki zajema koncepte, podatkovne modele, standarde in operacije, povezane s 3D-prostorskimi podatki. Poudarek je na trenutnih zmogljivostih prostorskih podatkovnih baz z vidika modeliranja in upravljanja 3D-prostorskih podatkov. Izredno aktualna tema je povezava oziroma vključevanje podatkov BIM v podatkovni model 3D-katastra. Vse več raziskav se v povezavi s podatki BIM usmerja tudi na modeliranje notranjosti stavb za potrebe registracije nepremičninskih enot (Oldfield et al., 2016; Atazadeh, 2017). Atazadeh et al. (2017) obravnavajo BIM kot osnovo za upravljanje pravic in omejitev na stavbah. Predlagajo razširitev podatkovnega modela tako, da bi ta podpiral tudi vnos podatkov o pravicah in omejitvah v stavbah in njihovo upravljanje. Med drugim se avtorji dotikajo aktualnega vprašanja odnosa 3D-enot pravnega pomena in fizičnega modela stavbe. Poleg prednosti avtorji navajajo omejitve predlaganega pristopa, ki vključujejo institucionalne ovire, preobširno podatkovno strukturo ter problem neusklajenosti med načrtovanim in dejansko zgrajenim objektom. V modeliranje notranjosti stavb je usmerjeno raziskovalno delo Zlatanove et al. (2016), kjer je prvič v okviru raziskav na področju 3D-katastra uporabljen v letu 2014 sprejet standard OGC – IndoorGML | 276 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | GEODETSKI VESTNIK | 62/2 | (2014). V prispevku avtorji obravnavajo možnosti za povezavo navedenega standarda s standardom TICLES LADM. Razširjene raziskave v tej smeri predstavlja Alattas et al. (2017). Standard IndoorGML temelji na večslojnem prostorsko-dogodkovnem modelu, prvotno namenjenem podpori navigaciji v notranjo-WED ARVIE sti stavb, predlaganem v letu 2009 (Becker, Nagel in Kolbe, 2009). Navedena skupina avtorjev je bila vodilna tudi pri razvoju starejšega standarda za modeliranje mest in pokrajin v 3D-okolju CityGML. Standard IndoorGML uvaja celični pristop modeliranja notranjosti stavb, ki je po načelu dualnosti ANI ČLANKI | PEER-RE (Munkres, 1984) povezan z matematičnim grafom. Ta omogoča izvajanje algoritmov iskanja optimalnih poti v podporo navigaciji, pa tudi drugih algoritmov, ki temeljijo na topoloških odnosih med grafičnimi RECENZIR gradniki. Standard omogoča razširitve podatkovnega modela v obliki razširitvenih modulov za različna področja. Kang in Li ( 2017) sta posebej izpostavila možnost realizacije razširitvenega modula standarda IndoorGML za področje 3D-katastra. Povezovanje modelov mest in modelov notranjosti stavb ter pod-zemnih objektov je prepoznan raziskovalni problem tudi v skupini strokovnjakov za področje globalnih prostorskih informacij pri Združenih narodih UN-GGIM (2015). 4 SKleP Začetki uveljavljanja termina 3D-kataster segajo v čas nastanka dokumenta Cadastre 2014 (Kaufmann in Steudtler, 1998). Za opredmetenje zapisanih ciljev je bila leta 2001 izvedena prva delavnica pod EN okriljem FIG, ki je spodbudila raziskave na področju 3D-katastra. Jasneje so se določili različni vidiki SI | razvoja 3D-katastrov: pravni, tehnični in administrativni, od katerih sta prva dva v raziskavah močneje zastopana. Za prvo desetletje je značilno veliko število analiz sistemov zemljiške administracije v posameznih državah in predlogi za njihovo nadgradnjo. Skupne ugotovitve so bile, da je za vzpostavitev 3D-registracije potreben dodaten razvoj na vseh navedenih področjih raziskovanja. Na tehničnem področju raziskovanja 3D-katastrov je velik pečat pustila doktorska disertacija Stoterjeve (2004). Večina raziskav, ki sledijo, se namreč navezuje na njene izsledke, opredelitve in predloge. Na pravnem področju izstopata Švedska, ki je leta 2004 v svoj pravni sistem uvedla možnost registracije 3D-nepremičninske enote, in Avstralija z dolgo tradicijo možnosti registracije samostojnih 3D-nepremičninskih enot. Med najodmevnejšimi študijami stanja pravnih sistemov na področju 3D-katastra je doktorska disertacija Paulssonove (Paulsson, 2007), ki podaja celovit pregled in uvid v pravni vidik 3D-katastra. Dokument Cadastre 2014 je sprožil tudi začetne zasnove podatkovnega modela CCDM katerega namen je poenotenje ključnih sestavin sistemov zemljiške administracije po svetu. Gre za neposrednega predhodnika mednarodnega standarda LADM (2012), ki v osnovi ne omejuje evolucije tradicionalnega 2D-katastra v 3D-kataster. Zadnje desetletje raziskav na področju 3D-katastra zaznamuje predvsem objava standardov LADM (2012) in CityGML (2012) ter tudi IndoorGML (2014). V večini raziskav v tem obdobju se proučujejo možnosti uporabe navedenih standardov, analizirajo prednosti in slabosti posameznih standardov ter se medsebojno primerjajo. Manj raziskav se nanaša na pravni vidik, kar ugotavljata že Paulsson in Pasch (2013). Nizozemski raziskovalci so v tem obdobju pomembno prispevali k uvajanju konceptov 3D-katastra v prakso. Ko se bo za razvoj katastra v smeri podpore tretji razsežnosti odločilo več držav, se bodo okrepile tudi raziskave. Raziskovalni izzivi na področju 3D-katastra se nadalje nanašajo na integracijo podatkov iz drugih domen, predvsem so pri tem aktualne raziskave glede uporabe oziroma vključevanja Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | | 277 | | 62/2 | GEODETSKI VESTNIK podatkov BIM v podatkovni model 3D-katastra in nasprotno. Dodatno je aktualna obravnava oziroma TICLES modeliranje notranje strukture stavb, ki je zahtevna tako z vidika strukture podatkov in kompleksnosti WED AR podatkovnih modelov kot tudi z vidika pridobivanja podatkov in integracije modelov notranjosti stavb VIE in zunanjih modelov mest ter pokrajin (UN-GGIM, 2015). Zahvala ANI ČLANKI | PEER-RE Pregledni članek je rezultat raziskovalnega dela v okviru raziskovalnega programa št. P2-0227 z naslovom RECENZIR Geoinformacijska infrastruktura in trajnostni prostorski razvoj Slovenije, ki ga sofinancira Javna agencija za raziskovalno dejavnost Republike Slovenije (ARRS) iz državnega proračuna. viri in literatura Glej literaturo na strani 263. ENSI| Tekavec J., ferlan M., Lisec A. (2018). pregled raziskav na področju 3D-katastra nepremičnin. Geodetski vestnik, 62 (2), 249-278. DOI: 10.15292/geodetski- vestnik.2018.02.249-278 Asist. Jernej Tekavec, univ. dipl. inž. geod. Izr. prof. dr. Anka Lisec, univ. dipl. inž. geod. Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo Jamova cesta 2, SI-1000 Ljubljana Jamova cesta 2, SI-1000 Ljubljana e-naslov: jernej-tekavec@fgg.uni-lj.si e-naslov: anka.lisec@fgg.uni-lj.si Viš. pred. dr. Miran Ferlan, univ. dipl. inž. geod. Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo Jamova cesta 2, SI-1000 Ljubljana e-naslov: miran.ferlan@fgg.uni-lj.si | 278 | Jernej Tekavec, Miran ferlan, Anka Lisec | pREGLED RAZISKAV NA pODROčJU 3D-KATASTRA NEpREMIčNIN | A REVIEW Of RESEARCh ON 3D REAL pROpERTy CADASTRE | 249-278 | PRILOGA B Indoor space as the basis for modelling of buildings in a 3D Cadastre Tekavec, J., Čeh, M., Lisec, A. (2020) Survey Review DOI: 10.1080/00396265.2020.1838761 Faktor vpliva za leto 2019: 1,660 (Q3) Survey Review ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ysre20 Indoor space as the basis for modelling of buildings in a 3D Cadastre Jernej Tekavec , Marjan Čeh & Anka Lisec To cite this article: Jernej Tekavec , Marjan Čeh & Anka Lisec (2020): Indoor space as the basis for modelling of buildings in a 3D Cadastre, Survey Review, DOI: 10.1080/00396265.2020.1838761 To link to this article: https://doi.org/10.1080/00396265.2020.1838761 © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group Published online: 11 Dec 2020. Submit your article to this journal Article views: 197 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=ysre20 Indoor space as the basis for modelling of buildings in a 3D Cadastre Jernej Tekavec ∗, Marjan Čeh and Anka Lisec This paper presents a 3D cadastral data model for buildings. A review of the relevant research shows that a common concept in the 3D cadastre domain is using the legal building unit, i.e. real property unit, as the core modelling unit. Alternatively, this study proposes using indoor space as a core modelling unit. The main reason is to enable the efficient integration of cadastral data with the data from other domains. On the conceptual level, the model is linked to the Land Administration Domain Model (LADM). The integration options are studied for three international standards: IFC, CityGML and IndoorGML. Keywords: 3D cadastre, indoor space, integration, BIM, CityGML, IndoorGML Introduction Modelling of buildings and its features represents one of the main driving forces for the introduction of 3D Within a cadastral system, i.e. a land administration sys- spatial data, not only in the land administration domain, tem, a large volume of relatively detailed, well-structured but also in city and landscape modelling, and AEC. Every and high-quality geospatial data are stored and managed listed domain has its particularities in terms of require- that have applications beyond the land administration ments, constraints, rules and solutions. Consequently, domain, e.g. in forestry, agriculture, and spatial planning. the data on buildings from different domains can vary a In the last two decades, the tendency towards the lot in many aspects – not only in data format but also inclusion of a vertical dimension in cadastral systems in quality, detailedness and completeness, semantics and has been evident in many countries worldwide and is gen- object definitions. These challenging gaps between build- erally characterised by the term ‘3D Cadastre’ (FIG 3D ing data models are the basis for many current research cadastres 2019). An upgrade of a traditional 2D cadastral activities attempting to link or integrate the 3D Building data model to 3D can make the stored data more useful Information Modelling data, i.e. BIM data, and 3D geos- for new and existing applications in many (also new) patial data from Geographic Information Systems, i.e. domains that are related to land administration. One of GIS data. Deng et al. (2016) propose an instance-based the options to achieve this is to design the 3D cadastral method for mapping between CityGML (OGC 2012) data model to allow cross-domain integration. and Industry Foundation Classes (IFC) (ISO 2018) sche- In the context of this study, the term cadastral data mas. Liu et al. (2017) provide a state-of-the-art review of refers to the data that specifies the spatial extent of real GIS-BIM integration methods. Ohori et al. (2018) present property units, which are a subject for the registration practical results of GIS-BIM integration project. The of rights, restrictions and responsibilities (RRR). cadastral systems can significantly benefit from cross- Although cadastral data modelling is closely related to domain data integration capability. The first gain is the the legal framework and initial real property registration, provision of high-quality AEC data, i.e. BIM, as input we address the 3D cadastral modelling of real property data for real property unit formation within a land admin- units concerning the object’s physical characteristics in istration system, and the second is the increased potential this work. The approach is not a novelty in the cadastre for linking and integrating cadastral data with other data domain. Various aspects of 3D cadastral data modelling sources to improve existing and to design new geospatial are summarised and discussed in van Oosterom et al. applications. (2018). Among the others, the authors state that in most The main aim of this study is to develop a cadastral cases, like in 2D cadastre, the ownership of a 3D parcel data modelling approach for buildings that uses indoor implies the ownership of all physical objects that are space as the core modelling object instead of a building located within the defined space. From this perspective, part or real property unit. This idea is communicated 3D data on physical space can be used to describe physical through the design and implementation of a 3D cadastral reality in cadastre, which should be related to the ‘legal data model. We address all types of buildings, where the reality’ (van Oosterom et al. 2018). division of real property is relevant (e.g. residential, com- mercial, industrial and their combinations). Apart from providing the reference, the paper does not investigate the concepts of 3D modelling of non-physical legal spaces Faculty of Civil and Geodetic Engineering, Department of Geodesy, Univer- sity of Ljubljana, Jamova cesta 2, Ljubljana SI-1000, Slovenia (e.g. legal spaces not related to buildings). It has to be ∗ emphasised that various cadastral systems have been Corresponding author, email: jernej.tekavec@fgg.uni-lj.si © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Received 27 March 2020; accepted 14 October 2020 DOI 10.1080/00396265.2020.1838761 Survey Review 2020 1 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre developed worldwide, with different origins and purposes. space. The model enables efficient modelling of complex The main aim of our study has been to provide a concept situations, especially in the built environment. It provides for 3D cadastral building modelling based on the physical various types of both legal and physical objects and their characteristics of buildings. For its implementation, a geometric representations. Although the model solved vir- relation to the legal space has to be defined in a selected tually all possible complex situations, Aien et al. (2015) jurisdiction, but this is beyond this study. Although the acknowledged its limitations, especially in the integration building’s indoor spaces as physical features may not be and implementation aspects. Knoth et al. (2018) aimed to relevant for managing all RRRs, they can be used to inte- design a building model by the identification of common grate cadastral data with the data from other domains. elements among the selected building models. By extend- In the next section, we present a synthesis of the ing the core model, the authors provided a feasible research related to the data models that allow the realis- model for a 3D cadastre that integrated physical and ation of a 3D cadastre and we identify their key relevant legal aspects of a building. Li et al. (2016) studied the inte-features. In the following section, we design a conceptual gration of CityGML and LADM with a focus on condo- 3D cadastral model, which we link to the standardised minium units in buildings based on the application Land Administration Domain Model (LADM) (ISO domain extension (ADE) for CityGML, which was also 2012). this is followed by the implementation in a spatial proposed by Góźdź et al. (2014) and Rönsdorf et al. DBMS (Database Management System). Based on the (2014). The common feature of all presented the 3D cadas- data model, we investigate the integration capabilities tral data models is the existence of a legal building unit, i.e. with data that corresponds to IFC, the CityGML stan- real property unit, that represents a core modelling object dard and IndoorGML standard (OGC 2014). for a 3D cadastre, either by explicit modelling in the model or linking to LADM classes representing the legal units. 3D cadastral data models Recently, Rajabifard et al. (2018) identified the relevant spatial information models and evaluated their ability to The term 3D cadastre is relatively broad and incorporates model legal interests and boundaries in Victoria, Austra- technical, legal and organisational aspects of 3D cadas- lia, with a focus on the built environment. The authors tral systems. At the international level, the research con- classify the cadastral data models to legal, physical and cerning all those aspects is concentrated mainly under integrated ones that combine legal and physical aspects the FIG working group on 3D cadastres. During the in one model. They have emphasised the need for further last two decades of intensive research, significant progress research if indoor spaces are used to define the geometry has been achieved in all the aspects, resulting in the recent of legal spaces within a 3D cadastre. The topic is especially comprehensive FIG publication (van Oosterom 2018) challenging with respect to, among others, the multi-pur- that also presents the data models related to a 3D cadastre pose cadastre and spatial analytics, for instance in relation with an emphasis on the LADM standard data model. to indoor navigation as proposed by Alattas et al. (2020) LADM is designed to provide a comprehensive data and Tekavec and Lisec (2020). This study aims to address model for land administration. It provides an inter- this need and contribute to the development and national framework for the cadastral data modelling but implementation of the idea of linking physical indoor does not prescribe the modelling approach and technical spaces with legal spaces in the context of a 3D cadastre. data formats. Its core consists of linked packages related to parties, RRR, basic administrative units and spatial Materials and methods units. These packages represent the core of a land admin- istration system. Lemmen et al. (2015) emphasise that The 3D cadastral data model presented in this section is LADM is designed to represent legal space and enables based on the result of the synthesis of the cadastral data 3D representation. Registration of physical spaces is models and considering the data model design objectives beyond the scope but, as the authors argue, investigation as presented below. As an alternative to other 3D cadas- of LADM should be related to other geo-information tral data models, we focus on a building’s indoor space standards (CityGML, LandXML, IFC). as the core spatial unit. In this study, we define indoor The absence of physical space representation in LADM space as the space that is bounded by the inward-facing gave rise to research activities studying the relationship three-dimensional surfaces (interior surfaces) of walls, between physical and legal space in the context of 3D floors, ceilings and other structural parts of the building. real property registration (Paasch et al. 2016, Larsson These spaces can be represented by 3D volumetric geome- et al. 2020) as well as various ways of physical data mod-tries, i.e. solids. At the passages between two indoor elling and visualisation to source legal spaces for 3D spaces, the solid geometries touch each other (Fig. 1) aim-cadastres (Shojaei 2014), especially for buildings. The ing to integrate topologic relations between neighbouring LADM forthcoming revision is strongly considering link- spatial units. ing physical and legal objects, linking outdoor and indoor The boundaries of a real property unit (legal abstrac- models, and is promoting the integration with several tion of the space) in the buildings may differ between jur- encodings (BIM/IFC, GML, CityGML, LandXML, isdictions, where we should distinguish, for example, the IndoorGML etc.) (van Oosterom et al. 2019). El-Mekawy definition of a legal boundary which can coincide with and Ostman (2012) argued that neither of the existing the inward-facing or outward-facing surfaces of walls, models allowed modelling of a 3D cadastral system and floors, ceilings, and other structural parts of the building, proposed an extension of the Unified Building Model or it might be defined in the middle of the wall, etc. (see (El-Mekawy 2010) to make it feasible for application in also Cemellini et al. 2020). However, the possession or 3D cadastral systems. The 3D cadastral data model occupation of the physical space in a building is mainly (3DCDM) developed by Aien (2013) was intended to related to indoor space. This makes it logical to focus address the complex relations of legal and physical on indoor spaces as core spatial units when modelling a 2 Survey Review 2020 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre registration of buildings that is compatible with traditional 2D parcel-based cadastral systems. (1) The cadastral data model for buildings should not require storage and maintenance of the data outside the cadastral domain. (2) The cadastral data model should provide data that is structured in a way that enables as wide a cross-domain integration as possible on the data input and output sides. (3) The cadastral data model should differentiate build- ing data from land parcel data (e.g. for buildings, it should provide a separate (but integrated) data model or separate object classes) aiming to provide a step- by-step development of a 3D cadastre within a tra- ditional 2D data model. (4) The implementation should be feasible in a database management system (DBMS). The first objective can be perceived as contradictory to our decision to focus on indoor spaces, but as it is stated in Figure 1 The passages between indoor spaces are realised the justification for this decision, indoor spaces represent by touching surfaces (green) the main spatial units to which RRRs are related and pro- vide less ambiguity in the representation of a legal situ- building for RRR registration. Therefore, we define the ation in a building. However, following the first real property unit as a group of indoor spaces (Fig. 2). objective is challenging, as what belongs in the cadastral The model is built on the presumption that for buildings, domain depends on the jurisdiction. Lemmen et al. the indoor space represents a key object of interest for (2015) provide a list of fields related to but outside cadastral systems. The extension of these ‘basic spatial LADM, according to which physical registration of build- units’ can be applied in the sense of adding ‘wall solids’ ings is out of the LADM (cadastral) domain. On the other (the whole width, half of the width, etc.). hand, much research shows that a very strong relationship In addition to being a feasible solution for the defi- exists between physical and legal space in the built nition and geometric representation of a legal situation environment (Aien et al. 2015, Li et al. 2016, Knoth in the building, the indoor space also represents the et al. 2018). Larsson et al. (2020) study the conversion most integrative object for cross-domain integration of of 2D analogue cadastral boundary plans into 3D digital building information with 3D cadastral data (Knoth information and discuss the integration with BIM. Raja- et al. 2018). bifard et al. (2019) suggest that cadastral systems cannot and should not ignore the physical space, especially considering the second objective, which emphasises the Data model design multi-purpose role of the cadastre. The terms ‘physical The development of the proposed cadastral data model space’ and ‘physical boundary’ used in this study refer has been guided by the objectives, identified by the to the real world and its physical features, while ‘legal authors, that are presented below. We believe that fulfill- space’ and ‘legal boundary’ refer to abstract space and ing these objectives is among the most important prere- its features that have legal meaning. There seem to be quisites to ensure an efficient system for 3D cadastral two options for cadastral systems regarding consideration of physical space. (1) Consider legal space in the cadastral model, as suggested by LADM (Lemmen et al. 2015): this way the first objective is fully met, but on the other hand, we lose the connection to physical space, which is cru- cial to have cross-domain data integration capabilities that constitute the essence of the second objective. (2) Use an integrative approach, based on several recent studies (Aien et al. 2015, Atazadeh et al. 2017, Oldfield et al. 2017, Thompson et al. 2017, Knoth et al. 2018, Atazadeh et al. 2019, Sun et al. 2019) following the second objective and make a trade-off by storing and maintaining data out of the cadastral domain. However, there is a third, alternative approach, as pro- posed and used by this study. Instead of using one of the approaches presented above, we constrain the legal spaces with physical boundaries, giving them physical character- istics. More precisely, we constrain the boundaries of legal spaces (i.e. legal boundaries) to inward-facing surfaces of Figure 2 Two real property units (red and green) rep- physical features (i.e. physical boundaries) that enclose resented by groups of spatially linked indoor indoor spaces. Many studies comprehensively deal with spaces various possibilities of legal boundary position relative Survey Review 2020 3 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre Figure 3 Options for real property unit boundary placement to physical objects in the building (middle of the wall, wall Standards (IPMS 2020) that are developed to overcome exterior or interior surface), which in essence depends on the differences between countries regarding the rules for each country’s legislation (Paulsson 2007). Wang (2015) measuring the buildings. In general, if the real property studies the boundary placement from a 3D visualisation unit boundary is defined by the wall interior surfaces, it perspective, Atazadeh et al. (2017) and Rajabifard et al. does not contain any bounding walls. In contrast, if (2019) discuss the boundary placement in relation to defined by the wall exterior, it contains all bounding BIM. Therefore, this approach seems to go against the walls. Besides these two cases, we add a case where the established knowledge and thus requires further boundary is defined in the middle of the wall or anywhere justification. between the inward-facing or outward-facing surfaces of This paper does not intend to argue nor deal much with walls; what’s more, the boundary might be defined also legal aspects of 3D cadastres, but rather provide a techni- as a buffer zone – in all these cases an indoor-space can cal view on this matter. There is no doubt that in theory, a still be used as a core spatial unit which is extended as legal boundary can be established anywhere in space, not defined by the law. related to any physical features. However, when it comes Most jurisdictions worldwide use condominium (Pauls- to registering RRRs in buildings, the physical features son 2007) to manage the RRRs, which is inherently con- become more important, which is also reflected in the nected with co-ownership since for buildings there are number of studies dealing with this matter. Therefore, a certain spaces that need to be owned by all individual legal boundary can lie either on the wall exterior, interior owners (land parcel, common spaces and installations, or somewhere in between. Atazadeh et al. (2017) discuss etc.). The exterior and middle wall boundaries open up these options while studying the purely legal, purely phys- several issues and dilemmas regarding RRR management ical and integrated approaches for managing the RRRs in the building (ownership is used as it represents the most for buildings. The purely legal approach excelled in visu- important of the RRRs): alisation and querying performance, but proved to be (1) If a boundary of a real property unit is defined in the inappropriate for communication of boundaries relative middle or exterior side of the wall, the wall is owned to physical structures. Our model minimises this only by some of the owners. If the wall is statically deficiency by constraining legal spaces to physical struc- important, this concerns all individual owners in the tures. Fig. 3 aims to bring a common understanding of building. the terms wall interior, wall exterior and the middle of (2) If a boundary of a real property unit is defined in the the wall for real property unit boundary placement in middle of the wall, how is the ownership determined for the context of this study. The same concepts can be outer walls that delineate the building interior and found in the International Property Measurement exterior? Figure 4 Different variants of exterior boundary placement 4 Survey Review 2020 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre Figure 6 An example of the building geometry with legal amendment (balcony space) Figure 5 Various sizes of legal units, presented in different the third objective, the model is designed to leave the tra- colours, for the same apartments (boundaries ditional 2D parcel-based cadastral system mostly between real property units lie in the middle of unchanged. For integration with 2D parcels, the ground the wall) contact area or maximal extent area (or both) of the building can be used. Depending on legal regulations, (3) If a boundary of a real property unit is on the these areas are restricted to one parcel, form a parcel or exterior side of the wall, how is a boundary in complex can be independent of the parcel structure and only pro- situations defined (Fig. 4)? vide the information about the spatial extent of the build- (4) If a boundary of a real property unit is defined in the ing. These approaches are well established in most middle or exterior side of the wall, how is the ownership cadastral systems worldwide. of various installations in this wall determined? The pure concept of using indoor spaces and building (5) If walls have a various thickness and the boundaries external geometry, enclosing indoor spaces, becomes pro- of a real property unit are defined in the middle or blematic when dealing with semi-indoor and semi-out- exterior of the wall, identical physical units have differ- door spaces that are partly connected to a building, but ent sizes of their respective real property units (Fig. 5). cannot be characterised as indoor space (Fig. 6). There Should the subject of their ownership be different? are countless variants of these spaces (balconies, covered (6) If a boundary is in the middle or exterior of the wall, or semi-covered terraces, atriums, etc.). Yan et al. (2019) is it the same with slabs? The slabs provide structural provide a comprehensive study on this matter from a navi- stability for the entire building and thus also concern gation perspective that illustrates that no clear boundary all individual owners in the building. exists between indoor and outdoor space. The determi- nation of what is part of a building and what is not is a All listed dilemmas suggest that individual owners can- general dilemma when dealing with RRR registration not fully exercise their ownership rights on the building on buildings. Our proposed model can be used to register structural parts but only on indoor spaces that are pro- these spaces in a similar way as indoor spaces in Atazadeh vided and made functional by the building structure. et al. (2017). There are several legal solutions, for example, the defi- Following the fourth objective, we have designed the nition of the object to which RRR refer in a contract, implementation of the proposed model in the PostgreSQL where building’s structural parts might also be included. DBMS with PostGIS and SFCGAL extensions. The data- However, if we use interior wall surfaces as boundaries base implementation represents a basis that can be further of real property units, we avoid the listed dilemmas and extended according to the specific needs of each jurisdic- issues. This represents a similar concept as the IPMS 3C tion. The selected DBMS supports 3D spatial data types, variant for measuring the buildings in the International including solid geometries. Additionally, it offers func- Property Measurement Standards (IPMS 2020). Interior tions that support the stored 3D data management and wall surfaces enclose indoor spaces which are used as a analyses. core spatial unit in our data model. The difference from the IPMS 3C is that the geometries of indoor spaces touch each other at passages, which means they are not Data model concept exclusively bounded by physical structures. Based on the objectives, we develop the data model con- Apart from already mentioned indoor spaces, the exter- cept (Fig. 7), defined by the main entities, their relations nal building, can be characterised as an important entity and their geometric representations. Compared to the for registration of RRRs on buildings. The external build- data models that are referenced in section 2, our model ing geometry should enclose all indoor space geometries. can be perceived as basic. The reason for this is that this It represents a key feature that delineates the building model is used to present and discuss the idea of using from the traditional continuous land parcel structure indoor spaces as the core cadastral data modelling entity, found in most land administration systems. Following not to provide a complete and all-inclusive data model. Survey Review 2020 5 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre Figure 7 The concept of the proposed cadastral data model for buildings 3D building modelling for cadastral purposes is the Despite the differences, the proposed data model is main focus of this paper. However, the presented concept partly compliant with the 3DCDM data model (Aien also includes the relation to the land parcels and the 2013) and the extended Core Model for 3D cadastre pro- optional relation to the 3D legal spaces, which are not posed by Knoth et al. (2018). Both models use an integra- defined by physical features. The proposed concept tive approach and represent physical features using assumes that each building is associated with at least multiple entities. However, if the same concept of using one real property unit, but this depends on the jurisdiction indoor spaces to define the real property units is used, and can be altered. For buildings where no condominium the data models would be similar to the one that is pro- is established, only basic data or no data is collected. posed in the paper. The data model is aligned with the LADM (Fig. 8). RealPropertyUnit is related to the LA_BAUnit class and IndoorSpace to LA_SpatialUnit and its subclass Results LA_LegalSpaceBuildingUnit. However, LADM expli- citly denies the restriction of legal spatial units to the DBMS implementation building’s physical structures (Lemmen et al. 2015), A DBMS represents the technical backbone of cadastral which differentiates the two models. As the focus of the (information) systems worldwide. It provides an efficient proposed data model is on indoor spaces, the solid rep- way for secured storage and maintenance of cadastral resentation of the geometry is selected as the most appro- data as well as for exploitation of cadastral data and infor- priate. It facilitates 3D representation, spatial analyses mation. Each new solution or upgrade of cadastral sys- and provides volumetric information (Rajabifard et al. tems should, therefore, include or be supported by an 2019), while LADM provides geometric representation advanced DBMS. Spatial data storage and maintenance of 3D geometries by boundary faces. are supported by the majority of DBMSs considering the ISO SQL/MM-Part 3 (ISO 2016) or OGC Simple fea- ture access (OGC 2010), providing spatial data types, spatial indexes and operations that allow the geometry of the objects to be stored alongside their thematic attri- butes. The proposed cadastral data model is implemented using open source DBMS PostgreSQL with PostGIS and SFCGAL extensions, which support storage of polyhe- dral surfaces and solids, and offers functions to perform 3D operations. Three main tables (Fig. 9) are used to represent three core features of the presented concept (Build- ing, RealPropertyUnit and IndoorSpace). Although the selected DBMS supports storage of solids, the internal holes are not fully supported. The polyhedral surfaces can be stored with internal closed boundaries that form holes. However, 3D operations require solid geome- tries, obtained by the ST_MakeSolid. The polyhedral sur- Figure 8 The relation of the proposed cadastral data model faces with an internal hole(s) cause the ST_MakeSolid (white) and LADM (green) classes function to crash with an invalid geometry error. This is 6 Survey Review 2020 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre Figure 9 DBMS implementation of the proposed cadastral data model not a problem in our case, as indoor spaces are bounded by instruments for public authorities to obtain accurate, physical structures. A hole in the indoor space would mean structured, relatively detailed spatial and thematic data the physical structure would be detached and would about buildings, including data about the indoor environ- ‘hover’ in space. The hole in indoor space could be created ment, which is not accessible through remote sensing only by another indoor space that belongs to the other real technology. Cadastral data has often been used for mul- property unit in the same physical indoor space. Similarly, tiple purposes beyond its core one, partly so that the the building external volumetric representation does not high costs of establishing and maintaining the system need internal holes, as an internal hole would still be can be spread and justified, and partly because it has indoor and not outdoor space. Despite this, we present been the only available data source. If the cadastral data the extension of the basic implementation that can accom- is structured so as to enable integration with data from modate holes in 3D solids. The database schema presented other related domains, it can significantly increase its in Fig. 10 introduces two new tables that can store several potential applications and consequently, its importance solid objects for one Building or indoor space represen- and value. Therefore, one of the most important objec- tation. The hole Boolean type attribute indicates whether tives of our research has been to develop a cadastral the geometry represents a hole. data model for buildings that can be integrated with the The CityGML standard widely introduced the concept dominant standards relating to 3D building modelling. of modelling of spatial entities in a 3D environment with The integration or at least linking is important to the multiple levels of detail (LOD). There are several benefits data input side, i.e. to obtain data for registration from of this approach, such as efficient visualisation and data various data sources, and to the data output side, i.e. to manipulation, and efficient spatial and other analyses include or use the 3D cadastral data in other models that can also be identified as important by cadastral auth- and increase the cadastral data usage/application orities. The extension (Fig. 11) can accommodate mul- potential. tiple levels of detail for both indoor spaces and outdoor geometry by adding an attribute that identifies the corre- Integration with IFC sponding level of detail for each geometry. The concept of The importance of integrating land administration pro- multiple LODs is useful for cadastral authorities to store cesses and BIM is increasing with the rapid adoption of additional, more detailed data to further clarify the RRR BIM in the AEC industry. Several studies have already ana- situation in the building. lysed various options for the integration of BIM and cadas- tral data (El-Mekawy and Ostman 2012, Liu et al. 2017, Data model integration Oldfield et al. 2017, Rajabifard et al. 2018, Atazadeh Registration of a building in the cadastral system, i.e. land et al. 2019, Rajabifard et al. 2019). All studies identified administration system, represents one of the few available the IfcSpace as the most important class for the integration Figure 10 DBMS implementation allowing storage of solids with holes Survey Review 2020 7 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre Figure 11 DBMS implementation allowing storage of multiple levels of detail of IFC with RRR data. Atazadeh et al. (2019) propose an and semantic complexity. An interesting data model spe- approach that uses IfcSpace and IfcZone classes and is cification as an extension of LODs for detailed building aligned with our modelling approach. The authors estab- modelling was proposed by Biljecki et al. (2016), where lish a link with LADM classes, by linking IfcSpace with a set of 16 LODs focused on the grade of the exterior geo- the LA_LegalSpaceBuildingUnit class and IfcZone with metry of buildings, while the indoor space was not dis- LA_BAUnit (Fig. 12). This indicates that our proposed cussed. Since the introduction of CityGML 2.0 in 2012, cadastral data model can be integrated with IFC data. the standard has also been intensively studied from a 3D cadastral perspective. Çagďaş (2012) proposed a Integration with CityGML CityGML extension for property taxation. Góźdź et al. CityGML is the dominant standard for 3D topographic (2014) proposed an ADE for the CityGML standard to modelling in the geospatial domain. The CityGML stan- link it with LADM classes, which is further studied by dard defines five LODs that increase in their geometric Li et al. (2016). The authors propose linking the Figure 12 The concept of the integration of the proposed data model with the IFC, CityGML and IndoorGML standards 8 Survey Review 2020 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre LADM LA_LegalSpaceBuildingUnit class with a legal the borders of the real property units to coincide with object class that is related to the BuildingPart and Build- the borders of indoor spaces. Therefore, the proposed ing classes. Ying et al. (2017) use CityGML LOD 3 data model cannot be directly introduced and models to construct volumetric objects, suitable for use implemented without amendments of the legal framework in 3D cadastres. or the proposed data model. The proposed data model As the core element of our model is indoor space, the can be amended with an additional entity, which rep- corresponding class in CityGML is the Room class. resents the building’s physical features. These features Like the indoor space in our model, the Room class has are linked to the corresponding real property unit the a solid geometry representation (Fig. 12). However, the same way as indoor spaces. Using this amendment, the Room class can be included only in the LOD 4 model, model can support all types of boundaries of real property where very detailed physical structures are required. Up units within the buildings, while still allowing the pre- to LOD 4, the CityGML standard does not provide an sented geospatial data integration options. On the other option to model the building interior. Boeters et al. side, the legal frameworks can also be amended to (2015) emphasise the need to include indoor geometries increase the compatibility with the presented data to lower levels of detail. The building geometry, on the model. These amendments are mostly related to the defi- other hand, can be better integrated with CityGML. nition of the RRRs on the physical features of the build- Unlike the Room class, the _AbstractBuilding class is ing (walls, slabs, etc.). One option is to define that all the not restricted to specific levels of detail. physical structures of the building are owned by all the owners, if not additionally defined differently. Another option is to define that the physical structures between Integration with IndoorGML two indoor spaces of the same real property unit also Of the three standards considered in this section, the belong to this real property unit. The physical structures IndoorGML has the simplest data model. To avoid dupli- between two indoor spaces of different real property cations, the standard allows linking with IFC and units can be defined as owned by both owners, or CityGML. The main role of IndoorGML is to provide owned by each owner to the middle. The data model a data model for indoor navigation purposes. Zlatanova can also be modified to allow the modelling of only the et al. (2016) and Alattas et al. (2017) studied the link outer boundaries of the real property units. The geometry between LADM and IndoorGML to provide additional of the real property unit can be defined as a union of all information about indoor spaces which can improve adjacent indoor spaces belonging to the same real prop- indoor navigation performance. Rajabifard et al. (2018) erty unit, also containing all physical structures of the discussed the options for linking IndoorGML with building. In case the indoor space geometries would be LADM and concluded that IndoorGML cells could be sourced from the BIM entities, this union should be used to define the geometry of legal spaces. To realise done additionally using 3D modelling software. If the this, the authors emphasise the need for investigation of indoor spaces are modelled based on measurements, complex 3D spatial analyses. Our approach is very simi- then all the spaces of the same real property unit can be lar, and we addressed the need for 3D spatial analyses joined into one 3D geometry representing the extent of by implementing the model in PostgreSQL with PostGIS the real property unit. However, this would reduce the and SFCGAL extensions that provide state-of-the-art options for integration with other data models and conse- DBMS support for 3D spatial operations. quentially the options to use the cadastral data for other As in the proposed model, the core element of the purposes. IndoorGML data model is the indoor space (CellSpace One of the challenging topics regarding RRR regis- class), which makes the two very aligned and interoper- tration on buildings and parts of buildings within a 3D able in this aspect (Fig. 12). Following the duality con- cadastre is the required geometric accuracy as well as cept, the IndoorGML data model contains navigation the level of detail of the geometric data model. This graph elements – nodes and edges (State and Transition research does not deal with this issue apart from allowing elements), which are not included in the proposed cadas- storage of multiple levels of detail. If the boundaries of tral data model. The indoor space geometries in the pro- legal spaces coincide with physical boundaries, the posed model touch each other on the surfaces that required accuracy is not as high as it is for boundaries represent the passage between them (doors and other with no physical counterpart. In the future, the BIM openings). This makes it possible to derive the indoor models will provide very detailed data on the building’s space topology from geometries using 3D operations physical elements for more and more buildings. The inte- available in the selected DBMS. If the physical indoor gration, as presented in this study, will therefore be very space is divided into several touching indoor spaces beneficial. The current problem is that many IFC models belonging to different real property units, these can be do not provide as-built data, essential for cadastral used to perform subspacing of IndoorGML CellSpace registration. entities. The developed data model is based on indoor spaces that are touching each other at passages, which means Discussion that 3D geometries of indoor spaces have shared faces. The proposed database implementation uses the polyhe- The legal frameworks that regulate the registration of dral surface to represent each indoor space, which RRRs on buildings differ very much among the countries means the ‘touching’ faces are duplicated. Consequently, and jurisdictions. The data model is focused on the defi- the required storage space is increased. Additionally, the nition of real property units using indoor spaces, indepen- duplication can cause the data to be inconsistent. How- dently on the legal definition of a real property unit. As ever, this approach has several advantages over storing already mentioned, the legal regulations do not require 3D geometries using the topological data structure. Survey Review 2020 9 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre (1) The 3D geometries can be managed by the available contribution of this study can also be perceived in the database 3D functionalities. presentation of a complete design process from initial (2) We have direct access to 3D geometries, without the idea to DBMS implementation. The aim of the study need to construct 3D geometries using lower-dimen- has not been to reject all the proposed 3D cadastral sional geometrical features and their topological models and research in this field, but to provide an relationships. alternative data modelling approach. The data model is (3) The 3D geometries can be stored in line with other designed to be aligned with the LADM standard and attribute data. standards related to physical modelling of buildings. The inconsistencies should be avoided by automatically This means it enables and facilitates the integration of checking the data at insertion and each manipulation of cadastral data with the data from other related domains, geometry. The PostgreSQL DBMS with PostGIS and which is becoming increasingly important. It can be con- SFCGAL extensions has functionalities to implement sidered and applied by the cadastral authorities in revising some data validation as an overlapping check also in 3D. or designing new cadastral data models with objectives The open question for cadastral building registration is aligned with the ones presented in this paper. which buildings to include. While in many countries detailed cadastral registration is foreseen only for build- Disclosure statement ings where a condominium is to be established, some countries, e.g. Slovenia and Sweden (see Drobež et al. No potential conflict of interest was reported by the author(s). 2017, Larsson et al. 2020), register all buildings, mainly for efficient real property valuation that serves many gov-Funding ernmental applications. In this case, the indoor spaces have to be measured to obtain the area information. This work was supported by Javna Agencija za Raziskovalno Dejavnost With some additional vertical measurements, it is possible RS: [Grant Number P2-0406 Earth observation and geoinformatics]; Minsitry of Environment and Spatial Planning of the Republic of Slove- to obtain enough information for 3D modelling of the nia (Surveying and Mapping Authority of the Republic of Slovenia): indoor spaces according to the proposed data model. [Grant Number V2-1934]. This also opens new possibilities for valuation as we obtain not only area but also the volume of the indoor spaces. Notes on contributors For the buildings with no established condominium, all Jernej Tekavec graduated from the University of Ljubl- indoor spaces can be grouped into one real property unit. jana, Faculty of Civil and Geodetic Engineering (UL Following this concept, also these buildings can be mod- FGG) in the field of geodesy. Currently, he is employed elled according to the proposed data model. If the cadas- as assistant researcher at the Chair of Geo-informatics tral system registers only the buildings with multiple real and Real Estate Cadastres, UL FGG, and is a PhD can- property units, only some basic data may be registered didate at the same faculty. His research is focused on real about the other buildings (e.g. external geometry) or estate cadastres, 3D city and indoor modelling, and 3D they can be entirely left out. The buildings having multiple spatial analyses. He is actively involved in current real property units can be modelled according to the pro- national and international projects and activities of the posed data model. chair. As an early stage researcher, he has also partici- However, it should be noted that the developed concept pated in various international workshops, forums and and data model is in accordance with the current practices conferences from the geospatial and land administration of RRR registrations on parts of buildings, e.g. condomi- domains. niums, based on floor plans. Marjan Čeh Storing detailed data on the interior of buildings can be is assistant professor at the Chair of Geo- problematic from a privacy and security aspect. This can informatics and Real Estate Cadastres at the University be efficiently managed by making parts of the data of Ljubljana, Faculty of Civil and Geodetic Engineering (indoor spaces) available only to authorised users. Given (UL FGG), Slovenia. His main research interests are in that the proposed model offers data that can be used for the field of semantic interoperability and ontology, geo- many applications, including public safety, crisis and dis- metric improvement of cadastral maps, strategic develop- aster response by police, firefighters, ambulance and other ment of Land Administration Systems, analyses of spatial first responders, the benefits should outweigh the data infrastructure, mass appraisal of real estate. He is in concerns. charge of several projects financed by the Surveying and Mapping Authority of Republic of Slovenia. At the inter- national level, he has contributed to several projects, such Conclusions as COST actions, Tempus projects, ESPON, FP7 and In our study, we look for generic solutions to 3D cadastral Horizon2020 project, and is actively involved in many building modelling that would be suitable for deployment international initiatives, such as EuroSDR, OGC. in various jurisdictions. The idea of using indoor spaces Anka Lisec is associate professor and head of the Chair of constrained to physical features as the core element for Geo-informatics and Real Estate Cadastres at the Univer- cadastral registration of buildings is used for 3D cadastral sity of Ljubljana, Faculty of Civil and Geodetic Engineer- modelling. Using the indoor space as the core spatial unit ing (UL FGG), Slovenia. Her research interests are differentiates the research presented here from other focused on geoinformatics and spatial information sys- related studies that have included indoor spaces in the tems, including land administration systems, with a par- data models. The model design process is thoroughly dis- ticular focus on 3D&4D geospatial data modelling and cussed, with references to the initial objectives used to 3D cadastre. At the international level, she is actively steer and guide the model design process. The involved in many international exchange programmes, 10 Survey Review 2020 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre associations and initiatives in geospatial and land admin-ISO, 2012. ISO 19152:2012 – geographic information – land adminis-istration domains. With her research, coordinating and tration domain model (LADM). Geneva, Switzerland: International Standard. knowledge-transfer activities, she has contributed to sev- ISO, 2016. ISO/IEC 13249-3:2016. information technology – database eral international projects, such as COST actions, Tempus languages – SQL multimedia and application packages – part 3: projects, ESPON, FP7 and Horizon2020 project. Beside spatial. Geneva, Switzerland: International Standard. her research work within national and international pro- ISO, 2018. ISO 16739-1:2018. Industry Foundation Classes (IFC) for jects, she is actively involved in the preparation of cadas-data sharing in the construction and facility management industries. Geneva, Switzerland: International Standard. tral legislation as well as in developing strategic guidelines Knoth, L., et al., 2018. Cross-Domain building models—A step towards for the national mapping and surveying service. interoperability. ISPRS International journal of geo-information, 7 (9), 363. doi:10.3390/ijgi7090363 Larsson, K., Paasch, J.M., and Paulsson, J., 2020. The land administration domain model. Land use policy, 98, 104178. doi:10.1016/j. ORCID landusepol.2019.104178 Lemmen, C., van Oosterom, P., and Bennett, R., 2015. The land admin-Jernej Tekavec http://orcid.org/0000-0001-6792-6885 istration domain model. Land use policy, 49, 535–545. doi:10.1016/ Marjan C ̌ eh http://orcid.org/0000-0001-5768-4712 j.landusepol.2015.01.014 Li, L., et al., 2016. 3D modeling of the ownership structure of condominium units. Computers, environment and urban systems, 59, 50–63. References doi:10.1016/j.compenvurbsys.2016.05.004 Liu, X., et al., 2017. A state-of-the-art review on the integration of build-Aien, A., 2013. 3D cadastral data modelling. Thesis (PhD). Melbourne, ing information modeling (BIM) and geographic information sys-Victoria: University of Melbourne. tem (GIS). ISPRS International journal of geo-information, 6 (2), Aien, A., et al., 2015. Integrating legal and physical dimensions of urban 53. doi:10.3390/ijgi6020053 environments. ISPRS international journal of geo-information, 4 (3), OGC, 2010. Open geospatial consortium. Implementation specification 1442–1479. doi:10.3390/ijgi4031442 for geographic information – simple feature access – part 2: SQL Alattas, A., et al., 2017. Supporting indoor navigation using access rights option. Available from: https://www.opengeospatial.org/standards/ to spaces based on combined use of IndoorGML and LADM sfs [Accessed 1 Mar 2020]. models. ISPRS international journal of geo-information, 6 (12), OGC, 2012. Open geospatial consortium. City Geography Markup 384. doi:10.3390/ijgi6120384 Language (CityGML) encoding standard, version 2.0. Available Alattas, A., et al., 2020. LADM-IndoorGML for exploring user move-from: http://www.opengeospatial.org/standards/citygml [Accessed ments in evacuation exercise. Land Use Policy, 98, 104219. doi:10. 1 Mar 2020]. 1016/j.landusepol.2019.104219 OGC, 2014. Open geospatial consortium. IndoorGML, version 1.02. Atazadeh, B., et al., 2019. Querying 3D cadastral information from BIM Available from: http://www.opengeospatial.org/standards/ models. ISPRS international Journal of Geo-information, 8 (8), 329. indoorgml [Accessed 1 Mar 2020]. doi:10.3390/ijgi8080329 Ohori, K.A., et al., 2018. Processing BIM and GIS models in practice: Atazadeh, B., Rajabifard, A., and Kalantari, M., 2017. Assessing per-experiences and recommendations from a GeoBIM project in The formance of three BIM-based views of buildings for communi- Netherlands. ISPRS International journal of geo-information, 7 cation and management of vertically stratified legal interests. (8), 311. doi:10.3390/ijgi7080311 ISPRS international Journal of Geo-information, 6 (7), 198. Oldfield, J., et al., 2017. Working with open BIM standards to source doi:10.3390/ijgi6070198 legal spaces for a 3D cadastre. ISPRS International journal of Biljecki, F., Ledoux, H., and Stoter, J., 2016. An improved LOD specifi-geo-information, 6 (11), 351. doi:10.3390/ijgi6110351 cation for 3D building models. Computers, environment and urban Paasch, J.M., et al., 2016. Building a modern cadastre: legal issues in systems, 59, 25–37. doi:10.1016/j.compenvurbsys.2016.04.005 describing real property in 3D. Geodetski vestnik, 60 (2), 256–268. Boeters, R., et al., 2015. Automatically enhancing CityGML LOD2 doi:10.15292/geodetski-vestnik.2016.02.256-268 models with a corresponding indoor geometry. International journal Paulsson, J., 2007. 3D property rights – an analysis of key factors based on of geographical information science, 29 (12), 2248–2268. doi:10. international experience. Thesis (PhD). Stockholm: Royal Institute 1080/13658816.2015.1072201 of Technology – KTH. Çagďaş, V., 2012. An application domain extension to citygml for immov-Rajabifard, A., Atazadeh, B., and Kalantari, M., 2018. A critical evaluation able property taxation: a Turkish case study. International journal of of 3D spatial information models for managing legal arrangements of applied earth observation and geoinformation, 21 (1), 545–555. multi-owned developments in Victoria, Australia. International journal doi:10.1016/j.jag.2012.07.013 of geographical information science, 32 (10), 2098–2122. doi:10.1080/ Cemellini, B., et al., 2020. Design, development and usability testing of 13658816.2018.1484125 an LADM compliant 3D cadastral prototype system. Land use pol- Rajabifard, A., Atazadeh, B., and Kalantari, M., 2019. BIM and urban icy, 98, 104418. doi:10.1016/j.landusepol.2019.104418 land administration. Boca Raton: CRC Press. doi:10.1201/ Deng, Y., Cheng, J.C.P., and Anumba, C., 2016. Mapping between BIM 9781351032346 and 3D GIS in different levels of detail using schema mediation and Rönsdorf, C., Wilson, D., and Stoter, J., 2014. Integration of land admin-instance comparison. Automation in construction, 67, 1–21. doi:10. istration domain model with CityGML for 3D cadastre. In 1016/j.autcon.2016.03.006 International workshop on 3D cadastres. Dubai. Drobež, P., et al., 2017. Transition from 2D to 3D real property cadastre: Shojaei, D., 2014. 3D cadastral visualisation: understanding user’s require-the case of the Slovenian cadastre. Computers, environment and ments. Thesis (PhD). Melbourne, Victoria: University of urban systems, 62, 125–135. doi:10.1016/j.compenvurbsys.2016.11. Melbourne. 002 Sun, J., et al., 2019. Utilising BIM and GIS for representation and El-Mekawy, M., 2010. Integrating BIM and GIS for 3D city modelling; Visualization of 3D cadastre. ISPRS International journal of geo-The case of IFC and CityGML. Thesis (PhD). Stockholm: Royal information, 8 (11), 5013. doi:10.3390/ijgi8110503 Institute of Technology – KTH. Tekavec, J., and Lisec, A., 2020. Cadastral data as a source for 3D indoor El-Mekawy, M. and Ostman, A., 2012. Feasibility of building infor-modelling. Land use policy, 98, 104322. doi:10.1016/j.landusepol. mation models for 3D cadastre in Unified city models. 2019.104322 International journal of e-planning research, 1 (4), 35–58. doi:10. Thompson, R., van Oosterom, P., and Soon, K., 2017. LandXML encod- 4018/ijepr.2012100103 ing of mixed 2D and 3D survey plans with multi-level topology. FIG 3D Cadastres. 2019. Website of FIG joint commission 3 and 7 work-ISPRS International journal of geo-information, 6 (6), 171. doi:10. ing group on 3D cadastres. Available from: http://www.gdmc.nl/ 3390/ijgi6060171 3dcadastres/ [Accessed 1 Mar 2020]. van Oosterom, P., et al., 2018. 3D cadastral information modelling. In: P. Góźdź, K., et al., 2014. The possibilities of using CityGML for 3D rep-van Oosterom, ed. 3D cadastres best practices. Copenhagen, resentation of buildings in the cadastre. In International workshop Denmark: International Federation of Surveyors – FIG, 95–132. for 3D cadastres. Dubai. Chapter 3. IPMS, 2020. International property measurement standards. Available van Oosterom, P., 2018. Best practices 3D cadastres. Copenhagen, from: https://ipmsc.org/standards/ [Accessed 8 Oct 2020]. Denmark: International Federation of Surveyors – FIG. Survey Review 2020 11 Tekavec et al. Indoor space as the basis for modelling of buildings in a 3D Cadastre van Oosterom, P., et al., 2019. Joint ISO/TC211 and OGC revision Yan, J., Diakité, A.A., and Zlatanova, S., 2019. A generic space definition of LADM: valuation information, spatial planning infor-framework to support seamless indoor/outdoor navigation systems. mation, SDG land indicators, refined survey model, links to Transactions in GIS, 23 (6), 1273–1295. doi:10.1111/tgis.12574 BIM, support of LA processes, technical encodings, and Ying, S., et al., 2017. 3D space shift from CityGML LoD3-based much more on their way! FIG Working Week 2019: multiple building elements to a 3D volumetric object. ISPRS Geospatial Information for a Smarter Life and Environmental International journal of geo-information, 6 (1), 17. doi:10.3390/ Resilience. ijgi6010017 Wang, C., 2015. 3D Visualization of cadastre: assessing the suitability Zlatanova, S., et al., 2016. Ladm and indoorgml for support of indoor of visual variables and enhancement techniques in the 3D model space identification. ISPRS annals of photogrammetry, remote sen- of condominium property units. Thesis (PhD). Québec: sing and spatial information sciences, IV-2/W1(October), 257–263. Université Laval. doi:10.5194/isprs-annals-IV-2-W1-257-2016 12 Survey Review 2020 PRILOGA C Cadastral data as a source for 3D indoor modelling Tekavec, J., Lisec, A. (2020a) Land Use Policy, 98(2020): 1-14 DOI: 10.1016/j.landusepol.2019.104322 Faktor vpliva za leto 2019: 3,682 (Q1) Land Use Policy 98 (2020) 104322 Contents lists available at ScienceDirect Land Use Policy journal homepage: www.elsevier.com/locate/landusepol Cadastral data as a source for 3D indoor modelling Jernej Tekavec*, Anka Lisec University of Ljubljana, Faculty of Civil and Geodetic Engineering, Department of Geodesy, Jamova cesta 2, SI-1000 Ljubljana, Slovenia A R T I C L E I N F O A B S T R A C T Keywords: Despite the rapid development of indoor spatial data acquisition technology, there are currently no solutions that IndoorGML enable large-scale indoor spatial data acquisition due to several limiting factors that characterize the indoor 3D cadastre space. This fact, together with the rapidly growing need for indoor models, is the main motivation for our Real property research. The focus is on the study of the appropriateness of existing cadastral data for 3D indoor modelling. Floor plan Within the study, a framework for 3D indoor modelling has been developed, comprising a chain of processes, 3D modelling starting from initial cadastral data and ending with the OGC IndoorGML compliant document. The process chain is divided into three parts, which are described individually and supported by UML activity diagrams. The Slovenian Building Cadastre data represents the basis for the framework design and data assessment. The IndoorGML standard is used for final outputs, as it provides a standardized data model for the representation and exchange of indoor spatial information designed for indoor navigation and location-based services. The data storage options using a spatially enabled database are presented for storing 2D and 3D geometries. The stored data enables fully automatic IndoorGML document generation on request, while also taking advantage of all spatial database functionalities. The proposed approach is software independent and can be implemented with various spatially enabled software packages. In addition to 3D indoor data modelling, the framework represents a comprehensive method for assessing the usability of input data for the purpose of 3D indoor modelling. The assessment is done for the case of the Slovenian Building Cadastre. The assessment of the cadastral data suitability for 3D indoor modelling can be used for decisions regarding future steps towards a multi-purpose 3D real property cadastre. The presented concept can be applied in many countries worldwide that have a similar condominium registration system. 1. Introduction particular Global Navigation Satellite Systems (GNSS), cannot be used in the indoor environment. As a result, there are a great number of The importance of 3D indoor models is growing due to a variety of highly developed and massively used location-based services (LBS) for applications, related in particular to location-based services and navi- the outdoor environment, such as Google Maps, Uber, Foursquare, gation (Afyouni et al., 2012; Yang and Worboys, 2015; Lin and Lin, Pokemon Go etc., and almost none for indoor space. Looking at the 2018). We are often faced with locating points of interest (POI) and market of outdoor LBS, one can easily see the great potential of indoor finding optimal paths to them, which can be effectively solved by using LBS, especially because nowadays we spend a major part of our time navigation principles. The navigation process requires two key com- indoors, whether working, resting, exercising, shopping, etc. The in- ponents. The first one is positioning and the second one is spatial data, door navigation applications are of particular interest with regard to which provides the spatial context to the position information. Com- public buildings (e.g. hospitals, schools, universities, bus and railway bining these two components, the navigation device can calculate an stations, etc.) and shopping centres to facilitate POI searching for their optimal route to the desired POI. visitors. Apart from that, several public services are potential users of Recent technological developments in the field of geospatial data indoor spatial data (e.g. police, emergency medical aid, firefighters, acquisition and outdoor positioning have made location-based services etc.). These services would benefit from indoor spatial data of every highly affordable and widely available on smartphones and other type of building, not just the public ones. Some studies have evaluated electronic devices (Huang et al., 2018). However, this is only true for the indoor 3D spatial data and indoor navigation support for first re-the outdoor environment, as technologies for geospatial data acquisi- sponders in emergency situations (Lee and Zlatanova, 2008; Tang and tion, including remote sensing, and technologies for positioning, in Ren, 2012; Chen et al., 2014; Tashakkori et al., 2015). The potential has ⁎ Corresponding author. E-mail addresses: jernej.tekavec@fgg.uni-lj.si (J. Tekavec), anka.lisec@fgg.uni-lj.si (A. Lisec). https://doi.org/10.1016/j.landusepol.2019.104322 Received 2 April 2019; Received in revised form 14 October 2019; Accepted 21 October 2019 Available online 22 January 2020 0264-8377/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 expanded even more in recent years with the increasing use of smart- suitability from the perspective of 3D indoor modelling. The latter can phones, which combine portability, computing power, data storage, provide valuable information for decision making regarding the future connectivity and sensors. changes of data models and processes in land administration systems. Several technologies are being developed to enable efficient indoor The article represents an extension of the research, published at 6th FIG spatial data acquisition using various combinations of passive image Workshop on 3D cadastres (Tekavec and Lisec, 2018). The whole fra-sensors and active ranging sensors (Gunduz et al., 2016; Kang and Lee, mework was thoroughly revised, and the proposed process has been 2016; Liang et al., 2016; Jiao et al., 2017; Lee et al., 2017; Lenac et al., modelled using UML activity diagrams. The research is supplemented 2017). Indoor positioning is also a rapidly evolving field with some with a comprehensive study on various options for data storage in a technologies already available and a lot of ongoing research activity spatial database. Additionally, 3D data visualization options are pre- (Vanclooster et al., 2016; Correa et al., 2017; Brena et al., 2017). The sented and discussed. major common problem of both spatial data acquisition and positioning technologies related to the indoor environment that sets them apart 2. Literature review from technologies for outdoors is that they are currently spatially very limited, i.e. they cannot be performed on a large scale. In practice, this In recent years, the topic of indoor LBS has been experiencing in- means that each building must be individually equipped with an indoor tensified attention in research as the services and available technologies positioning system, and a detailed as-built 3D model of the building lag behind compared to the ones developed for the outdoor environ- interior structure should be available. In addition, the model has to ment (Jensen et al., 2010; Gunduz et al., 2016). Important factors that include the topology information, i.e. an indoor space connectivity facilitate research and development and ensure their consistency are graph. international standards. There was no specific standard in the field of The complexity and consequently high costs of indoor spatial data indoor LBS until the Open Geospatial Consortium (OGC) standard In- acquisition noted above justify research on the usability of already doorGML was introduced in 2014, aiming to harmonize and foster re- acquired and stored data for existing buildings. The most detailed and search and development. IndoorGML standard provides an open data comprehensive documentation about buildings is a construction plan. model and XML schema for indoor spatial information (OGC, 2014). It For older buildings, the drawings and documentation were provided covers geometric, topological and semantic aspects of indoor space. The and archived in analogue form, while for newer ones digital vector origins of the standard date back to the year 2009, when the multi- models are provided, and for some recently built ones also conforming layered space-event model for navigation in indoor spaces was pub- to the standards in the fields of Building Information Modelling – BIM lished by Becker, Nagel and Kolbe (2009). They defined key principles (ISO, 2018). Construction models that are provided within BIM in ac-that are used in the IndoorGML standard. The standard follows the cordance with the Industry Foundation Classes (IFC) provide rich 3D cellular space concept, according to which the indoor space is modelled geometry and semantic information, which can be used, analysed, en- as a collection of non-overlapping cells. This sets it apart from the other riched and updated by a large range of domains included in a building’s standards in the field of 3D modelling (e.g. CityGML, IFC) as they do not lifecycle. This also includes the field of indoor LBS, which can benefit model the indoor space itself, but the building features (e.g. walls, from the growing utilization of BIM in practice. Much recent research is windows), which, on the other hand, can also define indoor space. The therefore focused on integration of BIM and geospatial domains to fully most similar class, named IfcSpace, is used in IFC to represent empty exploit BIM potential in the geospatial domain and vice versa (Li and spaces inside the building and can be used for linking the two standards He, 2008; Chen et al., 2014; Hong et al., 2015; Deng et al., 2016; (Teo and Yu, 2017). The overlap with other standards in the geometric Diakité and Zlatanova, 2016a, b; Xu et al., 2017; Teo and Yu, 2017; Ellul part is solved with the possibility to add external references. However, et al., 2018). Currently, BIM has not yet been widely introduced op-3D geometry can also be included in an IndoorGML document. erationally into the processes of building design, construction, main- Topology is the key component of the IndoorGML standard, as it is tenance, and facility management, but in the future, it has big potential vital for navigation applications (Lee, 2004). It is realized in the form of to become a valuable data source for indoor LBS. The biggest issues a Node-Relation Graph (NRG). The theoretical basis for derivation of surrounding the use of building construction documentation and data NRG from the indoor space geometry is the Poincaré duality, where a k- for indoor data extraction lie in their complexity, and in the fact that dimensional object in N-dimensional primal space is transformed to an generally, they do not provide as-built information about new buildings (N-k) dimensional object in dual space (Munkres, 1984). The topolo- (Atazadeh et al., 2017). Another possible source of data about build-gical relationships in IndoorGML are explicitly described using the ings, albeit lacking in detail, is land administration data. Although their XLinks concept of XML provided by GML. The referencing is realized content, structure, degree of detail and entry regulations depend on using href attributes ( xlink:href is used in the paper). Another important each country’s legislation, this data is generally centralized and easier concept of the IndoorGML standard is a multi-layered representation to access compared to construction documentation. (Becker et al., 2009). It allows the same indoor space to be modelled in This article focuses on 3D indoor modelling of buildings using ca-several layers according to the cellular space concept and therefore dastral data as an alternative data source to other indoor spatial data allows for separate modelling of WIFI, RFID and other spaces related to acquisition approaches. In this way, the article also aims to contribute indoor navigation (Fig. 1). The links between spaces are established via to the idea of a multipurpose 3D cadastre (Tekavec et al., 2018), of-interlayer relations with different possible relations, such as within, fering data modelling approach for indoor location-based services. We contain, overlap. decided to use cadastral data from Slovenian Building Cadastre, as it With the introduction of the IndoorGML standard, the research has several advantages over other data sources in Slovenia, in parti- community and developers got a data model to develop interoperable cular, the centralized storage, availability and relatively uniform data solutions for indoor LBS. Since then, several studies relating to this content are the most important ones. Our aim has been to develop a standard have been done on data acquisition, 3D modelling, visuali- framework for 3D indoor modelling which could be applied to Slove- zation and applications. Seo (2017) developed a software for creating nian cadastral data. It is designed to generate 3D indoor models, IndoorGML compliant models, which is is similar to our proposed fra- compliant with IndoorGML standard, that provides a data model for mework in terms of data input and output. However several differences indoor spatial information. The framework can be used for two pur- between them exist, which are discussed in section 5. Ryoo et al. (2015) poses, either to generate 3D indoor models or to assess the input data compared the OGC standards IndoorGML and CityGML (OGC, 2012), 2 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 1. IndoorGML multi-layered representation (OGC, 2014). which both cover 3D modelling of indoor space in the GIS domain. The UML activity diagrams are used to summarize and clarify the Although they are similar in the 3D modelling aspect, the authors framework (Figs. 3, 4, 7 and 14). The diagrams follow the division of showed their differences, especially in their scope and applications. the framework into three parts. Each diagram presents the activities Typically for rapidly developing fields, there are a large number of and their links in the corresponding part of the framework. Aiming to remaining unresolved issues regarding indoor mapping and modelling. design an applicable framework, we have tested each process using the These issues are systematically analyzed and categorized into existing input data from the Slovenian Building Cadastre in the form of a floor and future ones in Zlatanova et al. (2013). Recently Kang and Li (2017) plan of a residential house. The data is in raster format, containing a emphasized the potential of the IndoorGML standard and encouraged floor plan for each storey (see also Drobež et al., 2017). Similar floor the research community to include it in their research by proposing new plans are usually included in the documentation for condominium re- features, developing new extensions and performing case studies. Re- gistration in several other countries. GIS and ETL tools were used for search has also been conducted in relation to the data sources and the implementation of the framework. Besides the output in IndoorGML modelling processes for obtaining IndoorGML compliant models (Khan format, we developed options for 2D and 3D database data storage. The et al., 2014; Mirvahabi and Abbaspour, 2015; Kim and Lee, 2015; storage can serve as a final output or as an intermediate step from Diakité and Zlatanova, 2016a,b; Teo and Yu, 2017). Indoor spatial in-which the rest of the framework to obtain IndoorGML file can be done formation is related not only to physical structures but can also be automatically on request at any given time. For the Slovenian study combined with legal information (Zlatanova et al., 2016; Alattas et al., case, a usability analysis for 3D indoor modelling was further con- 2017). These studies deal with linking IndoorGML and LADM (ISO, ducted, aiming to identify advantages and disadvantages of the data 2012) mostly on a conceptual level. Alattas et al. (2018) further pro-used throughout the process. pose a database implementation of the conceptual link between LADM The framework is developed and illustrated based on the Slovenian and IndoorGML. The main aim of linking is to analyse how legal in- Building Cadastre data. The Slovenian Building Cadastre was in- formation from LADM can improve the semantic properties of In- troduced in 2000 as a database for condominium registration in the doorGML models and thus improve the process of indoor navigation. Land Registry (see also Drobež et al., 2017). In the following years, up The cadastral extension is also mentioned as a candidate for the se-to 2006, photogrammetric acquisition of 2D outlines (outdoor) for all mantic extension of IndoorGML in Kang and Li (2017). The link be-buildings was conducted for the whole country. In addition to building tween land administration and IndoorGML has therefore already been outlines, additional attributes were collected, including the ground established and studied, but until now only at a conceptual level. height and maximum height. In 2006, the Building Cadastre was legally and operationally introduced and detailed registration of buildings and their parts (legal subdivision), together with floor plans, became man- 3. Research methodology and materials datory. The Building Cadastre data is open and publicly available via the official website. The scanned building registration documentation The design of a conceptual framework for 3D indoor modelling with floor plans is not publicly available. However, it is available to the based on cadastral data represents the core of the presented research. authorized land surveyors. The framework is composed of a chain of manually or automatically In this study, we have used floor plans from the building in the performed processes, starting from initial cadastral data in the form of Building Cadastre, which have been available in raster digital form floor plans in raster format. The attention is given to minimize the need (Fig. 2) for more than a decade. The vector form of floor plans has been for manually performed activities, as they significantly increase the required for new cadastral entries since 2018. Mandatory content of amount of resources needed for execution of the implemented frame- floor plans are outlines of building parts for each floor. In many cases, work. The OGC IndoorGML standard is used for final outputs. The the documentation contains more detailed floor plans with room out- framework consists of three main parts that are described in detail in lines aiming to clarify the building division into building parts. Section 4. 3 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 2. Sample document of Slovenian Building Cadastre containing the building floor plan. 4. Framework design In this section, the entire process chain of the framework is de- scribed in detail. The framework is based on the research, published in Tekavec and Lisec (2018). The division into three parts follows the actual workflow of the framework implementation (Fig. 3). 4.1. Data preparation and digitization phase In the first phase, all the data should first be acquired, prepared and then digitized. Additionally, the attributes, necessary for the execution of processes in the next phases, have to be added. Fig. 4 presents the processes that are executed in this phase. For the implementation, we took raster floor plans of a residential house (with outlines of all rooms) modelled to fit into Slovenian Building Cadastre. Generally, two types of floor plans exist, depending on whether or not the wall thickness is considered (Fig. 5). The IndoorGML standard treats both concepts (thin-wall and thick-wall) as valid. Based on the available digital raster image (floor plan), the geo- metry and topology (room connectivity) of indoor spaces can be ob- tained through digitization (Fig. 6). First, we scaled the raster images to represent the true extent of the building. For each floor plan, we created three spatial layers to be able to later align the digitized data to the IndoorGML data structure: a polygon layer for room geometry, a line layer for connections (graph edges), and a point layer for graph nodes. As the IndoorGML standard does not allow cell overlapping, each room outline digitized using the polygon feature, representing the basis for Fig. 3. Generalized UML activity diagram of the proposed framework. 3D cells, has to be checked for overlap with other polygons. This can be 4 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 4. UML activity diagram for the digitization phase of the proposed framework. done automatically in QGIS software by the Topology Checker tool. To Once the room polygons are created, points representing graph enable the construction of 3D cell spaces with the extrusion, the floor nodes can be created automatically as centroids, while lines re- and ceiling heights have to be added for each floor. These attributes had presenting graph edges have to be added manually if the floor plan does not been available in Slovenian Building Cadastre until recently (2018) not include door openings. The possible connections between spaces so the heights have to be measured and added manually. The heights can be automatically narrowed down to connections between neigh- can be relative, absolute or a combination of both. For absolute vertical bour spaces, but the actual connections still have to be added manually. positioning of the model, we need one height that has both relative and Together with floor and ceiling heights, this represents significant absolute height. Navratil and Unger (2013) provided a comprehensive manual input as it requires physical inspection of the building. If the overview of heights in land administration. It is worth mentioning that floor plan contains the connection openings, they can be automatically polygon extrusion could produce overlapping of 3D cells in cases where identified using object recognition techniques on a raster image. Due to the floor and ceiling are not straight and horizontal. the duality concept, the points are given the same identifier as room 5 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 4.2. Transformation phase In the second phase, the data from the first phase goes through a set of transformation processes (Fig. 7). All layers of the same type, digitized separately for each floor, are first merged into one layer, followed by three different sets of transformations, one set for each feature type (polygon, line, point). Additionally, the options for 2D and 3D data storage are presented within this phase. At the end of this subsection, we present the option for 3D visualization of the data, which can be used to perform quality checks and for a clear representation of the modelled data. The polygons representing outlines of the rooms are transformed into 3D cells with a process of extrusion using the height attributes. Each digitized 2D polygon, representing a room outline, is “lifted” to the floor height and then extruded to ceiling height, which forms a closed 3D cell. After extrusion, the orientation of faces has to be checked and then the cell can be assembled into solid geometry. The final phase of the transformation of polygons is creating IndoorGML specific attributes (Fig. 8). A unique id is assigned to each cell using a Fig. 5. Raster floor plan from Slovenian Building Cadastre following the thin-global counter. When generating the models separately (at a different wall concept - the room annotations consist of building part number and room time or as separate processes) the uniqueness of cell ids should be en-usage code in brackets. sured by saving the last used id, using unique ids within the model combined with the building’s unique id, etc. The parent property and parent id attributes are static and contain information about the posi- tion and role in the hierarchy of the IndoorGML document. According to the concept of Poincaré duality, the cell and the corresponding node are connected. This connection is materialized with a duality xlink:href attribute. If we use the same numbering of cells and nodes (to obtain unique ids), only the node prefix has to be added to the numbering of cells to obtain a duality xlink:href attribute (“N1” for the node that corresponds to the “C1” cell). The height values needed to position the nodes inside the linked cells in 3D space can be derived from the corresponding polygons. For implementation, we chose the mean value between the floor and ceiling height. The same types of attributes are added to nodes as to cells, with an additional connects xlink:href attribute that contains a list of all edges that are connected to a given node. This list was not included in the digitization phase, as the node can have any number of connected edges, so the connection information cannot be stored in the attribute table of nodes. Having the information about the start and end node for each edge acquired in the digitization phase, the list of all connected lines to each node can be created by joining the edge and node table. Fig. 6. Digitized polygons, line strings and points with identifiers. We implemented the list of connected edges by leveraging the list at- tribute option in FME Desktop software. polygons. Each line should be snapped to the start and end point with The edges are more challenging to put in the 3D environment than the start and end point identifiers added as attributes. The lines re- the nodes. The start and end height can be derived from node heights. presenting connections between rooms in different floors (stairs and As long as the connection between nodes is placed on a single level elevators) have to be snapped to points on different layers with both these heights are sufficient. When the two connected nodes lie on se- point identifiers added as attributes. This enables the integration of parate levels, the edge should change the height along its way. separate connectivity graphs into one graph for a multi-storey building. Our solution simplified this problem using the start node height While the positional alignment of floors does not affect the topology, it until the last line break and then changing the height to the end node is still important to properly align the floor plans one above the other, height. This reduces the need for additional data input but can yield to enable proper 3D visualization of geometry. The alignment can be non-representative edges. Unique id, parent property and parent id automated if floor plans contain a common reference feature (outer attributes were also added to edges. We left out the duality xlink:href perimeter, reference point, etc.). There are two options from the co- attributes used to link the edges and faces, as we do not provide ordinate system perspective. We can use a local coordinate system for modelling of cell connection surfaces (doors, windows) within our each building, or we can choose to digitize all layers in a common framework. If they exist in input data, they can be additionally included national or global coordinate system, as we did. If the local coordinate in the framework as a fourth digitization layer, transformed and written system is chosen, we need to establish a link to a national or global into the IndoorGML document as CellSpaceBoundary. The connects coordinate system. This can be a point with national or global co- xlink:href attribute is sourced from the digitized line layer. The weight ordinates of the local coordinate system origin. Here it has to be attribute determines the movement difficulty along the edge and re- mentioned that the IndoorGML standard supports the conversion of presents key information in optimal path planning. The edge length and coordinate reference systems via anchor node element. height difference can be used to automatically estimate the weight values, as they are available within the framework. However, more accurate values can be set if additional information or manual input is provided. 6 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 7. UML activity diagram for the transformation phase of the proposed framework. 7 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 8. IndoorGML specific attributes for each feature type (Tekavec and Lisec, 2018). Fig. 10. 2D database storage with building table. and analyses consider only two dimensions, with a few exceptions of some basic functions supporting 3D properties, e.g. Z-coordinate value retrieval. The 2D data storage can be performed right after the start of the transformation phase, as can be seen from the UML activity diagram for the data transformation phase (Fig. 7). Another option is storing the combined layers from the transformation phase. The stored geometries and attributes can be accessed and visualized by various GIS software, web services, etc. The basic concept of 2D database storage contains the Fig. 9. The basic concept of 2D database storage. 4.2.1. Database data storage The storage of the data in a database has many advantages com- pared to file-based data storage, especially from a data management and data dissemination perspective. Since the result of the proposed framework is in a file format, we propose two options of intermediate database data storage within the transformation phase. The storage is designed to enable fully automated creation of IndoorGML files. It can be seen as a breakpoint in the framework, from where the processes can be used only on request and data sourced from the database. The dif- ference between options is the dimension of the stored geometries. 4.2.1.1. 2D database storage. Most of the databases support storing and management geometry following ISO SQL/MM-Part 3 (ISO, 2016) or OGC Simple Feature Access (OGC, 2010). They mostly support storage of geometry in the 3D space, but the functions for data management Fig. 11. The first level of 3D database storage. 8 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 geometries was chosen for our implementation. It can be performed after “lifting” of all features and extrusion of polygons into 3D solids. For nodes and edges, the storage is the same as in the first level, while cells are now stored as polyhedral surfaces, representing the outer shell of the solids (Fig. 12). To overcome the lack of current 3D support in GIS software, we propose a hybrid database storage, where “lifted” polygon geometry and cell relative height are still stored for cells. 4.2.2. 3D data visualization Cells, nodes and edges modelled in 3D space open the possibility for their 3D visualization. The purpose of data visualization is often nar- rowed down to just being a tool for user-friendly and clear data pre- sentation, but it is also an excellent tool to check data for incon- sistencies and errors. Within our framework implementation, we used FME Inspector to inspect the results of each process of the transfor- mation phase, as well as the end result – IndoorGML file. FME Inspector also allows visualization of the digitized layers and data stored in the database. To make the visualization of the data within the framework open and accessible, we propose the WebGL-based data visualization plugin for QGIS named Qgis2threejs. It takes 2D layers in QGIS and puts them in 3D space using height attributes (Fig. 13). The prepared visualization can be easily published to the web and opened in a web browser. We can further take advantage of QGIS advanced symbology Fig. 12. Second level of 3D database storage. capabilities to make data visualization as clear and descriptive as pos- sible. The visualization has built-in commands for controlling 3D view, cell, node and edge table (Fig. 9). Since the cells and nodes are related switching the layers on and off, adjusting transparency for each layer, according to the Poincaré duality, the storage can be further simplified and attribute-based labelling and access to feature attributes by their to store only the cell and edge table. In that case, node geometry and selection. Several studies deal with the challenges of 3D spatial data attributes would have to be derived from the cell table. visualizations, such as occlusions, etc. (Shojaei, 2014; Wang, 2015; The brackets for x_translation and y_translation attributes indicate Zhou et al., 2015). that these attributes are needed only when local coordinates are used. From version 3.0 on, QGIS has a built-in 3D viewer that has func- Instead, another point geometry column can be added. However, this tionalities to visualize the spatial data in 3D. Currently, the 3D viewer can cause problems with some software that supports retrieval of only does not provide sufficient functionalities to visualize volumetric ob- one geometry column per table. The “…” sign means that additional jects representing indoor spaces, together with the connectivity graph. custom attributes can be added for each entity. By adding the building Also, the visualization is limited to QGIS software and cannot be dis- table (Fig. 10) related to official registers the data can be linked to seminated online like the output of the Qgis2threejs plugin. address numbers and other data. The translation attributes if using local coordinates can also be stored in the building table. The rotation at- tribute can also be included but should be additionally accompanied by 4.3. Structuration phase the rotation origin point. The 2D data storage is designed to allow fully automatic derivation of IndoorGML files, but not without additional In the final step, the transformed data is structured according to the processing. IndoorGML structure and written into the IndoorGML document (Fig. 14). 4.2.1.2. 3D database storage. As mentioned in the previous section, The 3D cell solid geometry is assigned to a CellSpace element, node most databases support storage of geometries in 3D space, including geometry to a State element and edge geometry to a Transition element. point, line string and polygon geometries. Some of the most advanced All geometries are encoded in GML (ISO, 2007). The parent id and databases (Oracle Spatial, PostgreSQL with PostGIS) nowadays support parent property attributes enable the creation of cellSpaceMember, sta- storage of volumetric 3D geometries. Therefore, two levels of 3D teMember and transitionMember elements and their proper placement in database storage can be performed within the transformation phase. the element hierarchy. The duality elements for CellSpace and State The first level of 3D database storage can be performed when points, elements that establish links between them are created using duality line strings and polygons are “lifted” onto the appropriate heights xlink:href attributes. The connects elements for State and Transition (Fig. 11). For cells, only one height attribute is now required for later elements are created with connects xlink:href attributes. These attri-extrusion. Also, the “lifting” of points and line strings with the polygon butes contain a list of connected feature identifiers, solving the car- heights is not needed every time when deriving IndoorGML file out of dinality of the node – edge relation. The Transition element also get a the stored data. These geometries are also supported by most GIS weight element with a weight attribute assigned. All other IndoorGML software, web services, etc. elements and their attributes are created to comply with IndoorGML The second level of 3D database storage requires the support for standard (Tekavec and Lisec, 2018). volumetric 3D geometries. The open-source PostgreSQL database with PostGIS SFCGAL extension that offers state-of-the-art support for 3D 9 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 5. Discussion digitized. For ensuing processes, we used the ETL software FME from SAFE software that supports 3D geometries and reading and writing Our framework was implemented using QGIS open source GIS IndoorGML files. For the input data, the raster floor plans from the software in the digitization phase, in which raster floor plans were Slovenian Building Cadastre were selected. 10 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Fig. 13. 3D visualization of the transformed data (Tekavec and Lisec, 2018). Fig. 14. UML activity diagram for the structuration phase of the proposed framework. 5.1. Data assessment required changes in the data model if they decide to support 3D indoor modelling. The presented framework can be used as a tool to identify the ad- For the selected case, the Slovenian Building Cadastre, many ad- vantages and disadvantages of the (cadastral) data suitability for 3D vantages were identified in a very early stage of our research and were indoor modelling. This is important the organizations, such as cadastral the reason why we considered using it for 3D indoor modelling. The authorities that are responsible for the cadastral data, to devise the data is centrally stored and maintained which makes it easily 11 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 already stated, since 2018, the floor plans are submitted in a vector format but are strictly limited to outlines of building parts, without the possibility to include room geometries. This significantly limits the usability of the floor plans for 3D indoor modelling. The paper focuses on using the building floor plans, which are not available for most of the buildings in Slovenia, as the full building registration is mandatory only since 2006. This is a significant limitation due to the fact that the proposed framework is not allowing a nationwide 3D indoor modelling. 5.2. Findings The detailed description of processes needed to develop a 3D indoor model of a building based on cadastral data resulting in an IndoorGML document opens up several topics for discussion. In the final IndoorGML document, we decided to use the 3D geo- metry of the cells. The only method to obtain 3D geometries that in- cludes a reasonable amount of additional manual input is the extrusion of 2D polygons using floor and ceiling heights. While it is the most feasible approach, it also doesn’t bring many advantages over 2D geo- metries. 3D geometries enable more realistic visualizations but have limitations, such as occlusions, that should be considered, like occlu- sions. The additional heights can be useful to determine if the height of the space enables navigation through it (some spaces can have very low ceiling). If the navigation graph is placed in 3D space, we can determine the height difference of the calculated route and thus get additional information about how demanding it is, as locomotion in a vertical direction is more demanding. For the Slovenian case study, it has been shown that in the first part of the framework several missing data have to be added manually. Additional data are generally not cheap, and its acquisition is time- consuming, especially for the indoor environment. As already men- tioned, progress has been made by introducing the floor heights and vector floor plans in the Slovenian Building Cadastre in 2018. The inclusion of the information about which rooms are connected into the cadastral documentation would be a greater challenge, as it is far beyond the scope of the current land administration system. On the other hand, not much additional information would need to be col- lected and stored in the cadastral database. A basic connectivity graph without a detailed edge geometry could already be generated from room connections in tabular form. The nodes can be created auto- matically as centroids and then also edges, using the geometry of nodes Fig. 15. Automatically derived graph from room polygons and room connec-and connectivity information from the table. Each edge is therefore tion information (top) and graph containing digitized nodes and edges (bottom) (Tekavec and Lisec, 2018). constructed as a straight line using start and end node geometries. The algorithms for automatic generation of centroids can fail to place the accessible, provided one has the access rights (attribute data is public, centroid inside the polygon feature, but that can be automatically documentation is available to authorized land surveyors). In this aspect, checked and manually corrected. Generated in such a way, a graph the cadastral data is more useful than construction data, which is often would have weak geometric properties, but its topology would be valid. hard to get access to and is not centrally stored. Another important Teo and Yu (2017) construct the connectivity graph the same way from advantage of cadastral data, compared to construction plans or data IFC data and propose manual editing of the edges for complex and open provided within BIM, is that the cadastre provides as-built data, while spaces. Fig. 15 shows a comparison of an automatically generated graph construction data often provides only as-planned information that can from tabular room connectivity data and a graph generated with the significantly differ from as-built data. As the Slovenian Building digitization of nodes and edges. However, it should be emphasized that Cadastre provides a technical basis for condominium registration, the with increased building complexity, the difference between the two 3D indoor models derived from the cadastral floor plans can be directly graphs would also increase. linked to the established rights, restrictions and responsibilities in the To reduce the need for additional manual input, we considered only building. The 3D model can be further linked to the Register of house IndoorGML core module. However, the model can be semantically en- numbers, allowing integration of indoor and outdoor navigation. In riched with IndoorGML Indoor Navigation module that provides in- 2018, the Slovenian Building Cadastre introduced submission of floor formation, useful for indoor navigation applications, by classifying the plans in vector format together with floor heights which significantly core module elements into navigable and non-navigable ones. If the reduces the manual input in the digitization phase of the proposed local coordinates are used for modelling, the anchor space element can framework. be used to establish a link between local and global coordinate re- The disadvantages mostly originate from the fact that the data ference systems and thus enable the integration of indoor-outdoor na- model for the Slovenian Building Cadastre was not designed to support vigation. 3D modelling and indoor graph derivation. Until 2018, the floor plans The IndoorGML standard provides two different approaches for cell were submitted in raster format, requiring initial digitization. As modelling, a “thin wall” model and a “thick wall” model. We have chosen the “thin wall” model, as it is closely aligned with floor plans in 12 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 Table 1 implemented with various spatially enabled software packages. Comparison between JInedit solution and our proposed framework. For the Slovenian case study, we identified key missing data in the current documentation of the Building Cadastre that is needed for 3D Jlnedit The proposed approach indoor modelling. To produce proper results, the need for vector floor Implementation software standalone software various possibilities plans has been identified, and additionally, floor height information Floor heights one for each floor one for each room and room-to-room connectivity are currently missing in the cadastral Additional data inclusion not possible possible database. Although the paper does not focus on semantic enrichment of Export to other formats not possible possible Database integration not possible possible a 3D model, it should be stressed that additional data can be provided Required user skills low high by linking the source data to land administration databases, which has great potential for semantic enrichment of IndoorGML models with information about rights restrictions and responsibilities, the value of the Slovenian Building Cadastre, which does not account for the the real estate, house number etc. thickness of walls. If construction plans are used instead, it would be Although the framework is developed based on raster floor plans better to use the “thick wall” model, as there, the walls are drawn with from Slovenian Building cadastre, it can be used for any similar data their actual thickness. One of the drawbacks of the “thin wall” model is that represents 2D floor plans with some adjustments, especially re- that it does not allow creating the cells with the correct geometry while garding the manual input. A detailed description of all processes and maintaining the correct outer shape of the building. These two concepts the process diagram, together with the following remarks and con- also raise open questions regarding modelling rules for 3D geometries siderations, can serve as a starting point to assess the data in national in a 3D real property cadastre. If the 3D geometry provides the only land administration systems worldwide, whether and to what degree reference to the physical structures of the building, which define the the data can be used for 3D indoor modelling and what should be real extent of RRR in space, the correct geometry is not as important as changed. Future research will be focused on the analysis of other data in the case when 3D geometry defines the exact extent of RRR. The which is available by linking the source data to connected databases second option allows the reconstruction of the property unit geometry and their usability in the context of indoor navigation applications. without the building structure (not yet built or demolished). An open (legal) question is also how the border between two property units is Acknowledgements defined, the middle of the wall or on each side of the wall (Atazadeh et al., 2016). The authors acknowledge the financial support from the Slovenian The proposed framework has several features in common with Research Agency (research core funding No. P2-0406 Earth observation JInedit software (Seo, 2017), which is an open source software that and geoinformatics). enables the creation of IndoorGML documents based on raster floor plans. While it uses the same input and provides practically the same References output, there are several differences between our approach and Jlnedit software (Table 1): Alattas, A., Zlatanova, S., van Oosterom, P., Chatzinikolaou, E., Lemmen, C., Li, K.-J., 2017. Supporting indoor navigation using access rights to spaces based on combined - The JInedit is standalone software, while the process presented in use of indoorGML and LADM models. ISPRS Int. J. Geo-Inf. 6 (12), 32. https://doi. this paper is a set of software independent operations that can be org/10.3390/ijgi6120384. Alattas, A., van Oosterom, P., Zlatanova, S., Diakité, A.A., Yan, J., 2018. Developing a implemented using various spatially enabled software, which re- database for the LADM-IndoorGML model. In: 6th International FIG Workshop on 3D quires more skilled users. Cadastres. 2-4 October 2018, Delft, the Netherlands. pp. 261–278. - The proposed process enables more flexible height provision (each Atazadeh, B., Kalantari, M., Rajabifard, A., 2016. Comparing Three types of BIM-based models for Mana 3D ownership interests in multi-level buildings. In: 5th International room can have its own floor and ceiling height) and hence better FIG Workshop on 3D Cadastres. 18-20 October 2016, Athens, Greece. pp. 183–198. representation of a real-world situation. Atazadeh, B., Kalantari, M., Rajabifard, A., Ho, S., Ngo, T., 2017. Building information - Only limited data can be acquired in Jlnedit, while the proposed modelling for high-rise land administration. Trans. GIS 21 (1), 91–113. https://doi. process allows additional data, such as rights, restrictions and re- org/10.1111/tgis.12199. Afyouni, I., Ray, C., Christophe, C., 2012. Spatial models for context-aware indoor na-sponsibilities, currently not supported by the IndoorGML standard vigation systems: a survey. J. Spatial Inf. Sci. 1 (4), 85–123. https://doi.org/10.5311/ to be captured and stored in the spatial database. JO-SIS.2012.4.73. - The proposed process enables the export of 3D geometry in various Becker, T., Nagel, C., Kolbe, T.H., 2009. A multilayered space-event model for navigation in indoor spaces. In: Lee, J., Zlatanova, S. (Eds.), 3D Geo-Information Sciences. 3D formats. Springer-Verlag, Berlin, pp. 61–77. https://doi.org/10.1007/978-3-540-87395-2. - The proposed process allows direct storage of geometry and attri- Brena, R.F., García-Vázquez, J.P., Galván-Tejada, C.E., Muñoz-Rodriguez, D., Vargas-butes from the transformation phase into the database where Rosales, C., Fangmeyer, J., 2017. Evolution of indoor positioning technologies: a survey. J. Sens. 2017. https://doi.org/10.1155/2017/2630413. routing capabilities can later be used. Chen, L.C., Wu, C.H., Shen, T.S., Chou, C.C., 2014. The application of geometric network models and building information models in geospatial environments for fire-fighting simulations. Comput. Environ. Urban Syst 45, 1–12. https://doi.org/10.1016/j. 6. Conclusions compenvurbsys.2014.01.003. Correa, A., Barcelo, M., Morell, A., Vicario, J.L., 2017. A review of pedestrian indoor The study aimed to create a framework for 3D indoor modelling of positioning systems for mass market applications. Sensors 17 (8). https://doi.org/10. 3390/s17081927. buildings from input raster floor plans to an IndoorGML compliant Deng, Y., Cheng, J.C.P., Anumba, C., 2016. Mapping between BIM and 3D GIS in different document. A detailed workflow for 3D indoor modelling has been levels of detail using schema mediation and instance comparison. Autom. Constr. 67, provided, comprising a chain of processes starting from initial cadastral 1–21 https://doi,org/10.1016/j.autcon.2016.03.006. Diakité, A.A., Zlatanova, S., 2016a. Extraction of the 3D Free space from building model data in the form of a raster floor plan and ending with the OGC for indoor navigation. ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci. IndoorGML compliant document. We have tried to design the whole 241–248. https://doi.org/10.5194/isprs-annals-IV-2-W1-241-2016. IV-2/W1. workflow in a generic way as much as possible, while also considering Diakité, A.A., Zlatanova, S., 2016b. Valid space description in BIM for 3D indoor navi-the particularities of Slovenian cadastral data. The IndoorGML standard gation. Int. J. 3-D Inf. Model. 5 (3), 1–17. https://doi.org/10.4018/IJ3DIM. 2016070101. is used for final outputs aiming to provide a model suitable for indoor Drobež, P., Kosmatin Fras, M., Ferlan, M., Lisec, A., 2017. Transition from 2D to 3D Real navigation and location-based services. The data stored in the database property cadastre: the case of the Slovenian cadastre. Comput. Environ. Urban Syst enables fully automatic IndoorGML document generation on request 62, 125–135. https://doi.org/10.1016/j.compenvurbsys.2016.11.002. Ellul, C., Stoter, J., Harrie, L., Shariat, M., Behan, A., Pla, M., 2018. Investigating the State while also taking advantage of all functionalities of a spatial database. of play of GeoBIM across Europe. Int. Arch. Photogramm. Remote Sens. Spatial Inf. The proposed approach is software independent and can be Sci. 19–26. https://doi.org/10.5194/isprs-archives-XLII-4-W10-19-2018. XLII-4/ 13 J. Tekavec and A. Lisec Land Use Policy 98 (2020) 104322 W10. from OpenStreetMap. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 40, Gunduz, M., Isikdag, U., Basaraner, M., 2016. A review of recent research in indoor 459–462. https://doi.org/10.5194/isprsarchives-XL-1-W5-459-2015. modelling & mapping. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 41, Munkres, J.R., 1984. In: Mixter, R.W., Harris, C.W. (Eds.), Elements of Algebraic 289–294 https://doi.org/0.5194/isprsarchives-XLI-B4-289-2016. Topology. Addison-Wesley. Hong, S., Jung, J., Kim, S., Cho, H., Lee, J., Heo, J., 2015. Semi-automated approach to Navratil, G., Unger, E.M., 2013. Requirements of 3D cadastres for height systems. indoor mapping for 3D as-built building information modeling. Comput. Environ. Comput. Environ. Urban Syst 40, 14–23. https://doi.org/10.1016/j.compenvurbsys. Urban Syst 51, 34–46. https://doi.org/10.1016/j.compenvurbsys.2015.01.005. 2013.04.001. Huang, H., Gartner, G., Krisp, J.M., Raubal, M., Van, N., 2018. Location based services: OGC, 2010. Open Geospatial Consortium. Implementation Specification for Geographic ongoing evolution and research agenda. J. Location Based Serv. 12 (2), 63–93. Information - Simple Feature Access - Part 2: SQL Option. https://www. https://doi.org/10.1080/17489725.2018.1508763. opengeospatial.org/standards/sfs. ISO, 2007. ISO 19136. Geographic Information – Geography Markup Language (GML). OGC, 2012. Open Geospatial Consortium. City Geography Markup Language (CityGML) International Standard. Encoding Standard, Version 2.0. http://www.opengeospatial.org/standards/citygml. ISO, 2012. ISO 19152:2012. Geographic Information – Land Administration Domain OGC, 2014. Open Geospatial Consortium. IndoorGML, Version 1.02. Model (LADM). International Standard. Ryoo, H.-G., Kim, T., Li, K.-J., 2015. Comparison between Two OGC standards for indoor ISO, 2016. ISO/IEC 13249-3:2016. Information Technology – Database Languages – SQL space: CityGML and indoorGML. Proceedings of the Seventh ACM SIGSPATIAL Multimedia and Application Packages - Part 3: Spatial. International Standard. International Workshop on Indoor Spatial Awareness 1–8. https://doi.org/10.1145/ ISO, 2018. ISO 16739-1:2018. Industry Foundation Classes (IFC) for Data Sharing in the 2834812.2834813. Construction and Facility Management Industries. International Standard. Seo, D., 2017. JInedit (Java IndoorGML Editor), GitHub Repository. https://github.com/ Jensen, C., Lu, H., Yang, B., 2010. Indoor-a new data management frontier. IEEE Data STEMLab/JIneditor#jinedit. Eng. Bull. 1–6. Shojaei, D., 2014. 3D Cadastral Visualisation: Understanding User’s Requirements. Jiao, J., Yuan, L., Tang, W., Deng, Z., Wu, Q., 2017. A Post-rectification approach of depth Doctoral Thesis. University of Melbourne, Victoria, Australia. images of kinect v2 for 3D reconstruction of indoor scenes. ISPRS Int. J. Geo-Inf. 6 Tang, F., Ren, A., 2012. GIS-based 3D evacuation simulation for indoor fire. Build. (11), 1–15. https://doi.org/10.3390/ijgi6110349. Environ. 49, 193–202. https://doi.org/10.1016/j.buildenv.2011.09.021. Kang, H.-K., Li, K.-J., 2017. A standard indoor spatial data model—ogc indoorGML and Tashakkori, H., Rajabifard, A., Kalantari, M., 2015. A new 3D indoor/outdoor spatial implementation approaches. ISPRS Int. J. Geo-Inf. 6 (4), 116. https://doi.org/10. model for indoor emergency response facilitation. Build. Environ. 89, 170–182. 3390/ijgi6040116. https://doi.org/10.1016/j.buildenv.2015.02.036. Kang, J., Lee, I., 2016. 3D modelling of an indoor space using a rotating stereo frame Tekavec, J., Lisec, A., 2018. A framework for assessing cadastral data as a source for 3D camera system. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 303–308 XLI- indoor modelling. In: 6th International FIG Workshop on 3D Cadastres. 2-4 October B4, https://doi.org710.5194/isprsarchives-XLI-B4-303-2016. 2018, Delft, the Netherlands. pp. 243–260. Khan, A.A., Donaubauer, A., Kolbe, T.H., 2014. A multi-step transformation process for Tekavec, J., Ferlan, M., Lisec, A., 2018. A review of research on 3D Real property ca- automatically generating indoor routing graphs from existing semantic 3D building dastre. Geod. vestn 62 (2), 249–278. https://doi.org/10.15292/geodetski-vestnik. models. In: 9th 3D GeoInfo Conference. November 2014, Dubai, United Arab 2018.02.249-278. Emirates. Teo, T.A., Yu, S.C., 2017. The extraction of indoor building information from BIM to OGC Kim, M., Lee, J., 2015. Developing a method to generate IndoorGML data from the omni- indoorGML. . Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 42 (4W2), directional image. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 40, 17–19. 167–170 https://doi.org7/10.5194/isprs-archives-XLII-4-W2-167-2017. https://doi.org/10.5194/isprsarchives-XL-2-W4-17-2015. Yang, L., Worboys, M., 2015. Generation of navigation graphs for indoor space. Int. J. Lee, J., 2004. A spatial access-oriented implementation of a 3-D GIS topological data Geogr. Inf. Sci. 29 (10), 1737–1756. https://doi.org/10.1080/13658816.2015. model for urban entities. GeoInformatica 8 (3), 237–264. https://doi.org/10.1023/ 1041141. B:GEIN.0000034820.93914.d0. Vanclooster, A., Van de Weghe, N., De Maeyer, P., 2016. Integrating indoor and outdoor Lee, J., Zlatanova, S., 2008. A 3D data model and topological analyses for emergency spaces for pedestrian navigation guidance: a review. Trans. GIS 20 (4), 491–525. response in urban areas. In: Li, J., Zlatanova, S. (Eds.), Geospatial Information https://doi.org/10.1111/tgis.12178. Technology for Emergency Response. Taylor & Francis Group, London, pp. 143–168. Wang, C., 2015. 3D Visualization of Cadastre: Assessing the Suitability of Visual Variables Lee, K., Ryu, S.H., Nam, C., Doh, N.L., 2017. A practical 2D/3D SLAM using directional and Enhancement Techniques in the 3D Model of Condominium Property Units. patterns of an indoor structure. Intell. Serv. Robotics 11 (1). https://doi.org/10. Doctoral Thesis. Université Laval, Quebec, Canada. 1007/s11370-017-0234-9. Xu, M., Wei, S., Zlatanova, S., Zhang, R., 2017. Bim-based indoor path planning con-Lenac, K., Kitanov, A., Cupec, R., Petrović, I., 2017. Fast planar surface 3D SLAM using sidering obstacles. ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci. 4 (2W4), LIDAR. Robotics Auton. Syst. 92, 197–220. https://doi.org/10.1016/j.robot.2017.03. 417–423. https://doi.org/10.5194/isprs-annals-IV-2-W4-417-2017. 013. Zhou, Y., Dao, T.H.D., Thill, J.C., Delmelle, E., 2015. Enhanced 3D visualization techni- Li, Y., He, Z., 2008. 3D indoor navigation: a framework of combining BIM with 3D GIS. ques in support of indoor location planning. Comput. Environ. Urban Syst 50, 15–29. 44th ISOCARP Congress 2008 1–10. https://doi.org/10.1016/j.compenvurbsys.2014.10.003. Lin, W.Y., Lin, P.H., 2018. Intelligent generation of indoor topology (i-GIT) for human Zlatanova, S., van Oosterom, P., Lee, J., Li, K.J., Lemmen, C.H.J., 2016. LADM and indoor pathfinding based on IFC models and 3D GIS technology. Autom. Constr. 94 indoorGML for support of indoor space identification. ISPRS Ann. Photogramm. (2018), 340–359. https://doi.org/10.1016/j.autcon.2018.07.016. Remote Sens. Spatial Inf. Sci. IV-2/W1, 257–263. https://doi.org/10.5194/isprs- Liang, X., Chen, H., Li, Y., Liu, Y., 2016. Visual laser-SLAM in large-scale indoor en- annals-IV-2-W1-257-2016. vironments. 2016 IEEE International Conference on Robotics and Biomimetics, Zlatanova, S., Sithole, G., Nakagawa, M., Zhu, Q., 2013. Problems in indoor mapping and ROBIO 2016 19–24. https://doi.org/10.1109/ROBIO.2016.7866271. modelling. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 40, 63–68. https:// Mirvahabi, S.S., Abbaspour, R.A., 2015. Automatic extraction of IndoorGML core model doi.org/10.5194/isprsarchives-XL-4-W4-63-2013. 14 PRILOGA D Simulating Large-Scale 3D Cadastral Dataset Using Procedural Modelling Tekavec, J., Lisec, A., Rodrigues, E. (2020) ISPRS International Journal of Geo-Information, 9(10), 598 DOI: 10.3390/ijgi9100598 Faktor vpliva za leto 2019: 2,239 (Q3) International Journal of Geo-Information Article Simulating Large-Scale 3D Cadastral Dataset Using Procedural Modelling Jernej Tekavec 1, * , Anka Lisec 1 and Eugénio Rodrigues 2 1 Department of Geodesy, Faculty of Civil and Geodetic Engineering, University of Ljubljana, Jamova cesta 2, SI-1000 Ljubljana, Slovenia; anka.lisec@fgg.uni-lj.si 2 ADAI, Department of Mechanical Engineering, University of Coimbra, Rua Luís Reis Santos, Pólo II, 3030-788 Coimbra, Portugal; erodrigues@uc.pt * Correspondence: jernej.tekavec@fgg.uni-lj.si Received: 1 September 2020; Accepted: 8 October 2020; Published: 11 October 2020 Abstract: Geospatial data and information within contemporary land administration systems are fundamental to manage the territory adequately. 3D land administration systems, often addressed as 3D cadastre, promise several benefits, particularly in managing today’s complex built environment, but these are currently still non-existent in their full capacity. The development of any complex information and administration system, such as a land administration system, is time-consuming and costly, particularly during the phase of evaluation and testing. In this regard, the process of implementing such systems may benefit from using synthetic data. In this study, the method for simulating the 3D cadastral dataset is presented and discussed. The dataset is generated using a procedural modelling method, referenced to real cadastral data for the Slovenian territory and stored in a spatial database management system (DBMS) that supports storage of 3D spatial data. Spatial queries, related to 3D cadastral data management, are used to evaluate the database performance and storage characteristics, and 3D visualisation options. The results of the study show that the method is feasible for the simulation of large-scale 3D cadastral datasets. Using the developed spatial queries and their performance analysis, we demonstrate the importance of the simulated dataset for developing efficient 3D cadastral data management processes. Keywords: building models; 3D cadastral system; procedural modelling; SQL; 3D visualisation 1. Introduction As the built environment is increasingly becoming spatially complex, land administration systems (e.g., cadastral systems) are challenged by an unprecedented demand to support decisions in utilising space above and below the earth’s surface [1]. Consequently, cadastral systems tend to be very complex information systems that require the efficient organisation of data to support their management. Additionally, a cadastral system is usually developed for a whole jurisdiction (e.g., a county or a country) which requires handling of a large amount of data. With the introduction of 3D geospatial models representing 3D real property units in the so-called 3D cadastre or 3D land administration system, this has become a very challenging task [2,3]. The development of 3D cadastre has been the subject of intensive research activity in the past two decades. Generally, the research has been focused on three main aspects of a 3D cadastre: the legal, the institutional and the technical. The legal aspects of 3D cadastre have been studied by Paulsson [4]. The study provides an international overview of legal systems and the analysis of several forms of 3D property rights. Kitsakis and Dimopoulou [5] studied and compared the establishment of multi-surface property in Common and Civil law jurisdictions. Paulsson and Paasch [6] analysed the research on the legal aspects and emphasised the need for additional research in this field. The institutional aspects of 3D cadastres are mainly studied within ISPRS Int. J. Geo-Inf. 2020, 9, 598; doi:10.3390/ijgi9100598 www.mdpi.com/journal/ijgi ISPRS Int. J. Geo-Inf. 2020, 9, 598 2 of 18 the research, concerning the developments in specific countries [7–11]. Ho et al. [12] studied the institutional factors in the development of 3D cadastral systems. Stoter [13] provided a comprehensive study of the 3D cadastral topics, with the focus on the technical aspects. Several options for realisation of 3D cadastral systems were investigated, including DBMS data storage and maintenance. The DBMS storage and management of 3D spatial data has been studied in [14–16] and its validation in [17–19]. The research in the field of 3D cadastre has also been related to the challenges for the visualisation of 3D cadastral data [20,21]. Regarding 3D cadastral data modelling, several data models have been studied and proposed. Aien et al. [22] provide an assessment and comparison of the existing data models for 3D cadastre applications. The authors conclude that none of the discussed models, including the conceptual model provided within the international standard ISO 19152:2012 on the Land Administration Domain Model (LADM) [23], fully supports 3D data modelling. Another study, focusing on linking physical space with legal boundaries, has been conducted by Rajabifard et al. [24] for the case of the jurisdiction of Victoria, Australia. The authors discuss Open Geospatial Consortium standards, namely CityGML [25], IndoorGML [26], and LandInfra [27], and their integration with LADM, together with the 3D cadastral data model (3DCDM), which was designed to support both 3D legal objects and their physical counterparts [28]. According to Kalogianni et al. [1], there is currently no country with a fully operational 3D cadastral system. The authors emphasise that some jurisdictions already support some aspects of 3D cadastre, but this is mainly a part of pilot projects and prototypes (see [29–31]). Consequentially, this means that no real large-scale 3D cadastral datasets are available, except for some individual cases that were the subject of existing studies. While advancements in remote sensing technology enable us to obtain large-scale models of building exteriors [32], this is not enough for cadastral use in case of buildings, as some degree of indoor spatial information is needed for unambiguous delineation of a building into real property units, to which rights, restrictions, and responsibilities refer. Indoor spatial data acquisition and modelling is challenging and requires an individual approach for each building [33]. Studies regarding this issue have also been done related to a 3D cadastre. Kitsakis and Dimopoulou [34] investigate the usability of existing cadastral documentation for 3D modelling. Vučić et al. [8] focus on the Croatian land administration system and investigate the possibilities of upgrading the cadastral system to 3D by linking topographic data. Tekavec and Lisec [35] provide a framework for 3D indoor modelling based on existing cadastral data. All studies deal with a limited number of individual buildings. Consequently, their usability in obtaining a large-scale dataset is limited. To overcome this issue, it is hypothesised that the 3D cadastral system development can benefit from using a synthetic dataset that simulates some aspects of the real data. In this paper, the method that uses procedural modelling for simulating the synthetic 3D cadastral dataset is developed and discussed. The research is limited to buildings, as they represent the dominant feature in the field of geospatial data modelling within a 3D cadastre. Using the existing cadastral data for the study area, we aimed to obtain a more representative placement and configuration of the simulated buildings. A similar approach, without procedural modelling, was proposed by Ledoux and Meijers [36] and Biljecki et al. [37] aiming to generate 3D city models based on 2D footprints and attribute data. Procedural modelling methods are rooted in computer graphics and refer to several techniques to create 3D models and textures from sets of predefined rules, combined with randomising some of the parameters. We focus on methods that generate whole buildings from indoor space arrangements. In this context, the study of Watson et al. [38] has to be mentioned, who reviewed the applications of procedural modelling in computer graphics. Merrell et al. [39] focus on the automated generation of building layouts based on a Bayesian network trained on real-world data. Smelik et al. [40] provide a survey of the methods for procedural modelling of virtual worlds. Camozzato [41] classifies the approaches to subdivision algorithms [42], tile placement algorithms [43], inside out algorithms [39] and growth-based algorithms [44]. Another classification of automatic approaches for indoor layout generation is provided by Rodrigues et al. [45]. The focus is on the architectural domain, where procedural modelling is divided into area assignment, area partitioning, space allocation, hierarchical construction, conceptual exploration ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 3 of 19 ISPRS Int. J. Geo-Inf. 2020, 9, 598 3 of 18 design adaptation. Procedural modelling is also used in the geoinformation domain. Gröger and Plümer [46] generated building indoor models to derive route graphs. Zhu et al. [47] used procedural and design adaptation. Procedural modelling is also used in the geoinformation domain. Gröger and modelling to generate virtual scenes. Based on the work of Tsiliakou et al. [48], Kitsakis et al. [49] Plümer [46] generated building indoor models to derive route graphs. Zhu et al. [47] used procedural used ESRI CityEngine to produce 3D building models of a traditional village. CityEngine procedural modelling to generate virtual scenes. Based on the work of Tsiliakou et al. [48], Kitsakis et al. [49] capabilities have also been studied by Ribeiro et al. [50] to test 3D visualisation of 3D cadastral data used ESRI CityEngine to produce 3D building models of a traditional village. CityEngine procedural and by Neuenschwander et al. [51] to generate and visualise green space patterns in a 3D capabilities have also been studied by Ribeiro et al. [50] to test 3D visualisation of 3D cadastral data environment. Biljecki et al. [52] used the procedural modelling approach to simulate the CityGML and by Neuenschwander et al. [51] to generate and visualise green space patterns in a 3D environment. datasets, which were used to analyse different variants of LODs (level of detail) and their influence Biljecki et al. [52] used the procedural modelling approach to simulate the CityGML datasets, which were on the performance of spatial analyses. The procedural modelling engine Random3DCity was used used to analyse different variants of LODs (level of detail) and their influence on the performance of by Kumar et al. [53] to generate artificial TIN terrain models at different LODs in CityGML format. spatial analyses. The procedural modelling engine Random3DCity was used by Kumar et al. [53] to In the following chapter, we present the developed approach for simulating the real 3D cadastral generate artificial TIN terrain models at different LODs in CityGML format. dataset by generating a synthetic dataset using a Hybrid Evolution Strategy [54] as a procedural In the following chapter, we present the developed approach for simulating the real 3D cadastral modelling method to create buildings with different geometries and functional programs. The dataset by generating a synthetic dataset using a Hybrid Evolution Strategy [54] as a procedural generated buildings are linked to the Slovenian cadastral data. The simulated dataset was stored in modelling method to create buildings with different geometries and functional programs. The generated a spatial database management system (DBMS), where we illustrate the performance testing for some buildings are linked to the Slovenian cadastral data. The simulated dataset was stored in a spatial 3D spatial queries related to data consistency and we evaluate options for 3D visualisation. database management system (DBMS), where we illustrate the performance testing for some 3D spatial queries related to data consistency and we evaluate options for 3D visualisation. 2. Materials and Methods 2. Materials Our and study Methods combines reality‐based and procedural building modelling. Namely, the simulation of the cadastral dataset is based on available data about existing buildings, acquired from the official Our study combines reality-based and procedural building modelling. Namely, the simulation records at the National Surveying and Mapping Authority, combined with the simulated data of the cadastral dataset is based on available data about existing buildings, acquired from the generated by procedural modelling, where randomised 3D building models are constructed based official records at the National Surveying and Mapping Authority, combined with the simulated data on the predefined rules. The developed approach aims to satisfy the following requirements: (i) the generated by procedural modelling, where randomised 3D building models are constructed based process of generation of the synthetic dataset is guided by the existing cadastral data from the current on the predefined rules. The developed approach aims to satisfy the following requirements: (i) the cadastral database; (ii) the 3D building models in the dataset should be randomised; and (iii) the process of generation of the synthetic dataset is guided by the existing cadastral data from the current dataset should be stored in a spatial DBMS, where data storage, access, maintenance, and cadastral database; (ii) the 3D building models in the dataset should be randomised; and (iii) the dataset manipulations can be performed and evaluated. Using the existing cadastral data for the study area, should be stored in a spatial DBMS, where data storage, access, maintenance, and manipulations can we have tried to obtain more representative placement and configuration of the simulated buildings, be performed and evaluated. Using the existing cadastral data for the study area, we have tried to where we focus on residential buildings. Figure 1 summarises the framework of our study. The obtain more representative placement and configuration of the simulated buildings, where we focus process starts by defining the type of residential units. This information is then used for generating on residential buildings. Figure 1 summarises the framework of our study. The process starts by the synthetic dataset using the procedural modelling method [54]. The generated models are then defining the type of residential units. This information is then used for generating the synthetic dataset transformed to fit into a GIS environment to be suitable for data storage in the spatial DBMS. The using the procedural modelling method [54]. The generated models are then transformed to fit into procedurally generated 3D building models are then linked to real cadastral data on buildings by a GIS environment to be suitable for data storage in the spatial DBMS. The procedurally generated location and the number of storeys. In this study, spatial DBMS is created and run locally. Its structure 3D building models are then linked to real cadastral data on buildings by location and the number of is focused on the storage of the simulated 3D dataset in the form building models, including their storeys. In this study, spatial DBMS is created and run locally. Its structure is focused on the storage interior spaces, i.e., rooms. Due to different legal frameworks, historical background and other of the simulated 3D dataset in the form building models, including their interior spaces, i.e., rooms. factors, countries have various definitions of real property units as well as various implementations Due to different legal frameworks, historical background and other factors, countries have various of cadastral databases. The stored simulated data are designed to allow linking data with additional definitions of real property units as well as various implementations of cadastral databases. The stored country‐specific data. simulated data are designed to allow linking data with additional country-specific data. Figure 1. Study concept framework. Figure 1. Study concept framework. ISPRS Int. J. Geo-Inf. 2020, 9, 598 4 of 18 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 4 of 19 The The study study is is generic generic with with respect respect to to procedural procedural modelling modelling of of 3D 3D building building geometries geometries and and their their transformations. transformations. The The reality-based reality‐based part part of of the the study study is is related related to to the the Slovenian Slovenian cadastral cadastral data data with with its its specific specific featur features es [55 [55].]. For For the the case case study study,, all all rresidential esidential buildings buildings in in Slovenia Slovenia stored stored in in the the Slovenian Slovenian building building cadastr cadastree were were used. used. Based Based on on the the simulated simulated data, data, we we evaluated evaluated the the performance performance of of several several queries queries aimed aimed at at 3D 3D cadastral cadastral database database maintenance maintenance and and data data quality quality contr ol. control. 2.1. 2.1. Procedural Procedural Modelling of Buildings Buildings For this study stud , y, the Evolutionary Program Program for the Space Space Allocation Allocation Problem Problem (EPSAP) (EPSAP) algorithm algorithm was used [54 [5 ]. 4]. The EPSAP EPSAP algorithm algorithm consists consists of a two-stage two‐stage appr a oach pproach that that has has an Evolution Strategy Strategy (ES) (ES) framework, framew wher ork, e the where mutation the muta operation tion is r operationeplaced is by replac a ed Sto bychastic a Hill Stocha Cli stic mbing Hill C (SHC) limbin metho g (S d. HC) Figure 2 method. depicts Figure the 2 overall depicts generation the overall pr ge ocess. nerationThe algorithm process. The starts algorit by hm cr eating starts a by set of candidate creating a set of designs that candidate have design their s thatr ooms have randomly their dispersed rooms random in ly the dis 2D per space sed in on the each 2D stor sp ey ace . Then, on each each design storey. is Then, evaluated each desi accor gn is ding to a weighted-sum evaluated according to cost a function weighted‐ of eighteen sum cost fu penalty nction functions of eighteenthat are minimised. penalty functions At this that moment, are the minimised. second At this stage starts moment, the by iteratively second stage changing starts by each iterat design ively with chan random ging each geometric design with and topologic random geomtransformations, etric and topolo such gic as tran translation, sformations, r otation, such as str tr etching, anslation,r eflection, rotation, and swapping. stretching, re If the flection, transformed and swappi design ng. If pr the oduces transf an impr ormed oved design solution, the produces new an design improved is sopreserved. lution, the When new the de SHC sign is stage is incapable preserved. When of finding the SHC further stage is impr incap ovements, able of fin the ding second further stage ends and improvements, the first stage second resumes stage ends by selecting and the first solutions stage resu that mes have by se a fitness lecting sollower utions than that the average have a fitnes fitness s lower of all than designs. the averag The e fit worst ness of solutions all designar s. e replaced The worst by so new lu randomised tions are repl ones, aced by which new ra ar nde also omise evaluated. d ones, wh This ich arcycle e alsois then ev repeated aluated. This until cycle ES is stage then reaches repeated a maximum until ES number stage re of aches iterations a ma or ximum is unable number ofto it find erat mor ions e or impr is oved unable design to find solutions more (for the improved complete design sodescription lutions (for of the the EPSAP algorithm complete descripti see on [54 of ]). theLastly , EPSA the P whole algori 3D thm building see [54]). is produced Lastly, the by whoadding le 3D height building to is each storey produced (measur by addi ed ng from height the to stor each ey floor storey to the (measu ceiling), red from extr the uding storey rooms floor for to one the ceor mor iling), e storeys extrud (e.g., ing a stair rooms forserves one a or range more of stor storey eys), s (e positioning .g., a stair the serves openings a range in of relation storeys), to the room positioning floo the r (i.e., the openings distance in fr relation om to the the floor room to floo the r (i. base e., theof the dista opening) nce from vertically the floor , to and the giving base of height the to the opening) opening vertica (i.e., lly, an the d gi void ving height height in to the the wall). open In ing the (i.e. case , the when void a heistor ght ey in multiplier the wall). is In pr th esent, e case the stor when a ey is r store epeated. y multiplier is present, the storey is repeated. Figure Figure 2. 2. Evolutionary EvolutionaryProg Pr ram ogram for for the the Space Space Allocation Allocation Problem Pr oblem (EPSAP) (EPSAP) algorithm algorithm workflow workflow (adapted (adapted from from ref. ref. [54]) [54 . ]). ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 5 of 19 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 5 of 19 ISPRS Int. J. Geo-Inf. 2020, 9, 598 5 of 18 For For our our study study ,, six six residential residential bu bu il il d d iing ng types types (t (t wo wo types types of of det det ached ached houses, houses, three three types types of of multi multi ‐‐ storey storey ap ap art art m m ent ent build build in in gs, gs, and and aa sin sin g g le le type type of of multi multi ‐‐storey storey ap ap ar ar ttm m ent ent build build in in g g with with comm comm erce erce on on For our study, six residential building types (two types of detached houses, three types of the the ground ground floor) floor) were were ge ge nerated. nerated. Two Two of of the the bu bu il il d d iin n g g types types are are detached detached sin sin g g le le ‐‐fa fa mil mil y y ho ho uses uses of of multi-storey apartment buildings, and a single type of multi-storey apartment building with commerce one one and and two two storeys, storeys, respe respe cctively. tively. The The si si ng ng llee‐‐store store y y house house comprises comprises bedroom bedroom s, s, bathrooms, bathrooms, livin livin g g on the ground floor) were generated. Two of the building types are detached single-family houses of room, room, kitchen, kitchen, corridor, corridor, and and entrance entrance hall. hall. The The two two ‐‐storey storey bu bu ildi ildi n n gs gs ha ha v v ee an an int int eerior rior stair stair tha tha tt one and two storeys, respectively. The single-storey house comprises bedrooms, bathrooms, living separ separ aates tes the the public public pa pa rrtt of of the the house house from from the the bedrooms, bedrooms, which which are are locat locat eed d in in the the upper upper storey. storey. room, kitchen, corridor, and entrance hall. The two-storey buildings have an interior stair that separates Three Three of of the the bu bu iilld d iing ng types types are are mult mult ii‐‐stor stor ey ey ap ap art art m m ent ent build build ing ing ss of of fou fou rr storeys storeys .. Th Th eessee bu bu il il d d iings ngs are are the public part of the house from the bedrooms, which are located in the upper storey. Three of the different different in in th th ee number number of of ap ap art art m m ent ent ss per per floor floor an an d d the the nu nu mbe mbe rr of of bedroo bedroo ms. ms. The The la la st st type type of of building types are multi-storey apartment buildings of four storeys. These buildings are different build build ing ing is is an an eight eight ‐‐store store y y build build ing ing tha tha tt has has commerce commerce space space ss on on the the ground ground flo flo or or and and access access to to the the in the number of apartments per floor and the number of bedrooms. The last type of building is an roof roof on on the the top top floor. floor. On On the the iin n termedia termedia te te floors, floors, two two th th rreeee‐‐bedroo bedroo m m ap ap art art m m ent ent ss are are serve serve d d by by aa stair stair eight-storey building that has commerce spaces on the ground floor and access to the roof on the top and and an an el el eev v aattor or .. floor. On the intermediate floors, two three-bedroom apartments are served by a stair and an elevator. 2.2. Transforming the Generated Building Models 2.2. 2.2. T ransforming Tr the ansforming Generated the Genera Building ted Build Models ing Models The building models comprise enclosed volumes with surfaces for walls, floors, ceilings, doors, The building The build models ing modelscomprise compriseenclosed enclosedvolumes volumeswith withsurf su ac rfa es cesfor forwalls, walls,flo ors, floors,ceilings, ceilings,doors, doors, and windows (see Figure 3). The adjacent rooms are in a touching relationship, having common and andwindows windows(see (seeFigur Figuere3 ). 3). The The adjacent adjacent r ooms rooms ar e are in ina atouching touchingr elationship, relationship,having havingcommon common interior walls, slabs, and door surfaces. interior inte walls, rior slabs, walls, slab and s, door and surfaces. door surfaces. Figure Figure Figure 3. 3. 3. Simul Simul aated Simulated ted two two ‐‐store two-stor store y ey y bu bu ilding building ilding mod mod eel modell com com p p rising comprising rising of of of wall, wall, wall, door door door and and and window window window su su rfaces surfaces. rfaces .. The The da da ta data ta model model of of the the sim sim u u lated simulated lated bu bu ild ild in in gs buildings gs ca ca n can n be be li li nked linked nked to to the the LADM LADM da da ta data ta mo mo del, model, del, as as presen pr presen ted esented ted in in Fig Fig u u re Figure re 4. 4. 4. Th Th e Thee si si mul mul aated simulated ted models models of of buildings buildings repre repr repre ssent esent ent thei thei r theirr physical physical co co mpo mpo n n een n ts components. ts .. It It mu mu sst mustt be be noted noted tha tha t thatt the the the boun boun dar dar iies boundaries es of of r of re re al ealal property property property un units un its its may may may not notnot co co incide coincide incide wi with wi th th the the the physical physical physical fe featur fe aat es t ures of a ures of of aa building building [56]. building [5 [5 6] 6] .. Howeve However, in Howeve r, r, in case in ca of ca se se of of buildings, buildings, a str buildings, aa strong ong link strong link link exists ex ex ist ist ss between between between physica physical and physica ll and legal and le le g g aall sp spaces, sp aces aces ,, wh whichwh ich has ich has been has been been stu studied stu d d ied byied by Aien by Ai et Ai een al.n [ et et 28 al ], al .. [2 Li [2 8] et 8] ,, Li al. Li [ et et 57 al ]al .. [5 and [5 7] 7] and and Knoth Knot et Knot h al.h [ et et 58 al ]. al .. [5 [5 8] 8] .. These Th Th eessee stud studies stud ies also ies al al so so inc include inc llud the ud ee the links the li toli nks the nks to to the the LADM LAD LAD M M standar stan d stan dar data dar d d data model data model that are model tha tha tt ar similar ar ee to si si mila our mila rr to study to . ou In ou rr stu stu d ecent d y. y. In In recent years,recent many years, years, ma studiesma ny ny stud stud ies investigated ies invest the invest ig ig at at ed potential ed th of th ee potential integrating potential of of int int eeg g r building r at at in in g g build build ing information ing in in fo fo rmat models rmat ion (BIM), ion models which models (BIM (BIM ), contain), s wh wh ich ich contains physical contains ph ph y y s information s iiccaal onl inform inform at at ion buildings,ion on with on buildings, 3D cadastr buildings, e with [59– with 3D 61]. 3D ca In ca das das t thistrree [[59– study, 59– 61 61 ]. the ]. In rIn thi ealthi s prs stu stu d d y, operty y, th th ee re unitsre al are al property realised property un by un its the its are are re aggrre aalliissed egation ed by of by r the the aggr oomsaggr e thate g g ation ation belong of to of roo the roo m m ss corr tha tha tt belong esponding r belong to eal to the pr the co co rrespondin operty units. rrespondin g g re Therre aall efor property e, the property units. units. There investigation is There ffore, ore, the limited to the invest invest ig ig at physically at ion ion is is lim defined lim it it ed 3Ded to to p p h h ysic ysic al boundaries. al ly ly defined defined 3D 3D boun boun dar dar iies. es. Figure 4. Linking Land Administration Domain Model (LADM) classes (green) to the data model of Figure Figure 4. 4. Link Link iin n g g Land Land Ad Ad mini mini ssttrati rati o o n n Dom Dom aain in Mode Mode ll (L (L AD AD M M )) cla cla sse sse ss (green) (green) to to the the dat dat aa mod mod eell of of simulated buildings (white). sim sim u u lated lated build build iin n gs gs (white) (white) .. ISPRS Int. J. Geo-Inf. 2020, 9, 598 6 of 18 ISPRS ISPRS Int . Int .J. J. Geo ‐ Geo Inf. ‐ Inf. 2020 , 2020 , 9 , 9 ,x x FO R FO R PE ER PE ER RE V RE I V EW I EW 6 6 of of 19 19 After Aftertheir theirgeneration, ge neration, ge the the3D 3Dbuilding bu ild bu ing ild models modelsar e aretransformed tra n tra sf n ormed sf into ormed in t in o t a adata da ta da format form at form that tha t tha is issuitable s u s it u a it b a le b le for forstorage storage stor in ina a spatial spatial sp DBMS. DBMS.First, Firs Fir t, s each each eac building building build needs needs ne an an assignment assi as gn si m gn e m nt e of of a a unique uni un q i ue q identifier identi ident fie i r fie followed foll fol owed l owed by bythe thedecomposition decompositionof ofthe thegeometrie s geometriesinto intoindividual individ u individ a u l a parts. pa rts. pa The The decomposed de c de o c mpos o e mpos d e geometries g e g ometries e enabl en e ablethe the application applicat applic ion at of ion the of of filter the the fi lt fi that er er selects that that only selects selects on surface ly ly sur f sur type ace ace geometries. type type The geometries. geometries. selected The Th e se surfaces lected lected su rfa su ar c e es es grouped are are gr by ouped ouped the by by room the the identifier room room , which identifier, identifier, enables which wh ich the enables enable s modelling the the of solid modelling modelling of of geometries solid solid ge for ometri ometr e i s e each fo r fo r room each each (see room room Figur (see (see e Fig5). ure u re 5). 5). Figure Figure 5. 5. Sol i Sol d Solid i d geometries geometries cre cr a cre ted eated a ted from fr om from the the predefined pr edefined predefined su rfaces su . surfaces. rfaces . The The The gener a gener te a d generated te d 3D 3D 3D building building building models models models ar ar e e ar e pla c pla ed placed c ed in in in a a a lo cal local lo cal coordinat coor e dinate coordinat e sy stem system sy stem but but but are ar e are not not not cent r cent a r lly centrally a lly plac pla ed. placed. c ed. The The placement placement of of the the models models varies varies re r lativ re e lativ elative e to to coordin coor a coordin te a dinate te origin, orig origin, in, as as shown shown in in Figu Fig re Figur u e re 6a 6 . 6a a. . Therefore Therefore,, Therefore , the the the central central central point point is point is is calc ca ul calculated lc a ul t a e t d fore d for each for eac ea h c h generated generated 3D generated 3D 3D building, building, based on building, based the ba r sed on oom on the the room geometries. room geometries. The Z coor geometries. Th e Th e Z dinate Z co of ordin the co a ordin te a te of central of the the central point is set central point to the point is is set set to to the minimum the minim Z u value minim m of u m Z Z value the value of of the bounding the bounding box. The r bounding box. oom box. The The room room ge geometries ometries ar ge e then ometries ar e ar e th moved e th n e n moved using an o moved us ff i us n set i g n g an an of vector fs of e fs t e tvector connecting vector conn the ecting local conn ecting the coo r the loc a loc l a dinatel coordin system a coordin te a te sy s origin sy t s em and t em origin the origin and and the calculated the calculated central calculated ce point, n ce t n r ast a r l a lpoint, shown point, as in as shown Figur e 6 shown in b. in Fi g Fi u g re u re 6b . 6b . Figure Figure Figure 6. 6. 6. ((a (a) a) )Or O iginal r Original iginal placement placem placement ent of of of generated generated generated 3D 3D 3D building building building mode mod ls; e models; ls; ((b (b) b) )the th the e same sam same e mode mod ls e models ls after after after their their their placement placement placement to to to the the the loca l loca local lcoordi coor nate coordi dinate nate sy stem sy system stem ori g ori in. g origin. in. The The The 3D 3D 3D building building building geometries geomet geometries ar rieseare are then then then inse ins inserted rt e ed rt in ed in the in the the PostgrPostgre eSQL S Postgre QL S QL DBMS DBMS with DBMS with wi th PostGIS PostGIS and PostGIS and SFCGAL and SFCGAL SFCGAL extensions, extensions, which extensions, whic wh support h ic h support 3D spatial support 3D 3D data spatial spatial storagedata da and ta storage storage and an d maintenance.maintena ma Before nce. intena nce. Befor inserting 3De Befor e inserting insert ing 3D geometries, to 3D geometries, avoid geometries, issues to to avo with i avo d i d issu is e su geometry s e s with wi th geom conversion etry geom from etry conversion spatial ETL conversion from from sp software atial sp atial (FME ETL ETL soft Desktop ware soft ware (FME Softwar (FME Desktop e)-supported Desktop Software) Softwa ‐ re) supported ‐ geometries to supported geo PostGIS, m geo the etries m etries to to Po geometries stGIS, Powere stGIS, the the geo m geo etries m triangulated. etries were were triang ulated. triang ulated. Besides Besid Besides es the the the building building building inte int r e i r or, i interior, or, a a a3D 3D 3D model model model of of of the the the buildi bu n ildi g n building g exterior exterior exterior extents extents extents is is is deri der v i e v d e derived d from from from the the the generated generated generated 3D 3D 3D bu ild bu ing ild building ing mo dels. mo models. dels. A A A (bu (buffe (buff red) ffe ered) red) 3D 3D 3D model model model of of of the the the building building building in in in the the the context context context of of of a a a 3D 3D 3D cadastre cadastre cadastre can can can also also also serv e serv serve eas as as a a a representa repr ti representa on ti esentation on of of of a a a bu i bu lid l i d ng i building ng le ga le l ga legal lspace. space. space. First, First, First, th e th the e2D 2D 2D o u o t u line t outline line has has has to to to be be be gen e gen rated e generated rated from from from room room room g e g ometries. e geometries. ometries. All All All su rfa su c rfa es c es surfaces of of of th e th e the building building building model model model need need need to to to be be be tra n tra sf n ormed sf ormed transformed int o int o into 2D 2D 2D sur f sur ace f s ace . s . surfaces. The The The vertical vertical vertical surfa su c rfa es c surfaces es tran tra sform n transform sform int into o int o lin line e lin e shape sh shape ape and and and thus thus thus need n need to eed to be to be be filtered filtered out. filtered out. out. Our Our Our appr approa a oach ch pproa has ch has ha been s been to been to to ca l ca c l calculate u c l u a l 2D t a e t e 2D 2D sur surface f sur ac f ar e ac ea earea and area and filter and fi lt fi all er lt er al l al lsur surfaces f sur ac f es ac es hav having i hav ng i zero ng zer or o zer o or very or very very smallsm ar all smea all are a are a (comput a (comput tional a (computational error tional error and r error and and rounding ounding). The ). roundingr ). The The r e r main e emaining 2Ding main ing 2D 2D sur surfaces f sur ar ace fe s ace s are are disso dissolved lved disso into lved int one o int 2D o one one 2D surface 2D su r rfa su epr c rfa e c e representing esenting the representing the building the building build ing 2D 2D outline out line (Fig (F u ig r u e r e 7). 7) . The The 2D 2D building building outline out line can can be be buffered buffered to to enclose enclose all al l room room ISPRS Int. J. Geo-Inf. 2020, 9, 598 7 of 18 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 7 of 19 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 7 of 19 2D outline (Figur geometries. The e 7 3D). r The epr 2D es building entation of outline the can buildingbe is bu ge ffered to nerated enclose by ex all room truding the geometries. 2D building The out 3D line. repr Theese ntation geometries. extrusion of The the 3D building r vectore pr is es calc is enta generated tion ulated of the from thby e extr build uding ing is vertical ge the 2D bounds of building nerated by the ex outline. truding room theThe 2D geometries.extr If usion building we opt vector ou tline to us .e is calculated The buffering, fr extrusion we om canthe vector vert is plac e ical calc the bounds ulated 2D of from outline the the r oom below geometries. vertical the bounds minimum If of we the opt vertical to room use bound bu of ff ro ering, geometries. om If we we can opt place to geometries use and the 2D buffe outline ring extrude , we over below can the ma the place xim minimum the um 2D vertical outline vertical bound below bound. The of the geo room metry geometries minimum of vertical each 3D and extr bound ude of building ro over om exterior the maximum geometries model and should vertical extrude also be bound. over moved The the ma us geometry xim ing u the m disp of lac each vertical e me 3D nt building bound. The ge between oexterior m the etry of central model each point should 3D and cooralso building dina be te moved exterior system using model or igin the should and displacement also tria be ngulated between moved using before the the being central displacepoint me inserted nt into and spati coor between al dinate the DBMS. system central origin point and and cootriangulat rdinate ed befor system or e ig being in and inserted tria into ngulated spatial before DBMS. being inserted into spatial DBMS. Figure 7. 3D room geometries (random colours), 2D building outline (yellow) and extruded 3D Figure Figure bu 7. 7. ilding 3D 3D r r exteri oom oom or geometries geometries geometry (random (random (transparent colou colours), blue). rs), 2D 2D building building ou outline tline (y (yellow) ellow) and and extr ext uded r 3Duded buildi3D ng building exterior exterior geometrygeometry (transpar (transparent ent blue). blue). In the spatial DBMS, two tables are created, one for storage of generated 3D models of rooms and In In the the one sp for atial spatial 3D DBMS, DBMS, models of two two ta build bl ing es tabless are are cr cre (Figure ated, eated, 8). one one These for for two stora storage tabl g es e of of are gen li erated generated nked 3D by 3D the mo modelsdels of r commonof rooms ooms and building and one one for model for 3D id 3D models models entifier ( i of d_bld of g_ build buildings gen). ing The s (Figu (Figure 3D 8). ge re 8). TheseThese two ometries of rotwo tables oms ta bl are are es are store linked dli nked by as the by th polyhedre alcommon common surfaces ,building building model together model with id identifier theentifier ( uniq ( ue i d_bl id_bldg_gen d ). g room _ i g d e The n). 3D ent The ifier 3D ( id ge _r ometries geometries oo of m_ r gen), of ooms ro are the oms stor are ed identi store as fier of d as the polyhedr polyhedral correspo al n surfaces,surf ding aces buil , together together with ding the and with the unique r uni oom optionally q the ue room identifier number ( id of ent id_r the ifie r ( id oom_gen), storey _r in oo the m_ge which n), the the identifierro of ident the om is ifier corr locat e of d ( the sto correspo esponding rey_numbe n building r ding and ). For bu theil di bun ilg d optionallyi and the ngs, optionally number the 3D of ge the the number stor ometry ey is in al soof the which stor storey the ed asr a in oom which is polyh edral the located ( su room stor rface. is locat ey_number Alongs e ). d ide ( sto For ge rey_n the umb ometry, e the r). buildings, a For the ttrib the 3D utes building geometry s, is containing the also3D ge stor uniq ometr ed as ue a y is also identifier stor polyhedral of th ed e as a surface. polyh building edral Alongside and su the rface. nu mbeAlong geometry,rthe of side ge storeys ometry, attributes are in the serte containing da. ttri the butes containing unique identifier Optionally, the 2D the of uni the que geometries identifier building of and rooms of the th and e building number buildof ings stor and eys can the ar be e numbe stored forr inserted. of storey applicat s Optionally, ion ar s, e the insert 2D where ed on . ly Optionally, geometries 2D of ge r the ooms ometries 2D and are geometries buildings needed or of can rooms be stor supported. and ed for buildings can applications, be stored where only for 2D application geometries s, arwhere e on needed ly or 2D geometries supported. are needed or supported. Figure 8. Database Database management system system (DBMS) (DBMS) storage storage of the transformed proce pr dur ocedurally ally generated Figure 3D bu 8. Data ilding building base model management s. models. system (DBMS) storage of the transformed procedurally generated 3D building models. 2.3. 2.3. Georefer Georefer encing encing Using Using the the Link Link to to the the Existing Existing Cadastral Cadastral Data Data 2.3. Georefer All the encing Using transformed the Link to building the Existing models are Cad stor a ed stra in l Data the All the transformed building models are stored in the spatial spatial DBMS DBMS in in a a local local coordin coor ate dinate system. system. All the However, However, tra this this n is sformed is not not closebuilding close to to the the r models real eal case case ar of e of stored 3D 3D ca in da the str cadastralal spatial data data. . The DBMS The in gene a generated lo rated cal bu coordin ilding building ate models sy modelsstem. need need to However, to be be spat this ia spatially lly r is efer not close reference enced d and to the and real ca random randomised, se ise d as of , 3D as stat edcada stated in str in the al the da ta init initial . iaThe l ai aims gene ms of of thisrated this bu stu study. d ild y. ing This This models pa part i rst is need based based on to onbe the spat the ially reference characteristics characteristics of d of and Sl random ovenia Slovenian n ise cad cadastrald a , as stral datastated da on ta in onthe initial buildings—the aims of buildings—the thi Slovenian s stu Sloveni d a y. n This building par building t cadastr is ca e. ba da sed stre Ther . e, on the the There, characteristics the bu buildings ild ar ings e stor of are ed Sl in ovenia stored the in n form cad the of a a fo stral rm of vector da a ta layeron vector buildings—the la with yer with building Sloven building outline i ou an building tline polygons and an ca polygons dast and re. an attribute There, attrib the ute database. bu A ild data ings base. common are A stored building in common the fo identifier rm of building a vector ident connects ifie the rla two yer with connects datasets. building the Models ou two are tline data geor polygons sets. eferenced and Models by an are two attribute dat georeference parameters to a d base. by the r eal A two common para (Slovenian) meters building to the data: (i) real ident (Slo building ifier veni a connects n) da location, ta: (i) defined the in two building the dat loca a ti national sets. on, Mode defined geodetic r ls in efer are the ence georeference national coor geo dinate d by detic system two re D96/para ferenc TM meters e based to coordin on athe te real ETRS89; (S system lo and veni D96 (i / i) Tan M the ) da ba ta: sed (i number ) on building of storeys. ETRS89; loc ana The t d ion, (ii) defi the building ned in numbe the r location of is national storeys. geo The calculated fr detic om a build referen ing lo centr c ca oid eti coordin on of ais cal acte ula building system ted D96 from polygon.a /T ce M nt Those r ba oi sed d of centr on a oids ETRS89; buil ardeing and consider (i ed i) polygon. as the numbe Those their ce refer r of ntroids ence storeys. are con centre s in The id the build ered as local ing thei coorlo r ca re t feion dinate is rence cal system cula centre te in d origin. from the loc a al ce co ntroid ordina of te a sy build stem ing orig polygon. in. Those centroids are considered as their reference centre in the local coordinate system origin. ISPRS Int. J. Geo-Inf. 2020, 9, 598 8 of 18 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 8 of 19 The The r eferenci referenc ng ingi s isp erforme perform d e u si us n i g n S QL SQLq ueri quer es ie a s as a ll al t he the d ata dat ( rea (re l a c ada cad s a tral stral d ata dat a nd and g en ge er ne at ra ed te buildi build ng ing m odels) models) a re are s tored store i n in t he the D BMS. DBMS. F or For t he the s imu sim l u ated lated c ad ca as da t s r t a r l a d at da as ta et, set, w e we h ave have c rea cre t a ed te t wo two tab ta le bl s, es, fir fi st rst for for data dat on on the the building building int in e t r e ior rior (r ( oom_si room_s mulated) imulated and and second second for for the the building building exter exte i r o i r or (b ( uildi build ng_s ing_ imulated) simulated (Fig (Fi u g re ur 9). 9) The The two two ta ta bl bl es es ar are connected connected by by the the cada cad s a tral stral bui bu ldi ild ng ing ide id nti ent f i ier fie (i(d_b id_ u b ildi uild ng) ing.).U sing Usin t gh e id the _ building id_build , th ing e , ava the ilabl ava e il c a a b d lea stra cad laattrib stral u a tes ttri c ban be utes add can ed be o pt ad iona ded lly, for instan optionally, ce, for the i inst dent ance i,f ier theo f the p identi arc fiere l to of wh the i ch pa the rcel b ui to l ding is which rela thet ed. For ea building ch is building related. , th Foe r corre eachs ponding d building, ata the on building in correspondin te g rior data( roo on m geome building trie int s e ) is rior i nserte (roomd into geom the roo etries) m _ is simu inse la rt ted ed ta intb ol e. T the here, the ident room_simulate i d f ier ta o bl f e. the rea There, l pro the pe id rty ent is s ifie t r or ofe d, w the hic realh is used t property o a is ggreg store a d, t e the r whicho o is m s us in edt o r to eal p aggr rop eg e at r e t y u the n its. O roomsp tion into ally, real the storey property num unit b s.e r where the Optionally, r oom the is loca storey ted nu can mbe b r e stored where . Va the rious roomo p ist ions f locat o e r d D B ca M n S s be torage o stored. f sim Variilar ous data are options dis for c uss DB ed MS i n [35]. storage of similar data are discussed in [35]. Figure Figure 9. 9. DBMS DBMS storage storage for for the the simulated simulated 3D 3D cadastral cadastral dataset. dataset. Each Each building building from from re r al eal cadastral cadastral data data ((id_building id_building)) gets gets one one proc pr edurally ocedurally generated generated 3D 3D model model assigned assigned ((id_bldg_gen id_bldg_gen), ), which which is is moved moved to to the the locatio location n of of the the building. building. There There arare e various various approa appr ches oaches to to the the selectio selection n of of the the co corr rresponding esponding pr procedura ocedurally lly generated generated model: model: • Generate Generate only only one one 3D 3D model model for for each each parameter: parameter: in in our our case, case, one one model model for for each each number number of of storeys. storeys. The The same same 3D 3D model model is is selected selected for for all all buildings buildings with with the the same same number number of of store stor ys. eys. • Generate Generate aa given given number number of of 3D 3D models models for for each each parameter parameter (number (number of of storey stor s). eys). One One of of the the 3D 3D models models is is randomly randomly selected selected and and re r feren efer ced enced to to re r al eal cadastral cadastral data data.. • Generate Generate a a random random 3D 3D model model for for each each re r al eal building. building. The first first approa appr ch oach is is the the simples simplest t and and ca can n be be used used in in case cases s where where the the diversity diversity of of the the models models is is not not a k a ey key i issue. ssue. SQL SQL statem statements ents (Appe (Appendicesndix A.1 A1 and and A.2) A2) are are used use to d to select select the the data dat that a ar etha to t ar be e to be inserted insert into ed the into the building_s building_simulated imulated and r and room_simulate oom_simulated d tables tabl (the es r (t eal he real cadastral cadastral data da are ta are stor stored ed in thein the table table cadastral_build cadastral_building). ing In ). In both both SQL SQL queries, querie the s, ST_Tthe ST_Translate ranslate function funct is i usedon to is use move d the to pr move the ocedurally procedura generated lly 3D generated ge ometries 3D to geometrie the s location to of the the location building. of the The bu ST_T ilding. ranslate The ST_Transla function te fu parameters nct are ion the parameters geometry are and the geometry component of and the component displacementof the displac vector in the ement X, Y vector and Z in dir the X, ections. Y and In the Z dir SQL ections. queries, In the the SQL geom_c queries, attribute the is a geom_c point attribut geometrye is of a thepoint geometry building centr of oid, the andbuilding h is a centroid, building and terrain h is a height building attribute.terrain Refer height encing a isttribute. Refe performed renc with ing the is performed join with operation on the the join operat number of ion stor on eys the nu (one mber model of is storeys (one available for model each is stor ava ey). ilab The le r for efer each encingstorey). of the r The referenc ooms is ing similar of to the the rooms first is query, simi with lar the to the first additional quer join y, of with the r the oom addit table. ional join of the room table. If diversity diversity of the models is require requir d, ed, the second and the thir thir d approa appr ch oach can can be used. The second second approa appr ch oach represents epresents a a balance balance between between the the first first and and the the thi thir r d d approa appr ch, oach, prov pr iding oviding a a flexible flexible degr degree ee of of randomness randomness in in the the simu simulated lated dat dataset. aset. Fo Following llow the ing the second secon option, d we option, we randomly random select a ly corr select a esponding correspondin model for g each model for building. eac The h building. SQL query The that SQL qu selects ery data tha for t se thelects data for the build building_simulated ing table _ is simulated therefore table is ther modified efore to modi select a fied to random select a model random model (Appendix A.3 (Appendix A3—modif —modifications of ications Appendix of A.1Appendix in bold A1 text). in bold The text) join . The join operation operation joins jo multiple ins mult models ipl to e models each to eac building, h buildi which n arg, e which then grare then ouped grou and ped and randomly randomly ordered. or Fr dered. om From each gr each oup, gro the up, first the fir joined st joined model model is is selected selected using r usin ow g row number. number. The Th SQL e SQL query quer that y that se selects lects data data for for the th r e room_simulated oom_simulated ta table bl ise is al also so modified modified (Appendix (Appendix A4 A.4 —modifications —modifications of Appendix A2 in bold text). Instead of joining the building table, the building_simulated table is joined ISPRS Int. J. Geo-Inf. 2020, 9, 598 9 of 18 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 9 of 19 Appendix A.2 in bold text). Instead of joining the building table, the building_simulated table is joined to to the cadastral_ building table. By doing this, we get information as to which model is assigned to the the cadastral_building table. By doing this, we get information as to which model is assigned to the building. With this information, the room table can be joined. building. With this information, the room table can be joined. 3. Results 3. Results The Slovenian building cadastre classifies buildings based on their usage. In this study, we focus The Slovenian building cadastre classifies buildings based on their usage. In this study, we focus only on buildings that are classified as residential. The reason is that more complex building types only on buildings that are classified as residential. The reason is that more complex building types such as schools and hospitals require more complex functional programs. They cause the exponential such as schools and hospitals require more complex functional programs. They cause the exponential growth of the required computational time due to the combinatorial nature of the problem. growth of the required computational time due to the combinatorial nature of the problem. The buildings were generated in four runs, each for a building type, in a 2.8‐GHz Dual‐Core The buildings were generated in four runs, each for a building type, in a 2.8-GHz Dual-Core CPU CPU with 8 GB of RAM. Multi‐threading was used. In the generation of each building type, the with 8 GB of RAM. Multi-threading was used. In the generation of each building type, the duration duration of the processing time ranged from 49 min to 15 h, depending on the complexity of the of the processing time ranged from 49 min to 15 h, depending on the complexity of the building building requirements. First, the number of storeys in the Slovenian building cadastre was requirements. First, the number of storeys in the Slovenian building cadastre was investigated. The test investigated. The test showed that 94% of the residential buildings have four or fewer storeys. Based showed that 94% of the residential buildings have four or fewer storeys. Based on this, we decided to on this, we decided to use 1, 2, 4 and 8 storey building models for the case study. In total, nine use 1, 2, 4 and 8 storey building models for the case study. In total, nine algorithm runs were carried algorithm runs were carried out, producing 282 building models. Figure 10 displays a few examples out, producing 282 building models. Figure 10 displays a few examples of the generated buildings of the generated buildings (all generated buildings can be accessed from the link provided in the (all generated buildings can be accessed from the link provided in the supplementary materials). supplementary materials). Figure Figure10. Ex 10. amples of Examples the of generated the buildi generated ngs rangin buildings g from ranging the fr sing om le the ‐storey fam single-stor ily ey house family to mul house ti‐ to storey mixed multi-storey ‐use apartment mixed-use building apartment s. buildings. The Thesimulation simulation ran ranin ina acomputer computerwith w ithan ani7‐8700 i7-8700CPU, CPU,32 32GB of GB RA of M. The RAM. Postg The reSQL Postgr eSQLdatabase database was was runnin r g unning locall locally y on on th this is computer computer. . The The building_sim building_simulated ulated and r and room_simulated oom_simulated tables ta werebles were populated popula with ted the with the tra transformed n prsformed procedurally ocedurally generated gener 3D ated models, 3D models, which were wh r ich efer were enced toreferenc the ed data fr to omth t e he data from the Slovenian building cadastre for the whole country (the data can be accessed from the ISPRS Int. J. Geo-Inf. 2020, 9, 598 10 of 18 Slovenian building cadastre for the whole country (the data can be accessed from the link provided in the supplementary materials). We evaluated the performance of the simulation process and of selected SQL queries, aiming to check the data validity. We selected the second approach for referencing the generated models to real cadastral data and thus get a degree of randomness in the simulated dataset. The transformation of the generated models was performed using FME Desktop software (the processing routine can be accessed from the link provided in the supplementary materials). The models with a higher number of storeys were more demanding in terms of processing time (a single-storey took 3 min, 2-storey 4 min, 4-storey 14 min, and 8-storey 37 min). The most demanding step in the transformation process is the creation of solid geometries. The referencing of the transformed models was performed in the PostgreSQL database using the SQL queries. The building_simulated table was created using a modified A3 SQL query (Appendix A.5—modifications of Appendix A.3 in bold text). The modification adds the classification of the buildings into four classes by the number of storeys. The SQL query A4 that selects data for room_simulated data was used without modifications. The SQL query execution times for creating building_simulated and room_simulated tables took 2 min 44 s and 12 min 35 s to process 512,191 (table size of 352 MB) and 10,985,775 (9274 MB) rows, respectively. After creating the tables, an index on the geometry column was created, and the VACUUM ANALYZE operation was performed to increase the performance of spatial queries. With the data in the database, the performance of various processes and applications related to 3D cadastral data can be tested. We tested the performance of 3D queries that are designed to check the data for inconsistencies. For this, 3D functions from the PostGIS SFCGAL extension were used. SFCGAL is a C++ wrapper library that supports ISO 19107:2013, and OGC Simple Features Access 1.2 for 3D operations. The library is built around CGAL (Computational Geometry Algorithms Library), which provides algorithms for geometric computation [62]. The first SQL query checks for the intersection of 3D geometries in building_simulated table (Appendix A.6). The intersection of two building geometries represents an error, whether the geometries represent the physical or legal extent of the building. The placement of the models is performed using the centroids of real buildings. As the extents of real buildings and the procedurally generated models do not coincide, it is likely that some geometries intersect. The function that checks for the intersection is a Boolean-type function ST_3DIntersects. It has two arguments, representing two geometries that are checked for the intersection. A condition is added to exclude checking two identical geometries for the intersection (a.id_building != b.id_building). The function returns a TRUE value when parts of geometries are overlapping or if they touch each other. From the cadastral point of view, these two relationships are different, as touching of legal units is not forbidden, while overlapping is. There is no option to differentiate these two relationships without using additional 3D functions. The function uses the available spatial indexes to increase the performance. It took 2 min to find 128,812 intersections in the building_simulated table. While the first query checked for the intersection of exterior geometries, the second one focused on identifying the room geometries that intersect each other (Appendix A.7). These intersections need to be checked separately for each building as the first query has checked intersections between exterior geometries of buildings. In our case, the room geometries are used to define the extents of the real property units (see Figure 4). Overlapping room geometries thus represent an error also in 3D cadastral context, as real property units are generally not allowed to spatially intersect each other. All the adjacent room geometries in the simulated dataset are in a touching relationship, which means that the ST_3DIntersects function identifies these as intersecting ones. As stated, there is no straightforward option to differentiate between touching and overlapping geometries when using PostGIS SFCGAL functions. Therefore, in the second query, the ST_3DIntersection function was used, which constructs 3D geometries that represent the intersections. If we calculate the volume of this intersection, it is possible to differentiate between touching and overlapping relationships. However, the ST_3DIntersection function is more demanding to execute than ST_3DIntersects as it returns the ISPRS Int. J. Geo-Inf. 2020, 9, 598 11 of 18 ISPRS Int. J. Geo‐Inf. 2020, 9, x FOR PEER REVIEW 11 of 19 geometry of the intersection rather than the Boolean TRUE or FALSE. The query A7 selects the building identifier and identifiers of the pairs of rooms that overlap. The rooms are checked only against other intersection of a room with itself ( a.id_room != b.id_room). As the ST_3DIntersection function is rooms in the same building (a.id_building = b.id_building), excluding checking of the intersection of a computationally very demanding, only one building with four storeys was checked for overlap, room with itself (a.id_room != b.id_room). As the ST_3DIntersection function is computationally very which took 10 min to execute. The fact that the neighbouring room geometries are all in a touching demanding, only one building with four storeys was checked for overlap, which took 10 min to execute. relationship results in a large number of pairs of room geometries (389) to be checked with the The fact that the neighbouring room geometries are all in a touching relationship results in a large ST_3DIntersection function. number of pairs of room geometries (389) to be checked with the ST_3DIntersection function. The third error in the data that was checked was the intersection of room geometries with The third error in the data that was checked was the intersection of room geometries with building building exterior geometry. If the intersection is found, it means that the building exterior geometry exterior geometry. If the intersection is found, it means that the building exterior geometry does does not enclose all room geometries. The room geometries can be checked for intersection with the not enclose all room geometries. The room geometries can be checked for intersection with the corresponding building exterior geometry in the building_simulated table (Appendix A8). The query corresponding building exterior geometry in the building_simulated table (Appendix A.8). The query was tested on the dataset for the whole country. The query execution time was 5 h and 18 min. was tested on the dataset for the whole country. The query execution time was 5 h and 18 min. All 3D spatial queries except those containing the ST_3DIntersection function performed well All 3D spatial queries except those containing the ST_3DIntersection function performed well with the complete dataset. This function validates both input geometries, which uses the majority of with the complete dataset. This function validates both input geometries, which uses the majority the processing time. However, if the input geometries are correct, this validation is not necessary. of the processing time. However, if the input geometries are correct, this validation is not necessary. While the SFCGAL library allows disabling the validity check, the PostGIS SFCGAL implementation While the SFCGAL library allows disabling the validity check, the PostGIS SFCGAL implementation does not allow this. does not allow this. To evaluate the performance of 3D visualisation, the simulated dataset is converted to a KML To evaluate the performance of 3D visualisation, the simulated dataset is converted to a KML file for visualisation in Google Earth, and to 3D tiles for visualisation in Cesium. Both conversions file for visualisation in Google Earth, and to 3D tiles for visualisation in Cesium. Both conversions were performed using FME Desktop software. As the KML file containing all 512,191 buildings from were performed using FME Desktop software. As the KML file containing all 512,191 buildings from the building_simulated table failed to load in Google Earth (waiting time of 2 h), multiple KML files the building_simulated table failed to load in Google Earth (waiting time of 2 h), multiple KML files that varied in size to test the performance were created. The KML files were created with the feature that varied in size to test the performance were created. The KML files were created with the feature count limit of 10,000, 5000 and 1000. The performance of KML file generation and loading time are count limit of 10,000, 5000 and 1000. The performance of KML file generation and loading time are roughly proportional to the feature count: 1000, 5000 and 10,000 took 7 s, 25 s, and 54 s, respectively, roughly proportional to the feature count: 1000, 5000 and 10,000 took 7 s, 25 s, and 54 s, respectively, for file creation and 12 s, 70 s and 150 s, respectively, for loading into Google Earth. Once the files for file creation and 12 s, 70 s and 150 s, respectively, for loading into Google Earth. Once the files were were loaded, the performance of the 3D visualisation (pan, zoom, rotate) was good in all cases (Figure loaded, the performance of the 3D visualisation (pan, zoom, rotate) was good in all cases (Figure 11). 11). Figure 11. 3D visualisation of 3D visualisation of georeferenced simulated residential 3D building in Figure 11. 3D visualisation of 3D visualisation of georeferenced simulated residential 3D building in Google Earth. Google Earth. For the the visualisation visualisation in Cesium, 3D tiles tiles of buildings buildings were were prepared prepared from from the building_simulated building_simulated table table (512,191 (512,191 records) records) and rooms rooms in the room_simulate r d oom_simulated table table (10,969,100 (10,969,100 records). records). It took FM F E ME 6 min to create cr eate3D til 3D es for tiles buildings for and buildings 4h and 35 4 h min 35 for min room for r s. For ooms.testin For g purposes, testing Cesuim purposes, was run Cesuim loca was rlly un on an Apach locally on an e Tomcat Apache Tweb server omcat web using server the sa using me the PC tha same t was PC us that ed was for other used for tests. other A sim tests. p A le web simple applic web ation that application visualises that 3D visualises tiles 3D on tiles the on globe the was globe use was d. Com used. pared Compar to ed the to theperformance performance of of 3D visualisation in Google Earth, Cesium effectively handled 3D visualisation of buildings and rooms for the whole country. The proposed approach proved to be efficient for the simulation of large‐scale 3D cadastral datasets for buildings referenced to some basic characteristics of the existing cadastral data, such as ISPRS Int. J. Geo-Inf. 2020, 9, 598 12 of 18 visualisation in Google Earth, Cesium effectively handled 3D visualisation of buildings and rooms for the whole country. The proposed approach proved to be efficient for the simulation of large-scale 3D cadastral datasets for buildings referenced to some basic characteristics of the existing cadastral data, such as building location and number of storeys. The execution times of all steps needed to generate the simulated dataset are acceptable, given that we performed the simulation for the whole country, i.e., for the case of all 512,191 residential buildings in Slovenia. 4. Discussion In this study, the location of the real buildings and the number of storeys are used to make the generated dataset more representative. Although the building generation algorithm allows us to specify other aspects of real buildings (area and dimension limits, height, etc.), the algorithm was left free to generate buildings from interior spatial requirements, such as to satisfy minimum room dimensions and minimum storey areas, in the proposed approach. This led to real buildings with a large footprint being assigned simulated buildings that were too small and vice versa. Consequently, the SQL query A6 identifies the intersections of the building geometries, that do not intersect in reality. Therefore, future developments will require the generated buildings to be produced within the statistical distribution of the real data. For a perfect match scenario, the simulated buildings would have to be generated exactly according to each real building in the country. In addition, there are computational and information limitations to overcome in the future. In the presented approach, the simulation is limited to residential buildings. When the building type information in the real cadastral dataset and higher computational capacity for generation of buildings are available, the approach may be improved to include also other types of buildings, such as schools, hospitals and offices. Regarding the assignment of real property unit identifiers to rooms, various methods can be used. Using these identifiers, the 3D geometries of rooms can be aggregated to represent the extents of a real property unit. The identifier can be assigned using the storey and/or apartment number. Another option is to obtain the number of real property units from the existing cadastral data or real property registry. In this case, the assignment of identifiers to the simulated building is more challenging to simulate. One option is to assign the identifier to each room randomly. To obtain more representative real property units, the adjacency of rooms would have to be considered. This way, we would generate groups of adjacent rooms, which would get the same identifier assigned. Another option is to combine the apartment units to get the desired number of real property units. Depending on the data model that we test, the 3D geometries of adjacent rooms that belong to the same real property unit can be joined into one 3D geometry of a real property unit. This removes the internal borders of rooms and preserves only the exterior boundaries of the group of adjacent rooms. This principle, using 2D geometries of the joined rooms representing a real property unit in one storey, is used for registration of buildings in Slovenia, where for new cadastral entries, 2D outlines of real property units are registered separately for each storey. Besides the simulated buildings, the BIM datasets could also be integrated into the simulation process. In the studies that combine BIM and 3D cadastre, the IfcSpace entity is used for modelling 3D legal spaces [56,59]. The proposed method in this study uses FME Desktop software, which can be used to extract the IfcSpace entities and obtain the 3D building model. The structure of such a model is comparable to the simulated building model, but it is much closer to the real building. However, given the small amount of available BIM data that is available, the current potential of including BIM datasets is limited. Compared to the approaches of Ledoux and Meijers [36] and Biljecki et al. [37], which also use existing data on buildings for 3D modelling, the proposed method is focused on generating 3D building models from its interior arrangements. Regarding external 3D geometry of buildings, our method underperforms due to external 3D geometry being a consequence of the interior arrangement. Thus, the other two methods are demonstrated to be more representative of reality in this aspect. ISPRS Int. J. Geo-Inf. 2020, 9, 598 13 of 18 The executed queries demonstrate how the simulated 3D cadastral dataset can be used to define 3D cadastral data management strategies. Based on the executed queries, we can see that the queries containing the ST_3DIntersection function are too demanding to be used for regular checks of the whole cadastral database. Instead, these queries can be used to ensure the data validity at the moment of insertion or when some of the geometries are modified. In these cases, the new or modified geometry does not need to be checked against all geometries in the database, but only against a small subset. 5. Conclusions In this study, a methodology to generate a not yet existing 3D cadastral dataset for buildings has been presented, thus answering the paradox of “how to get a complete dataset of something that is still in the design phase”. The proposed method for simulating the 3D cadastral dataset was successfully applied for the case of Slovenia, thus proving that it can be efficiently applied to large areas. The evaluation of the database performance was illustrated with three spatial SQL queries, designed to check for the errors related to spatial relationships of 3D geometries. The adjacent geometries of the real property units in cadastre are generally in a touching relationship. The ST_3DIntersection function is needed to distinguish between the touching and overlapping relationship of two geometries, of which the first is allowed and the second represents an error. The tests showed poor performance of the queries that contain the ST_3DIntersection function. Therefore, based on the results of the performance tests of SQL queries, we can conclude that it is beneficial to have a representative dataset available in the design phase of 3D cadastral information systems. This allows us to identify problems and critical issues during the early stages of development and develop optimal DBMS storage and data management processes for cadastral data. The evaluation of 3D visualisation options shows that using 3D tiles and Cesium platform is a better option than using KML files and Google Earth, as 3D tiles are designed for visualisation of large 3D datasets. Cesium platform has therefore proved to be suitable for visualisation of large-scale 3D cadastral datasets. Further enhancements of the presented method can be performed if additional parameters and attributes are provided for the buildings and parts of buildings, such as room types, their number, and area of each room per storey. The presented method needs further investigations to include also other aspects of real buildings into the building generation process, such as footprint area or outline, volume, and height. The developed method can also be used outside of 3D cadastral domain to develop spatial analytical tools for buildings in a 3D environment, which is also an open research area in the geospatial domain. Supplementary Materials: The dataset of the 282 generated residential 3D buildings is available at the private link: https://doi.org/10.6084/m9.figshare.12283805. The dataset of building footprints from Slovenian building cadastre (Geodetska uprava Republike Slovenije, Grafični podatki katastra stavb, 11. 5. 2020) in shapefile format is available at: https://doi.org/10.6084/m9.figshare.12283760.v1. The projection is EPSG 3794. The processing routine for transforming each generated building (Section 2.2) is available at: https://doi.org/10.6084/m9.figshare.12288368.v1. The script can be run using FME desktop 2020 software. Author Contributions: Conceptualization, Jernej Tekavec and Anka Lisec; Data curation, Jernej Tekavec and Eugénio Rodrigues; Formal analysis, Jernej Tekavec; Funding acquisition, Anka Lisec; Investigation, Jernej Tekavec, Anka Lisec and Eugénio Rodrigues; Methodology, Jernej Tekavec and Eugénio Rodrigues; Project administration, Anka Lisec; Resources, Jernej Tekavec, Anka Lisec and Eugénio Rodrigues; Software, Jernej Tekavec and Eugénio Rodrigues; Supervision, Anka Lisec; Validation, Jernej Tekavec and Eugénio Rodrigues; Visualization, Jernej Tekavec and Eugénio Rodrigues; Writing—original draft, Jernej Tekavec; Writing—review and editing, Jernej Tekavec, Anka Lisec and Eugénio Rodrigues. All authors have read and agreed to the published version of the manuscript. Funding: The authors acknowledge the financial support from the Slovenian Research Agency (research core funding No. P2-0406 Earth observation and geoinformatics). Conflicts of Interest: The authors declare no conflict of interest. ISPRS Int. J. Geo-Inf. 2020, 9, 598 14 of 18 Appendix A Appendix A.1. SQL Query to Select the Data for the Table Building_Simulated SELECT id_building, id_bldg_gen, n_storeys, ST_Translate(geom3d, ST_x(ST_GeometryN(geom_c,1)), ST_y(ST_GeometryN(geom_c,1)),h) as geom3d translation of in the x ynd y direction using ST_Translate function from 0,0 to the x,y of the centroid of the real building FROM cadastral_building JOIN building ON building.n_storeys = cadastral_building.n_storeys To each building from existing cadastral data, join the simulated building with the appropriate number of storeys Appendix A.2. SQL Query to Select the Data for the Table Room_Simulated SELECT id_building, id_room_gen, ST_Translate(room.geom3d, ST_x(ST_GeometryN(geom_c,1)), ST_y(ST_GeometryN(geom_c,1)),h) as geom3d translation of in the x ynd y direction using ST_Translate function from 0,0 to the x,y of the centroid of the real building FROM cadastral_building JOIN building ON building.n_storeys = cadastral_building.n_storeys JOIN room ON room.id_bldg_gen = building.id_bldg_gen To each building from existing cadastral data, join the simulated building with the appropriate number of storeys and then join all the rooms that to this building Appendix A.3. Modified A1 SQL Query (Modifications in Bold Text) to Select the Data for the Table Building_Simulated SELECT id_building, id_bldg_gen, n_storeys, ST_Translate(geom3d, ST_x(ST_GeometryN(geom_c,1)), ST_y(ST_GeometryN(geom_c,1)),h) as geom3d FROM (SELECT *, row_number() over(PARTITION BY id_building ORDER BY random()) as rn More than one simulated buildings join to each real building. Randomly assign row number to those simulated buildings FROM cadastral_building JOIN building ON building.n_storeys = cadastral_building.n_storeys) a WHERE a.rn = 1 Take only the first of all the joined simulated buildings Appendix A.4. Modified A2 SQL Query (Modifications in Bold Text) to Select the Data for the Table Room_Simulated SELECT id_building, id_room_gen, ST_Translate(room.geom3d, ST_x(ST_GeometryN(geom_c,1)), ST_y(ST_GeometryN(geom_c,1)),h) as geom3d FROM cadastral_building JOIN building_simulated ON building_simulated.id_building = cadastral_building.id_building The simulated building is randomly selected. Instead of joining the ISPRS Int. J. Geo-Inf. 2020, 9, 598 15 of 18 building table, we join building_simulated table JOIN room ON room.id_bldg_gen = building_simulated.id_bldg_gen We can join the corresponding room geometries of the assigned simulated building Appendix A.5. Modified A3 SQL Query (Modifications in Bold Text) to Select the Data for the Table Building_Simulated SELECT id_building, id_bldg_gen, n_storeys, ST_Translate(geom3d, ST_x(ST_GeometryN(geom_c,1)), ST_y(ST_GeometryN(geom_c,1)),h) as geom3d FROM (SELECT *, row_number() over (PARTITION BY id_building ORDER BY random()) as rn FROM cadastral_building JOIN building ON CASE WHEN cadastral_building.n_storeys >= 8 THEN building.n_storeys = 8 WHEN cadastral_building.n_storeys >= 3 AND cadastral_building.n_storeys < 8 THEN building.n_storeys = 4 WHEN cadastral_building.n_storeys < 3 THEN building.n_storeys = cadastral_building.n_storeys END) a WHERE a.rn = 1 We limited the simulated buildings to 1, 2, 4 and 8 storeys. Real buildings have also other numbers of storeys. The above addition assigns the appropriate simulated building to the real building. For example, real building with 5 storeys gets 4 storey simulated building assigned Appendix A.6. SQL Query that Checks for the Intersection of 3D Geometries in Building_Simulated Table SELECT a.id_building, b.id_building FROM building_simulated a, building_simulated b WHERE ST_3DIntersects(a.geom3d, b.geom3d) The contition The output of the ST_3DIntersects function is boolean AND a.id_building != b.id_building With this condition we exclude the checking of the geometry against itself Appendix A.7. SQL Query that Checks for the Overlap of 3D Geometries in Room_Simulated Table SELECT a.id_room, b.id_room, a.id_building FROM room_simulated a, room_simulated b WHERE ST_3DIntersects(a.geom3d, b.geom3d) AND ST_Volume(ST_MakeSolid(ST_3DIntersection(a.geom3d, b.geom3d))) != 0 In addition to the ST_3DIntersects function, which is quick and removes all pairs that are not intersecting, we use ST_Intersection function and ST_Volume function to calculate the volume of the intersection. If the volume of the intersection is 0, then the geometries are in a touching relationship. AND a.id_room != b.id_room AND a.id_building = b.id_building In addition to excluding the checking of the geometry against ISPRS Int. J. Geo-Inf. 2020, 9, 598 16 of 18 itself, we use the condition to check only pairs of rooms in the same building Appendix A.8. SQL Query that Checks for the Intersection of Room Geometries with Building Exterior Geometry SELECT a.id_room, b.id_building FROM room_simulated a JOIN building_simulated b ON a.id_building = b.id_building WHERE ST_3DIntersects(a.geom3d, b.geom3d) AND a.id_building = b.id_building This query has similar structure as A6. Instead of checking the pairs of building geometries, this query checks for the intersection of building geometry with room geometries that belong to the building. References 1. Kalogianni, E.; van Oosterom, P.; Dimopoulou, E.; Lemmen, C. 3D land administration: A review and a future vision in the context of the spatial development lifecycle. ISPRS Int. J. Geo-Inf. 2020, 9, 107. [CrossRef] 2. Janečka, K.; Karki, S.; van Oosterom, P.; Zlatanova, S.; Kalantari, M.; Ghawana, T. 3D Spatial DBMS for 3D Cadastres. In 3D Cadastres Best Practices; van Oosterom, P., Ed.; International Federation of Surveyors (FIG): Copenhagen, Denmark, 2018; pp. 133–181. 3. Jaljolie, R.; van Oosterom, P.; Dalyot, S. Spatial data structure and functionalities for 3d land management system implementation: Israel case study. ISPRS Int. J. Geo-Inf. 2018, 7, 10. [CrossRef] 4. Paulsson, J. 3D Property Rights—An Analysis of Key Factors Based on International Experience. Ph.D. Thesis, Royal Institute of Technology, Stockholm, Sweden, 2007. 5. Kitsakis, D.; Dimopoulou, E. 3D cadastres: Legal approaches and necessary reforms. Surv. Rev. 2014, 46, 322–332. [CrossRef] 6. Paulsson, J.; Paasch, J. 3D property research from a legal perspective. Comput. Environ. Urban Syst. 2013, 40, 7–13. [CrossRef] 7. Stoter, J.; Ploeger, H.; van Oosterom, P. 3D cadastre in the Netherlands: Developments and international applicability. Comput. Environ. Urban Syst. 2013, 40, 56–67. [CrossRef] 8. Vučić, N.; Roić, M.; Ma ¯ der, M.; Vranić, S.; van Oosterom, P. Overview of the Croatian Land Administration System and the possibilities for its upgrade to 3D by existing data. ISPRS Int. J. Geo-Inf. 2017, 6, 223. [CrossRef] 9. Shojaei, D.; Olfat, H.; Rajabifard, A.A.; Kalantari, M. Moving Towards a Fully Operational 3D Digital Cadastre: Victoria, Australia. In Proceedings of the 6th International FIG 3D Cadastre Workshop, Delft, The Netherlands, 1–5 October 2018. 10. Janečka, K.; Souček, P. A Country Profile of the Czech Republic Based on an LADM for the Development of a 3D Cadastre. ISPRS Int. J. Geo-Inf. 2017, 6, 143. [CrossRef] 11. Alkan, M.; Gürsoy Sürmeneli, H.; Polat, Z.A. Design and development 3D RRR model for Turkish cadastral system using international standards. Surv. Rev. 2020. [CrossRef] 12. Ho, S.; Rajabifard, A.; Stoter, J.; Kalantari, M. Legal barriers to 3D cadastre implementation: What is the issue? Land Use Policy 2013, 35, 379–387. [CrossRef] 13. Stoter, J. 3D Cadastre. Ph.D. Thesis, TU Delft, Delft, The Netherlands, 2004. 14. Khuan, C.T.; Rahman, A.A.; Zlatanova, S. 3D solids and their management in DBMS. In Advances in 3D Geoinformation Systems; van Oosterom, P., Zlatanova, S., Penninga, F., Fendel, E., Eds.; Springer: Berlin/Heidelberg, Germany, 2008; pp. 279–311. [CrossRef] 15. Višnjevac, N.; Mihajlović, R.; Šoškić, M.; Cvijetinović, Ž.; Bajat, B. Prototype of the 3D cadastral system based on a NoSQL database and a Javascript visualization application. ISPRS Int. J. Geo-Inf. 2019, 8, 227. [CrossRef] 16. Zulkifli, N.A.; Rahman, A.A.; Siew, C.B. Database Design and Development of 3D Cadastral Registration based on LADM. In Proceedings of the 8th International FIG Workshop on the Land Administration Domain Model, Kuala Lumpur, Malaysia, 1–3 October 2019. ISPRS Int. J. Geo-Inf. 2020, 9, 598 17 of 18 17. Kazar, B.M.; Kothuri, R.; van Oosterom, P.; Ravada, S. On Valid and Invalid Three-Dimensional Geometries. In Advances in 3D Geoinformation Systems; van Oosterom, P., Zlatanova, S., Penninga, F., Fendel, E., Eds.; Springer: Berlin/Heidelberg, Germany, 2008; pp. 19–46. [CrossRef] 18. Karki, S.; Thompson, R.; McDougall, K. Development of validation rules to support digital lodgement of 3D cadastral plans. Comput. Environ. Urban Syst. 2013, 40, 34–45. [CrossRef] 19. Shojaei, D.; Olfat, H.; Ignacio, S.; Faundez, Q.; Kalantari, M.; Rajabifard, A.; Bri, M. Geometrical data validation in 3D digital cadastre—A case study for Victoria, Australia. Land Use Policy 2017, 68. [CrossRef] 20. Shojaei, D.; Rajabifard, A.A.; Kalantari, M.; Bishop, I.D.; Aien, A. Design and development of a web-based 3D cadastral visualisation prototype. Int. J. Digit. Earth 2015, 8, 538–557. [CrossRef] 21. Wang, C. 3D Visualization of Cadastre: Assessing the Suitability of Visual Variables and Enhancement Techniques in the 3D Model of Condominium Property Units. Ph.D. Thesis, Université Laval, Quebec City, QC, Canada, 2015. 22. Aien, A.; Rajabifard, A.; Kalantari, M.; Williamson, I. Review and Assessment of Current Cadastral Data Models for 3D Cadastral Applications. In Advances in 3D Geoinformation, Lecture Notes in Geoinformation and Cartography; Rahman, A.A., Ed.; Springer: Berlin, Germany, 2017; pp. 423–442. [CrossRef] 23. ISO 19152:2012—Geographic Information—Land Administration Domain Model (LADM); International Standard; International Organization for Standardization: Geneva Switzerland, 2012. 24. Rajabifard, A.; Atazadeh, B.; Kalantari, M. A critical evaluation of 3D spatial information models for managing legal arrangements of multi-owned developments in Victoria, Australia. Int. J. Geogr. Inf. Sci. 2018, 32, 2098–2122. [CrossRef] 25. Open Geospatial Consortium. City Geography Markup Language (CityGML) Encoding Standard, Version 2.0. 2012. Available online: http://www.opengeospatial.org/standards/citygml (accessed on 31 August 2020). 26. Open Geospatial Consortium. IndoorGML, Version 1.02. 2014. Available online: http://www.opengeospatial. org/standards/indoorgml (accessed on 31 August 2020). 27. Open Geospatial Consortium. Land and Infrastructure Conceptual Model Standard (LandInfra) Encoding Standard, Version 1.0. 2016. Available online: http://www.opengeospatial.org/standards/landinfra (accessed on 31 August 2020). 28. Aien, A.; Kalantari, M.; Rajabifard, A.; Williamson, I.; Wallace, J. Towards integration of 3D legal and physical objects in cadastral data models. Land Use Policy 2013, 35, 140–154. [CrossRef] 29. Stoter, J.; Ploeger, H.; Roes, R.; van der Riet, E.; Biljecki, F.; Ledoux, H.; Kok, D.; Kim, S. Registration of Multi-Level Property Rights in 3D in The Netherlands: Two Cases and Next Steps in Further Implementation. ISPRS Int. J. Geo-Inf. 2017, 6, 158. [CrossRef] 30. Cemellini, B.; van Oosterom, P.; Thompson, R.; de Vries, M. Design, development and usability testing of an LADM compliant 3D Cadastral prototype system. Land Use Policy 2020, 98, 104418. [CrossRef] 31. Larsson, K.; Paasch, J.M.; Paulsson, J. Representation of 3D cadastral boundaries—From analogue to digital. Land Use Policy 2020, 98, 104178. [CrossRef] 32. Remondino, F.; Georgopoulos, A.; Gonzalez-Aguilera, D.; Agrafiotis, P. (Eds.) Latest Developments in Reality-Based 3D Surveying and Modelling; MDPI: Basel, Switzerland, 2018. [CrossRef] 33. Gunduz, M.; Isikdag, U.; Basaraner, M. A review of recent research in Indoor modelling & mapping. ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2016, XLI-B4, 289–294. [CrossRef] 34. Kitsakis, D.; Dimopoulou, E. Contribution of Existing Documentation to 3D Cadastre. In Proceedings of the 4th International FIG 3D Cadastre Workshop, Dubai, United Arab Emirates, 9–11 November 2014. 35. Tekavec, J.; Lisec, A. Cadastral data as a source for 3D indoor modelling. Land Use Policy 2020, 98, 104322. [CrossRef] 36. Ledoux, H.; Meijers, M. Topologically consistent 3D city models obtained by extrusion. Int. J. Geogr. Inf. Sci. 2011, 25, 557–574. [CrossRef] 37. Biljecki, F.; Ledoux, H.; Stoter, J. Generating 3D city models without elevation data. Comput. Environ. Urban Syst. 2017, 64, 1–18. [CrossRef] 38. Watson, B.; Muller, P.; Wonka, P.; Sexton, C.; Veryovka, O.; Fuller, A. Procedural urban modeling in practice. Comput. Graph. Appl. 2018, 28, 18–26. [CrossRef] 39. Merrell, P.; Schkufza, E.; Koltun, V. Computer-generated residential building layouts. ACM Trans. Graph. 2010, 29, 181. [CrossRef] 40. Smelik, R.M.; Tutenel, T.; Bidarra, R.; Benes, B. A survey on procedural modelling for virtual worlds. Comput. Graph. Forum 2014, 33, 31–50. [CrossRef] ISPRS Int. J. Geo-Inf. 2020, 9, 598 18 of 18 41. Camozzato, D. A Method for Growth-Based Procedural Floor Plan Generation. Ph.D. Thesis, Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, Brazil, 2015. 42. Marson, F.; Musse, S.R. Automatic real-time generation of floor plans based on squarified treemaps algorithm. Int. J. Comput. Games Technol. 2010. [CrossRef] 43. Peng, C.H.; Yang, Y.L.; Wonka, P. Computing layouts with deformable templates. ACM Trans. Graph. 2014, 33, 99. [CrossRef] 44. Tutenel, T.; Bidarra, R.; Smelik, R.M.; de Kraker, K.J. Rule-based layout solving and its application to procedural interior generation. In Proceedings of the CASA Workshop on 3D Advanced Media in Gaming and Simulation, Amsterdam, The Netherlands, 17–19 June 2009. 45. Rodrigues, E.; Gaspar, A.R.; Gomes, Á. An evolutionary strategy enhanced with a local search technique for the space allocation problem in architecture, Part 1: Methodology. Comput. Aided Des. 2013, 45, 887–897. [CrossRef] 46. Gröger, G.; Plümer, L. Derivation of 3D Indoor Models by Grammars for Route Planning. Photogramm. Fernerkund. Geoinf. 2010, 191–206. [CrossRef] 47. Zhu, J.; Zhang, H.; Chen, M.; Xu, Z.; Qi, H.; Yin, L.; Wang, J.; Hu, Y. A procedural modelling method for virtual high-speed railway scenes based on model combination and spatial semantic constraint. Int. J. Geogr. Inf. Sci. 2015, 29, 1059–1080. [CrossRef] 48. Tsiliakou, E.; Labropoulos, T.; Dimopoulou, E. Procedural modelling in 3D GIS environment. Int. J. 3-D Inf. Model. 2014, 3, 17–34. [CrossRef] 49. Kitsakis, D.; Tsiliakou, E.; Labropoulos, T.; Dimopoulou, E. Procedural 3D modelling for traditional settlements. The case study of central zagori. ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2017, 42, 369–376. [CrossRef] 50. Ribeiro, A.; de Almeida, J.-P.; Ellul, C. Exploring CityEngine as a Visualisation Tool for 3D Cadastre. In Proceedings of the 4th International FIG 3D Cadastre Workshop, Dubai, United Arab Emirates, 9–11 November 2014. 51. Neuenschwander, N.; Wissen Hayek, U.; Grêt-Regamey, A. Integrating an urban green space typology into procedural 3D visualization for collaborative planning. Comput. Environ. Urban Syst. 2014, 48. [CrossRef] 52. Biljecki, F.; Ledoux, H.; Stoter, J. Generation of multi-LOD 3D city models in City GML with the procedural modelling engine Random3Dcity. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2016, IV-4/W1, 51–59. [CrossRef] 53. Kumar, K.; Labetski, A.; Ledoux, H.; Stoter, J. An improved LOD framework for the terrains in 3D city models. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2019, IV-4/W8, 75–82. [CrossRef] 54. Rodrigues, E.; Fernandes, M.S.; Gomes, Á.; Gaspar, A.R.; Costa, J.J. Performance-based design of multi-story buildings for a sustainable urban environment: A case study. Renew. Sustain. Energy Rev. 2019, 113. [CrossRef] 55. Drobež, P.; Kosmatin Fras, M.; Ferlan, M.; Lisec, A. Transition from 2D to 3D real property cadastre—The case of the Slovenian cadastre. Comput. Environ. Urban Syst. 2017, 62, 125–135. [CrossRef] 56. Atazadeh, B.; Kalantari, M.; Rajabifard, A.; Ho, S.; Ngo, T. Building Information Modelling for High-rise Land Administration. Trans. GIS 2017, 21, 91–113. [CrossRef] 57. Li, L.; Wu, J.; Zhu, H.; Duan, X.; Luo, F. 3D modeling of the ownership structure of condominium units. Comput. Environ. Urban Syst. 2016, 59, 50–63. [CrossRef] 58. Knoth, L.; Scholz, J.; Strobl, J.; Mittlböck, M.; Vockner, B.; Atzl, C.; Rajabifard, A.; Atazadeh, B. Cross-Domain Building Models—A Step towards Interoperability. ISPRS Int. J. Geo-Inf. 2018, 7, 363. [CrossRef] 59. Oldfield, J.; van Oosterom, P.; Beetz, J.; Krijnen, T.F. Working with open BIM standards to source legal spaces for a 3D cadastre. ISPRS Int. J. Geo-Inf. 2017, 6, 351. [CrossRef] 60. Barzegar, M.; Rajabifard, A.; Kalantari, M.; Atazadeh, B. 3D BIM-enabled spatial query for retrieving property boundaries: A case study in Victoria, Australia. Int. J. Geogr. Inf. Sci. 2020, 34, 251–271. [CrossRef] 61. Rajabifard, A.; Atazadeh, B.; Kalantari, M. BIM and Urban Land Administration, 1st ed.; CRC Press: Boca Raton, FL, USA, 2019. [CrossRef] 62. SFGGAL, Wrapper around CGAL Library for 2D and 3D Operations Based on OGC Standards. 2019. Available online: http://www.sfcgal.org (accessed on 31 August 2020). © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). PRILOGA E 3D Geometry-Based Indoor Network Extraction for Navigation Applications Using SFCGAL Tekavec, J., Lisec, A. (2020b) ISPRS International Journal of Geo-Information, 9(7), 417 DOI: 10.3390/ijgi9070417 Faktor vpliva za leto 2019: 2,239 (Q3) International Journal of Geo-Information Article 3D Geometry-Based Indoor Network Extraction for Navigation Applications Using SFCGAL Jernej Tekavec * and Anka Lisec University of Ljubljana, Faculty of Civil and Geodetic Engineering, Department of Geodesy, SI-1000 Ljubljana, Slovenia; anka.lisec@fgg.uni-lj.si * Correspondence: jernej.tekavec@fgg.uni-lj.si; Tel.: +386-1-4768-561 Received: 29 May 2020; Accepted: 27 June 2020; Published: 29 June 2020 Abstract: This study is focused on indoor navigation network extraction for navigation applications based on available 3D building data and using SFCGAL library, e.g. simple features computational geometry algorithms library. In this study, special attention is given to 3D cadastre and BIM (building information modelling) datasets, which have been used as data sources for 3D geometric indoor modelling. SFCGAL 3D functions are used for the extraction of an indoor network, which has been modelled in the form of indoor connectivity graphs based on 3D geometries of indoor features. The extraction is performed by the integration of extract transform load (ETL) software and the spatial database to support multiple data sources and provide access to SFCGAL functions. With this integrated approach, the current lack of straightforward software support for complex 3D spatial analyses is addressed. Based on the developed methodology, we perform and discuss the extraction of an indoor navigation network from 3D cadastral and BIM data. The efficiency and performance of the network analyses were evaluated using the processing and query execution times. The results show that the proposed methodology for geometry-based navigation network extraction of buildings is efficient and can be used with various types of 3D geometric indoor data. Keywords: SFCGAL; DBMS; ETL; indoor navigation; topology; spatial query; 3D cadastre; BIM 1. Introduction The 3D geospatial data domain with its derived 3D geospatial information and applications is experiencing rapid development in all its aspects, including 3D spatial data acquisition, spatial data modelling and processing, spatial analysis and visualization [1–4]. 3D geospatial data and 3D geoinformation applications have gained momentum in the past decade as a rapidly developing geospatial industry with new sensors and platforms, as well as the rapid progress of information and communication technology, opened up possibilities for acquiring, collecting, modelling and analysing big geospatial datasets. However, 3D spatial data management and analyses together with all aspects of indoor spatial data had been lagging behind until recently [5]. The spatial data queries and analyses for decision making and indoor applications require advanced data models with high-quality geometry and topology. While many publications deal with geometrical and topological approaches to 3D geospatial modelling [6–8], integrating and managing 3D geospatial data on both the geometrical and semantic levels remains a big challenge and is the focus of several research activities [9–11]. The importance of 3D spatial data integration and management lies in the fact that many spatial analyses that produce valuable information use data from various sources, which need to be harmonized and integrated. The current software support for 3D spatial data analysis is inadequate, as most algorithms support only 2D spatial data processing and analyses. The demand for and importance of indoor spatial information are rapidly growing, creating a need to integrate the data from different domains [2,12,13]. The extraction and modelling of data and information relevant for indoor navigation from various ISPRS Int. J. Geo-Inf. 2020, 9, 417; doi:10.3390/ijgi9070417 www.mdpi.com/journal/ijgi ISPRS Int. J. Geo-Inf. 2020, 9, 417 2 of 15 data sources has been intensively studied in recent years [14–18]. The main aim of this study is to develop a methodology to perform the extraction of an indoor navigation network based solely on 3D geometry using SFCGAL functions, and to discuss the issues and advantages in using this approach. The methodology is applied to two different indoor spatial data sources, i. e. 3D cadastre data and BIM data, which are well-known 3D geometric and attribute data sources for building interiors. In the first part of the paper, the integration of spatial extract transform load (ETL) software [19] and the spatial database used to develop the methodology for geometry-based indoor navigation network extraction are presented. It is implemented using FME Desktop (FME) and PostgreSQL with PostGIS and SFCGAL [20] extensions. Assuming the ETL software supports the handling of 3D geometry, it can be effectively used for 3D geospatial data decomposition, integration, and data extraction, which is an important part of complex 3D spatial modelling and analyses. FME has strong support for 3D geospatial data manipulation and recently adopted standard data file formats such as CityGML [21], IndoorGML [22] and IFC (industry foundation classes) [23], together with other 3D data file formats (OBJ, COLLADA, STL, FBX, 3DS, and X3D). However, FME has some, though very limited, functionalities for processing 3D geometries. For this, the PostgreSQL database was selected as it provides access to SFCGAL 3D functions that enable processing of 3D geometries. The SFCGAL is a C++ wrapper library that supports ISO 19107:2013, and OGC (Open Geospatial Consortium) Simple Features Access 1.2 for 3D operations. The library is built around CGAL (Computational Geometry Algorithms Library), which provides algorithms for geometric computation. Besides the common spatial feature types, the PostGIS and SFCGAL enabled database supports storage of TINs (triangular irregular networks), polyhedral surfaces and solids, and offers functions to perform 3D spatial operations using the stored 3D geometries. As ETL software is designed to work with databases, it is common for ETL processes to include the databases. Our approach is different, as it uses the database for the access to 3D functions needed for extraction of an indoor navigation network based on 3D geometries. The 3D cadastral case study aims to strengthen the idea of a multipurpose 3D cadastre. It builds on the 3D cadastral data model discussed in [24] with the difference that the solid geometries touch only at passages (e.g., doors). 3D geospatial data are to be used within a 3D cadastre to efficiently model the complex divisions of real property units in a 3D environment to which rights, restrictions and responsibilities are related [25]. Indoor navigation applications have generally been beyond the scope of traditional land administration, i.e., a cadastre. However, developments in 3D spatial data acquisition, management and visualization, especially in the last decade, have brought these two fields much closer [26]. Using cadastral data has some advantages over other data sources as the data is official, follows a predefined data model, and its quality is (or should be) controlled. The biggest current disadvantage is that the digital 3D cadastral systems are not yet introduced operationally at a larger scale. The BIM data meeting IFC standards (ISO 16739-1:2018) represent a rich data source for many building-related applications, including indoor navigation. Diakité and Zlatanova [12] and Lin et al. [27] have already considered IFC data as a data source for indoor navigation applications. Xu et al. [28] and Lin et al. [17] propose reducing the problem to 2D space and rasterization of vector geometry. Xiong et al. [14] used voxelization to avoid complex vector processing tasks. A voxel model was also used by Staats et al. [29]. Teo and Cho [30] and Khan et al. [31] use semantic information (IfcRelSpaceBoundary) in IFC files for navigation network extraction. In contrast, our study is focused on the extraction of a navigation network based solely on the 3D geometry of IfcSpace, IfcDoor and IfcStair entities. Our approach is comparable to the approaches that propose the semantic-based extraction of a navigation network from an IFC file [30,31]. 2. Methodology The study focuses on the processing of 3D geometric data for extraction of an indoor navigation network. The 3D geometries that are used in this context are navigable 3D indoor spaces. To perform ISPRS Int. J. Geo-Inf. 2020, 9, 417 3 of 15 the 3D operations and to derive the navigation network from input data, we integrate the FME spatial ETL softwar ISPRS Int. J. e Geo- and Inf. Postgr 2020, 9, eSQL 417 database with PostGIS and SFCGAL extensions (Figure 1). 3 of 15 Figure 1. The concept of ETL-based data integration and 3D analytics. Figure 1. The concept of ETL-based data integration and 3D analytics. Before the data is transferred into the database, it should be pre-processed. The aim is to extract Before the data is transferred into the database, it should be pre-processed. The aim is to extract 3D geometries and attributes from input data and transform them into a form that can be imported 3D geometries and attributes from input data and transform them into a form that can be imported into the PostGIS database and analysed with SFCGAL functions. Firstly, all the relevant entities that into the PostGIS database and analysed with SFCGAL functions. Firstly, all the relevant entities that represent 3D geometries of indoor features have to be extracted from input data. Some input datasets represent 3D geometries of indoor features have to be extracted from input data. Some input datasets (for instance IFC files) have to be decomposed into separate entities first. The decomposition and (for instance IFC files) have to be decomposed into separate entities first. The decomposition and extraction of relevant data can be performed during the import of the dataset into FME. After that, extraction of relevant data can be performed during the import of the dataset into FME. After that, various attribute or geometry filters can also be applied to select the relevant data for 3D processing various attribute or geometry filters can also be applied to select the relevant data for 3D processing with SFCGAL functions. In this phase, depending on the data source, the geometries can be validated with SFCGAL functions. In this phase, depending on the data source, the geometries can be validated using FME tools and, to some degree, corrected for errors. The data storage usually terminates the using FME tools and, to some degree, corrected for errors. The data storage usually terminates the ETL process. In this case, the data writing into the PostGIS database is performed during the process, ETL process. In this case, the data writing into the PostGIS database is performed during the process, allowing it to be continued after the data is loaded. The greatest attention in this phase should be given allowing it to be continued after the data is loaded. The greatest attention in this phase should be to the geometry, as only PostGIS-supported geometry types can be inserted into the database. In this given to the geometry, as only PostGIS-supported geometry types can be inserted into the database. paper, the focus is on 3D solid geometries that are bounded by polyhedral surfaces. The bounding In this paper, the focus is on 3D solid geometries that are bounded by polyhedral surfaces. The surfaces of the geometries can be obtained in FME by deaggregation of input geometries. When the bounding surfaces of the geometries can be obtained in FME by deaggregation of input geometries. input geometries are deaggregated (split into parts), it is important to preserve their unique attributes. When the input geometries are deaggregated (split into parts), it is important to preserve their unique These attributes are used to group the surfaces that bound each solid geometry when creating solid attributes. These attributes are used to group the surfaces that bound each solid geometry when geometries. Due to different tolerances and definitions of solid geometries in FME and PostGIS, some of creating solid geometries. Due to different tolerances and definitions of solid geometries in FME and the geometries can be reported as invalid in PostGIS. Figure 2 shows an example of a valid 3D solid in PostGIS, some of the geometries can be reported as invalid in PostGIS. Figure 2 shows an example of FME (no solid validation errors reported) (a) before and (b) after insertion into PostGIS database. In this a valid 3D solid in FME (no solid validation errors reported) (a) before and (b) after insertion into case, the bounding surfaces in PostGIS are not the same as in FME. This happens with the surfaces PostGIS database. In this case, the bounding surfaces in PostGIS are not the same as in FME. This that have internal holes. Such geometries are reported by PostGIS as invalid. PostGIS also reports the happens with the surfaces that have internal holes. Such geometries are reported by PostGIS as validity error, when the points of the same surface do not lie on the same plane. Such geometries do invalid. PostGIS also reports the validity error, when the points of the same surface do not lie on the not allow the use of SFCGAL 3D functions in the PostGIS database. The generic solution for this issue same plane. Such geometries do not allow the use of SFCGAL 3D functions in the PostGIS database. is to perform the triangulation of all surfaces in FME before creating solid geometries. This additional The generic solution for this issue is to perform the triangulation of all surfaces in FME before creating step makes all bounding surfaces of solids triangles. With triangles, we avoid the validity issues as solid geometries. This additional step makes all bounding surfaces of solids triangles. With triangles, they do not have internal holes and also cannot have points not lying on the same plane. we avoid the validity issues as they do not have internal holes and also cannot have points not lying on the same plane. It should be noted that the triangulation changes the structure of the bounding surfaces, but this is not an issue for the presented application for geometry-based navigation network extraction. The geometries should be imported into the database together with identifiers to enable referencing the results of 3D operations in FME for later processing and writing of the results. ISPRS Int. J. Geo-Inf. 2020, 9, 417 4 of 15 ISPRS Int. J. Geo-Inf. 2020, 9, 417 4 of 15 Figure 2. (a) Solid geometry in FME and (b) polyhedral surface inserted into PostGIS. Figure 2. (a) Solid geometry in FME and (b) polyhedral surface inserted into PostGIS. It should be noted that the triangulation changes the structure of the bounding surfaces, but this When the geometries ( geom), identifiers ( sn) and optional additional attributes are imported into is not an issue for the presented application for geometry-based navigation network extraction. the database (temporary table spaces), the SQL queries using SFCGAL 3D functions can be executed The geometries should be imported into the database together with identifiers to enable referencing to derive the topological relationships of the inserted geometries. Generally, the navigable spaces that the results of 3D operations in FME for later processing and writing of the results. are in a touching relationship allow the transition from one to another. Some data models also allow When the geometries (geom), identifiers (sn) and optional additional attributes are imported into the overlapping of navigable spaces that allow the transition. Here, we presume that the navigable the database (temporary table spaces), the SQL queries using SFCGAL 3D functions can be executed to spaces that are spatially disjoint do not allow the transition from one to another. Depending on the derive the topological relationships of the inserted geometries. Generally, the navigable spaces that are required complexity of the network, various methods can be used. in a touching relationship allow the transition from one to another. Some data models also allow the The first and most basic method uses the FME centroid placement tool and the SFCGAL overlapping of navigable spaces that allow the transition. Here, we presume that the navigable spaces ST_3DIntersects Boolean function to identify the pairs of geometries that intersect. The function that are spatially disjoint do not allow the transition from one to another. Depending on the required returns TRUE for both overlapping and touching geometries. Using the query (Appendix A1), we complexity of the network, various methods can be used. select two symmetrical pairs of identifiers (as start and end) per each connection of spaces that are The first and most basic method uses the FME centroid placement tool and the SFCGAL found. If we want only one of the two pairs, the a.sn! = b.sn condition should be changed to a.sn > b.sn. ST_3DIntersects Boolean function to identify the pairs of geometries that intersect. The function returns The query results in the identifiers of the connected spaces. Parallel to query execution in the TRUE for both overlapping and touching geometries. Using the query (Appendix A.1), we select two database, the 3D solid geometries are transformed to graph nodes using the FME centroid placement symmetrical pairs of identifiers (as start and end) per each connection of spaces that are found. If we tool. The algorithm has an option to force the placement of the centroid inside the polygon, thus want only one of the two pairs, the a.sn! = b.sn condition should be changed to a.sn > b.sn. The query solving the issue of the external centroid for complex polygons (concave and doughnut polygons). results in the identifiers of the connected spaces. Parallel to query execution in the database, the 3D As only 2D polygons are supported, each 3D solid geometry is converted to a 2D polygon solid geometries are transformed to graph nodes using the FME centroid placement tool. The algorithm representing an outline of each solid in the horizontal plane. Each node is vertically placed in the has an option to force the placement of the centroid inside the polygon, thus solving the issue of the middle between the vertical bounds of the corresponding solid. The final navigation network is external centroid for complex polygons (concave and doughnut polygons). As only 2D polygons are realized by joining the nodes to the result of the SQL query and creating the lines between connected supported, each 3D solid geometry is converted to a 2D polygon representing an outline of each solid nodes. The first approach results in a basic navigation network, also known as node-relation structure in the horizontal plane. Each node is vertically placed in the middle between the vertical bounds of (NRS) [32], with connections that do not represent the path between the nodes (Figure 3). the corresponding solid. The final navigation network is realized by joining the nodes to the result Consequently, such a network can be used for basic navigation applications without an option for of the SQL query and creating the lines between connected nodes. The first approach results in a visualization of the path. Boguslawski et al. [33] and Liu and Zlatanova [34] use this type of network basic navigation network, also known as node-relation structure (NRS) [32], with connections that do as the basis for derivation of more complex navigable networks. not represent the path between the nodes (Figure 3). Consequently, such a network can be used for basic navigation applications without an option for visualization of the path. Boguslawski et al. [33] and Liu and Zlatanova [34] use this type of network as the basis for derivation of more complex navigable networks. The second method provides enrichment of the original NRS to node-relation structure and entrance (NRSE) [32]. To perform this using SFCGAL, the ST_3DIntersection function is introduced, which outputs the geometry of the intersection (Appendix A.2). Using the intersection geometries and FME centroid placing tool, the additional nodes, placed at space connections, can be derived and added to the navigation network (Figure 4). Figure 3. Basic indoor navigation network (NRS) with red nodes and green connections. ISPRS Int. J. Geo-Inf. 2020, 9, 417 4 of 15 Figure 2. (a) Solid geometry in FME and (b) polyhedral surface inserted into PostGIS. When the geometries ( geom), identifiers ( sn) and optional additional attributes are imported into the database (temporary table spaces), the SQL queries using SFCGAL 3D functions can be executed to derive the topological relationships of the inserted geometries. Generally, the navigable spaces that are in a touching relationship allow the transition from one to another. Some data models also allow the overlapping of navigable spaces that allow the transition. Here, we presume that the navigable spaces that are spatially disjoint do not allow the transition from one to another. Depending on the required complexity of the network, various methods can be used. The first and most basic method uses the FME centroid placement tool and the SFCGAL ST_3DIntersects Boolean function to identify the pairs of geometries that intersect. The function returns TRUE for both overlapping and touching geometries. Using the query (Appendix A1), we select two symmetrical pairs of identifiers (as start and end) per each connection of spaces that are found. If we want only one of the two pairs, the a.sn! = b.sn condition should be changed to a.sn > b.sn. The query results in the identifiers of the connected spaces. Parallel to query execution in the database, the 3D solid geometries are transformed to graph nodes using the FME centroid placement tool. The algorithm has an option to force the placement of the centroid inside the polygon, thus solving the issue of the external centroid for complex polygons (concave and doughnut polygons). As only 2D polygons are supported, each 3D solid geometry is converted to a 2D polygon representing an outline of each solid in the horizontal plane. Each node is vertically placed in the middle between the vertical bounds of the corresponding solid. The final navigation network is realized by joining the nodes to the result of the SQL query and creating the lines between connected nodes. The first approach results in a basic navigation network, also known as node-relation structure (NRS) [32], with connections that do not represent the path between the nodes (Figure 3). Consequently, such a network can be used for basic navigation applications without an option for ISPRS visu Int. ali J. zat Geo-Inf. ion 2020 of the , 9 p , a 417 th. Boguslawski et al. [33] and Liu and Zlatanova [34] use this type of net5 of w 15 ork as the basis for derivation of more complex navigable networks. ISPRS Int. J. Geo-Inf. 2020, 9, 417 5 of 15 ISPRS Int. J. Geo-Inf. 2020, 9, 417 5 of 15 The second method provides enrichment of the original NRS to node-relation structure and The second method provides enrichment of the original NRS to node-relation structure and entrance (NRSE) [32]. To perform this using SFCGAL, the ST_3DIntersection function is introduced, entrance (NRSE) [32]. To perform this using SFCGAL, the ST_3DIntersection function is introduced, which outputs the geometry of the intersection (Appendix A2). Using the intersection geometries and which outputs the geometry of the intersection (Appendix A2). Using the intersecti FME centroid placing tool, the additional nodes, placed at space connections, can on geometries and be derived and FME ce added Figure to the 3. navig Basic ntroid placing a indoor tool tion ne navigation , the addition twork (Figurenetwork 4). (NRS) al nodes with , placed red at nodes space and green connections. connections, can be derived and added Figure 3. to the navigBasic a indoor navigat tion network (F i ig on network (NRS) ure 4). with red nodes and green connections. Figure 4. Figure 4. Node-r Node elation struct -relation struct ure and entrance (NRSE) ure and entrance (NRSE) navigation network with red navigation network with red node nodes and s and green Figure 4. Node-relation structure and entrance (NRSE) navigation network with red nodes and green green connections. connections. connections. The method works for spaces that have only one connection to each connected space (Figure 4). The method works for spaces that have only one connection to each connected space (Figure 4). The method Figure 5a shows works for an sp example aces th of twoat have spaces only that one con are nection to connected eac with h connected multiple space passages. (Figure Ther 4). efore, Figure 5a shows an example of two spaces that are connected with multiple passages. Therefore, the Fig theure 5a sho method ws an for exam NRSE ple of tw derivation o sp hasaces t to h be at are connected with modified to multiple passage accommodate such s cases. . Ther In efo thesere, the cases, method for NRSE derivation has to be modified to accommodate such cases. In these cases, the method for the NRSE der ST_3DIntersection ivation has to b function returnse modified t multiple o accommod intersection ate such c geometries a as ses. In these c one feature. ases, the The FME ST_3DIntersection function returns multiple intersection geometries as one feature. The FME ST_3DI deaggr ntersec egator tion tool funct was ion used ret to urns split mu the ltiple int spatially ersec disjointtion geomet intersection ries as one geometries feat and ure. thus The place FME a node deaggregator tool was used to split the spatially disjoint intersection geometries and thus place a dea to ggrega each tor tool wa connection s used to sp (Figure 5b). lit the spatially disjoint intersection geometries and thus place a node to each connection (Figure 5b). node to each connection (Figure 5b). Figure 5. Basic NRSE derivation method (a) and the modified method for spaces with multiple Figure 5. connections B ( a b si ). c NRSE derivation method (a) and the modified method for spaces with multiple Figure 5. Basic NRSE derivation method (a) and the modified method for spaces with multiple connections (b). connections (b). As shown in Figures 3 and 4, the complexity of indoor spaces requires more advanced approaches that As shown increase in the Fig nu ure mber s 3 of and 4, nodes t and he complexit optimize y o their f indoor sp placement. aces re Many quires m studies pr ore adv oposed anced various As shown in Figures 3 and 4, the complexity of indoor spaces requires more advanced approache methods s to that incre derive a the se the number of navigation nodes network for and optimize the optimal path ir placement. planning. Most M of any stud them use ies proposed methods that approaches that increase the number of nodes and optimize their placement. Many studies proposed various methods to deri include medial axis ve the navi transformation ga [ tion network 30,32,35], str for opti ucture mal path pla segmentation nni using ng. Most of them use triangulated irregular various methods to derive the navigation network for optimal path planning. Most of them use methods t networks hat include m (TINs) [18, 33]. edi The al axis t SFCGAL ransform function ation ST_Appr [30,32,35], struct oximateMedialAxis isure s used egm in ent the at thir ion d usin pr g oposed methods that include medial axis transformation [30,32,35], structure segmentation using triangulated irregular networks (TINs) [18,33]. The SFCGAL function ST_ApproximateMedialAxis is triangulated irregular networks (TINs) [18,33]. The SFCGAL function ST_ApproximateMedialAxis is used in the third proposed method that densifies the navigation network inside each space. As the used in the third proposed method that densifies the navigation network inside each space. As the function works only on 2D geometries, the solid geometries are transformed to their 2D outlines in function works only on 2D geometries, the solid geometries are transformed to their 2D outlines in FME and inserted into the PostGIS database (temporary table spaces2d) as polygons ( geom) together FME and inserted into the PostGIS database (temporary table spaces2d) as polygons ( geom) together with the identifiers of spaces ( sn). There, the approximate medial axis is calculated for each polygon, with the identifiers of spaces ( sn). There, the approximate medial axis is calculated for each polygon, obtaining the network edges (Appendix A3). The ST_Dump function is used to derive the nodes of obtaining the network edges (Appendix A3). The ST_Dump function is used to derive the nodes of the calculated edges (Appendix A4). At this stage, the calculated networks are not connected. The the calculated edges (Appendix A4). At this stage, the calculated networks are not connected. The connection nodes of spaces are derived according to the second method and inserted into the PostGIS connection nodes of spaces are derived according to the second method and inserted into the PostGIS ISPRS Int. J. Geo-Inf. 2020, 9, 417 6 of 15 method that densifies the navigation network inside each space. As the function works only on 2D geometries, the solid geometries are transformed to their 2D outlines in FME and inserted into the PostGIS database (temporary table spaces2d) as polygons (geom) together with the identifiers of spaces (sn). There, the approximate medial axis is calculated for each polygon, obtaining the network edges (Appendix A.3). The ST_Dump function is used to derive the nodes of the calculated edges (Appendix A.4). At this stage, the calculated networks are not connected. The connection nodes of ISPRS Int. J. Geo-Inf. 2020, 9, 417 6 of 15 spaces are derived according to the second method and inserted into the PostGIS database (temporary table connections) with unique identifier (id) geometry (geom) and identifiers of both connected spaces database (temporary table connections) with unique identifier ( id) geometry ( geom) and identifiers of (start, end). After joining the calculated edges (the result of the Appendix A.3 query) to the connections both connected spaces ( start, end). After joining the calculated edges (the result of the Appendix A3 table, the nearest point on edge can be found using ST_ClosestPoint function that is included in standard query) to the connections table, the nearest point on edge can be found using ST_ClosestPoint function PostGIS functions (Appendix A.5). This enables the connection of individual navigation networks into that is included in standard PostGIS functions (Appendix A5). This enables the connection of the final navigation network (Figure 6). individual navigation networks into the final navigation network (Figure 6). Figure 6. The result of the third proposed method. Figure 6. The result of the third proposed method. 3. Results 3. Results 3.1. Navigation Network from 3D Cadastral Data 3.1. Navi For gati the on Ne 3D twork fr cadastral om 3 case D Cadastr study, a the l Data input data is focused on buildings and can be characterized as a 3D For the 3D version of cadastral c traditional ase 2D study, floor the inpu plans used t dat for a is focused on bu condominium r ildings egistration. and c The an be scope char of act the erized selected as a 3D v dataset ersi used on of t in the radit case io stnal 2D f udy is loor p limitedlans to used fo cadastralr condom data, iniu which m is registra focused tion on . The scop the repr e of the esentation select of ed the dat legal aset used situation in t (e.g., hre cas eal e pr stud opertyy is lim units) ited using to cada indoor stral dat spaces a as , which a core is fo spatial cused o unit, n the defined rep by resent the ation of t physical strhe leg uctur a e l sit of uat the ion (e.g., re building. al p Refer roperty encing unit the s) us legal ing indoor sp situation to aces the as a core building’s spatial physical unit str , define ucture d for by the physical st condominium or ructure o strata f title trhe building. egistration Refe using rencin 2D g t floor he leg plans al is situation t known in o the buil many c ding’s ountries physical struc worldwide. ture for c The 3D ondominium o indoor spaces, r strata ti modelled tle reg with istrat solid ion using 2D geometries, floor p repr lan esent s is known the core ofin many the test countries wo data model t rldwide. The hat has also 3D beenindoor spaces, mode discussed in [24]. ll Ased with the solid focus of tgeometries, r he case e study present the was on core o testing f the the test data performancemodel that h of the as also been developed discus methods, sed four in [24]. A generic 3D s the focu cadastral s of the c datasets ase s wer t e udy w used, as on testing consisting of the performance of the de one, three, six and nine veloped meth storeys with theods, four same generic horizontal 3D cadastral d layout and a one tasets were used, co vertical connection nsistin per g floor of one, three (Figure 7). , six The and nine test storeys w datasets were ith the same modelled using horizontal layo SketchUp 3D ut and one vert modelling ical connectio software. The n per connections floor (Fig between ure 7) indoor . The t spaces est dat (e.g., asets we doors) arere modelled materialized usin by g Sk shar et ed chUp 3D mod surfaces e which lling soft enable ware. The extraction of connections the between indo topology information or spaces using the (e.g., doors) ar SFCGAL 3D e materialized geometry by sh functions. ared The surfac real pr es which operty enable units can extra be ction of derived t byhe topol aggr ogy inf egation oforma the tion usi corr ng the SF esponding CGAL 3D spaces, i.e., geo sp m atialetry fun units. c It tio is ns. The real p possible to roperty apply the unit pr s can be oposed derived by methods to a any ggreg other ation o 3D f the correspo cadastral dataset nding sp if the ace data s, i.e., sp format is atial units. It supported is by pos the sible t FME o and appl if it y the proposed methods to follows the same data any other modelling 3D ca principle,dastral da having ta 3D set if the da geometries ta of format is indoor supported by the spaces in touching FME r and if it elationships. follow The s th appr e sa oach me for data modell modelling ing such principle, h datasets fr avin om g 3D geomet existing 2D ries of indoo documentation,r spaces which in can touching r be found e in lationships many . The appro countries ach fo worldwide, r model has ling been pr such da oposed ta in sets from exi [24]. sting 2D documentation, which can be found in many countries worldwide, has been proposed in [24]. Figure 7. The 3D cadastral datasets. ISPRS Int. J. Geo-Inf. 2020, 9, 417 6 of 15 database (temporary table connections) with unique identifier ( id) geometry ( geom) and identifiers of both connected spaces ( start, end). After joining the calculated edges (the result of the Appendix A3 query) to the connections table, the nearest point on edge can be found using ST_ClosestPoint function that is included in standard PostGIS functions (Appendix A5). This enables the connection of individual navigation networks into the final navigation network (Figure 6). Figure 6. The result of the third proposed method. 3. Results 3.1. Navigation Network from 3D Cadastral Data For the 3D cadastral case study, the input data is focused on buildings and can be characterized as a 3D version of traditional 2D floor plans used for condominium registration. The scope of the selected dataset used in the case study is limited to cadastral data, which is focused on the representation of the legal situation (e.g., real property units) using indoor spaces as a core spatial unit, defined by the physical structure of the building. Referencing the legal situation to the building’s physical structure for condominium or strata title registration using 2D floor plans is known in many countries worldwide. The 3D indoor spaces, modelled with solid geometries, represent the core of the test data model that has also been discussed in [24]. As the focus of the case study was on testing the performance of the developed methods, four generic 3D cadastral datasets were used, consisting of one, three, six and nine storeys with the same horizontal layout and one vertical connection per floor (Figure 7). The test datasets were modelled using SketchUp 3D modelling software. The connections between indoor spaces (e.g., doors) are materialized by shared surfaces which enable extraction of the topology information using the SFCGAL 3D geometry functions. The real property units can be derived by aggregation of the corresponding spaces, i.e., spatial units. It is possible to apply the proposed methods to any other 3D cadastral dataset if the data format is supported by the FME and if it follows the same data modelling principle, having 3D geometries of indoor spaces in ISPRS to Int. J. Geo-Inf. uching relati 2020, 9, onships 417 . The approach for modelling such datasets from existing 2D document 7 of at 15 ion, which can be found in many countries worldwide, has been proposed in [24]. Figure 7. The 3D cadastral datasets. Figure 7. The 3D cadastral datasets. All the test datasets were processed with all three presented methods in order to assess the performance of each method and the SQL queries included in each method. The processing times are used as results. While they enable relative comparisons, they depend heavily on the hardware that is used for processing. All the tests were performed on a PC equipped with i7-8565U CPU, 16 GB RAM and an SSD hard drive. Table 1 summarizes the processing results for the first and second method. Table 2 summarizes the results for the third method. For each method, a complete processing time is given, including all processes in FME and PostGIS. Additionally, the query execution times are given for each query that is included in the method. The processing times for Query Appendix A.2 in Table 2 are repeated from Table 1, as both the query and the data are identical for both methods Appendix A. Table 1. Results of processing of the 3D cadastral datasets for the first and the second method. Dataset No. of Spaces Method 1 Query Appendix A.1 Method 2 Query Appendix A.2 1 storey 31 29 s 0.08 s 1 min 10 s 42 s 3 storeys 93 59 s 1.4 s 3 min 8 s 2 min 25 s 6 storeys 186 1 min 46 s 7.1 s 6 min 39 s 4 min 28 s 9 storeys 279 2 min 58 s 16.2 s 10 min 16 s 6 min 30 s Table 2. Results of processing of the 3D cadastral datasets for the third method. No. of Query Query Query Query Dataset Method 3 Spaces Appendix A.2 Appendix A.3 Appendix A.4 Appendix A.5 1 storey 31 1 min 11 s 42 s 0.02 s 0.02 s 0.04 s 3 storeys 93 3 min 13 s 2 min 25 s 0.04 s 0.06 s 0.1 s 6 storeys 186 6 min 40 s 4 min 28 s 0.08 s 0.1 s 0.2 s 9 storeys 279 10 min 20 s 6 min 30 s 0.1 s 0.2 s 0.3 s 3.2. Navigation Network from IFC Data The second case study is based on the example project IFC file from BIMcollab [36] for the office type building (Figure 8). Since many versions of IFC models exist, it should be emphasized that our method can be successfully applied only to the models that contain the IfcSpace, IfcDoor and IfcStair entities. The input IFC file was imported and decomposed in FME, allowing the selection of individual IFC entities. First, we limited the selection of entities to IfcSpace, IfcDoor. We filtered the geometries by the type of geometry, where only solid geometries were passed through. Another filter passes only door solids with Name attribute value “Box” and space solids with Name attribute value “Body”. For doors, the Box geometry representation was selected due to complex Body geometry representation. The geometries of doors and spaces were inserted into the PostGIS database (tables spaces and doors), where a query (Appendix A.6) was executed, selecting the doors and spaces that intersect in 3D space. This process is aligned with our first method (using only the ST_3DIntersects function), but gives results similar to the second method (NRSE), as doors are included as spaces. This approach can be used ISPRS Int. J. Geo-Inf. 2020, 9, 417 7 of 15 All the test datasets were processed with all three presented methods in order to assess the performance of each method and the SQL queries included in each method. The processing times are used as results. While they enable relative comparisons, they depend heavily on the hardware that is used for processing. All the tests were performed on a PC equipped with i7-8565U CPU, 16 GB RAM and an SSD hard drive. Table 1 summarizes the processing results for the first and second method. Table 2 summarizes the results for the third method. For each method, a complete processing time is given, including all processes in FME and PostGIS. Additionally, the query execution times are given for each query that is included in the method. The processing times for Query A2 in Table 2 are repeated from Table 1, as both the query and the data are identical for both methods. Table 1. Results of processing of the 3D cadastral datasets for the first and the second method. Dataset No. of Spaces Method 1 Query A1 Method 2 Query A2 1 storey 31 29 s 0.08 s 1 min 10 s 42 s 3 storeys 93 59 s 1.4 s 3 min 8 s 2 min 25 s 6 storeys 186 1 min 46 s 7.1 s 6 min 39 s 4 min 28 s 9 storeys 279 2 min 58 s 16.2 s 10 min 16 s 6 min 30 s ISPRS Int. J. Geo-Inf. 2020, 9, 417 8 of 15 Table 2. Results of processing of the 3D cadastral datasets for the third method. Dataset No. of Spaces Method 3 Query A2 Query A3 Query A4 Query A5 to extract a navigation network for one floor as it lacks vertical connections that are modelled with 1 storey 31 1 min 11 s 42 s 0.02 s 0.02 s 0.04 s IfcStair entities. To add them, we select IfcStair solid geometries with Name attribute value “Box” and 3 storeys 93 3 min 13 s 2 min 25 s 0.04 s 0.06 s 0.1 s insert them into the database (table stairs). The union of three queries (Appendix A.7) selects pairs of 6 storeys 186 6 min 40 s 4 min 28 s 0.08 s 0.1 s 0.2 s intersecting doors 9 storeys and spaces, 279 pairs of intersecting 10 min 20 s spaces 6 min 30 s and stairs, 0.1 s and pairs of 0.2 s intersecting 0.3 s stairs. Figure 9 shows the navigation network derived from the result of these queries. The result contains the 3. connections 2. Navigation between Network all fro featur m IFC es that Data are in an intersecting relationship. Some doors intersect with spaces that are on a different floor, resulting in non-existent vertical connections between spaces (blue The second case study is based on the example project IFC file from BIMcollab [36] for the office lines in Figure 9). To avoid the extraction of these connections, the query Appendix A.7 was replaced type building (Figure 8). Since many versions of IFC models exist, it should be emphasized that our with three separate queries (Appendix A.8) with one using the ST_3Dintersection function that outputs method can be successfully applied only to the models that contain the IfcSpace, IfcDoor and IfcStair the geometry of the intersection. The non-existent connections were filtered in FME using the vertical entities. extent of the intersection geometries. ISPRS Int. J. Geo-Inf. 2020, 9, 417 8 of 15 function), but gives results similar to the second method (NRSE), as doors are included as spaces. This approach can be used to extract a navigation network for one floor as it lacks vertical connections that are modelled with IfcStair entities. To add them, we select IfcStair solid geometries with Name attribute value “Box” and insert them into the database (table stairs). The union of three queries (Appendix A7) selects pairs of intersecting doors and spaces, pairs of intersecting spaces and stairs, and pairs of intersecting stairs. Figure 9 shows the navigation network derived from the result of these queries. The result contains the connections between all features that are in an intersecting relationship. Some doors intersect with spaces that are on a different floor, resulting in non-existent vertical connections between spaces (blue lines in Figure 9). To avoid the extraction of these connections, the query A7 was replaced with three separate queries (Appendix A8) with one using the ST_3Dintersection function that outputs the geometry of the intersection. The non-existent Figure 8. IFC test dataset [36] visualized using the FZK Viewer. connections were filt Figure 8. ered in FMIFC E test usin dataset g th [36] e vert visu ical exalize tent d us of t ing h the F e inte ZK Vie rsecti wer. on geometries. The input IFC file was imported and decomposed in FME, allowing the selection of individual IFC entities. First, we limited the selection of entities to IfcSpace, IfcDoor. We filtered the geometries by the type of geometry, where only solid geometries were passed through. Another filter passes only door solids with Name attribute value “Box” and space solids with Name attribute value “Body”. For doors, the Box geometry representation was selected due to complex Body geometry representation. The geometries of doors and spaces were inserted into the PostGIS database (tables spaces and doors), where a query (Appendix A6) was executed, selecting the doors and spaces that intersect in 3D space. This process is aligned with our first method (using only the ST_3DIntersects Figure 9. Navigation network derived from the IFC dataset using the first method. Figure 9. Navigation network derived from the IFC dataset using the first method. Secondly, we applied the third proposed method that uses the SFCGAL function Secondly, we applied the third proposed method that uses the SFCGAL function ST_ApproximateMedialAxis to densify the navigation network and make it suitable for path planning ST_ApproximateMedialAx is to densify the navigation network and make it suitable for path planning applications. Figure 10 shows the navigation network obtained by applying the third method. applications. Figure 10 shows the navigation network obtained by applying the third method. Table 3 summarizes the results of processing the test IFC file that contains 51 IfcSpace, 58 IfcDoor and 3 IfcStair entities. The first method that generated the NRSE network structure was implemented using Appendices A.6–A.8 queries, while the third method was implemented using the Appendix A.8 queries, together with queries Appendices A.3–A.5 for calculating the approximate medial axes and finding the closest connection points on them. Figure 10. Navigation network derived from the IFC dataset by applying the third method. Table 3 summarizes the results of processing the test IFC file that contains 51 IfcSpace, 58 IfcDoor and 3 IfcStair entities. The first method that generated the NRSE network structure was implemented using A6, A7 and A8 queries, while the third method was implemented using the A8 queries, together with queries A3, A4 and A5 for calculating the approximate medial axes and finding the closest connection points on them. ISPRS Int. J. Geo-Inf. 2020, 9, 417 8 of 15 function), but gives results similar to the second method (NRSE), as doors are included as spaces. This approach can be used to extract a navigation network for one floor as it lacks vertical connections that are modelled with IfcStair entities. To add them, we select IfcStair solid geometries with Name attribute value “Box” and insert them into the database (table stairs). The union of three queries (Appendix A7) selects pairs of intersecting doors and spaces, pairs of intersecting spaces and stairs, and pairs of intersecting stairs. Figure 9 shows the navigation network derived from the result of these queries. The result contains the connections between all features that are in an intersecting relationship. Some doors intersect with spaces that are on a different floor, resulting in non-existent vertical connections between spaces (blue lines in Figure 9). To avoid the extraction of these connections, the query A7 was replaced with three separate queries (Appendix A8) with one using the ST_3Dintersection function that outputs the geometry of the intersection. The non-existent connections were filtered in FME using the vertical extent of the intersection geometries. Figure 9. Navigation network derived from the IFC dataset using the first method. Secondly, we applied the third proposed method that uses the SFCGAL function ST_ApproximateMedialAxis to densify the navigation network and make it suitable for path planning ISPRS Int. J. Geo-Inf. 2020, 9, 417 9 of 15 applications. Figure 10 shows the navigation network obtained by applying the third method. Figure 10. Navigation network derived from the IFC dataset by applying the third method. ISPRS Int. J. Geo-Inf. 2020, 9, 417 9 of 15 Figure 10. Navigation network derived from the IFC dataset by applying the third method. Table 3. Results of processing of the 3D cadastral datasets for the third method. Table 3. Results of processing of the 3D cadastral datasets for the third method. Table 3 summarizes the results of processing the test IFC file that contains 51 IfcSpace, 58 IfcDoor and 3 IfcStair enti Method Method ties. The No. of first method tha Edges No. of Edges t genera Query Query ted the NRSE netw Query Time ork str Query T u ime cture was Processing Ti implem Processing Time me ented using A6, A7 Method and A 1 Method 1 8 queri 109 es, wh 109 ile the third met A6 hod Appendix was A.6 impleme 0.2 s nted us 0.2 s ing the A8 9.6 s quer 9.6 ies, s together with queries Method A3, 1 Method 1 A4 and 113 A5 113 for calculating t A7 he ap Appendix p A.7 roximate m 0.2 s edial 0.2 axes s and fin 10 s ding 10 t s he closest Method 1 110 Appendix A.8 56.6 s 1 min 5 s connection points on them Method 1 . 110 A8 56.6 s 1 min 5 s Appendix A.8+ Appendix A.3 + Method 3 Method 3 157 157 A8 + A3 + A4 + A5 56.8 s 56.8 s 1 min 9 1 s min 9 s Appendix A.4 + Appendix A.5 4. Discussion 4. Discussion The proposed methods for indoor navigation network extraction use advanced SFCGAL The proposed methods for indoor navigation network extraction use advanced SFCGAL functions functions that process the geometries in 3D. This enables us to work fully automatically with 3D data, that process the geometries in 3D. This enables us to work fully automatically with 3D data, depending depending only on geometry. In 2D, a similar approach can be used, but it requires the building to only on geometry. In 2D, a similar approach can be used, but it requires the building to be divided be divided into floors to avoid vertical overlap of features, which cannot be properly handled by 2D into floors to avoid vertical overlap of features, which cannot be properly handled by 2D algorithms. algorithms. The division depends on semantic information in building models or must be done The division depends on semantic information in building models or must be done manually. There is manually. There is also the problem of assembling all processed floors into one vertically connected also the problem of assembling all processed floors into one vertically connected model. However, model. However, in some parts of the process, we also have to use 2D geometries. The first case is in some parts of the process, we also have to use 2D geometries. The first case is the centroid placement, the centroid placement, and the second is the calculation of the approximate medial axis. Both and the second is the calculation of the approximate medial axis. Both processes are used to calculate processes are used to calculate the placement of the navigation network nodes. Before the the placement of the navigation network nodes. Before the transformation of geometries to 2D, transformation of geometries to 2D, their vertical bounds are calculated to enable 3D placement of their vertical bounds are calculated to enable 3D placement of the calculated nodes. We avoid the the calculated nodes. We avoid the external centroids in 2D with the option in FME that guarantees external centroids in 2D with the option in FME that guarantees the point to be inside the 2D polygon. the point to be inside the 2D polygon. The nodes are placed in 3D using the middle value of the The nodes are placed in 3D using the middle value of the vertical bounds of space geometries. In some vertical bounds of space geometries. In some cases, where the geometries of the spaces are more cases, where the geometries of the spaces are more complex, the nodes can be placed outside of the 3D complex, the nodes can be placed outside of the 3D geometry of the space (Figure 11). geometry of the space (Figure 11). Figure 11. The node (red) placed out of the solid geometry due to variable ceiling height. Figure 11. The node (red) placed out of the solid geometry due to variable ceiling height. These errors can be found with additional processing in PostGIS. The nodes have to be inserted into PostGIS, where we can find the identifiers of solid geometries, which do not contain the corresponding node using the SQL query (Appendix A9). The query uses the computationally demanding ST_3DIntersection function. The reason is that the ST_3DIntersects function does not work properly when comparing solid geometry and point geometry. Instead, the ST_3DIntersection function is used to check if the intersection geometry is empty, meaning the node is outside the corresponding solid geometry. Another option that avoids the presented errors is to create a vertical line for each node and intersect it with the space geometry (Figure 12) using an SQL query (Appendix A10). The line is placed on the node location and has a larger vertical extent than the space geometry. The node that is inside the space geometry (red point) is between the calculated intersections (green points). Due to the usage of ST_3DIntersection function, both approaches significantly reduce the performance of the proposed methods, especially of the third method, where many network nodes are created for each space. Additional research is needed to improve the performance of the presented approaches for node placing inside 3D geometries. ISPRS Int. J. Geo-Inf. 2020, 9, 417 10 of 15 These errors can be found with additional processing in PostGIS. The nodes have to be inserted into PostGIS, where we can find the identifiers of solid geometries, which do not contain the corresponding node using the SQL query (Appendix A.9). The query uses the computationally demanding ST_3DIntersection function. The reason is that the ST_3DIntersects function does not work properly when comparing solid geometry and point geometry. Instead, the ST_3DIntersection function is used to check if the intersection geometry is empty, meaning the node is outside the corresponding solid geometry. Another option that avoids the presented errors is to create a vertical line for each node and intersect it with the space geometry (Figure 12) using an SQL query (Appendix A.10). The line is placed on the node location and has a larger vertical extent than the space geometry. The node that is inside the space geometry (red point) is between the calculated intersections (green points). Due to the usage of ST_3DIntersection function, both approaches significantly reduce the performance of the proposed methods, especially of the third method, where many network nodes are created for each ISPRS Int. J. Geo-Inf. 2020, 9, 417 10 of 15 space. Additional research is needed to improve the performance of the presented approaches for node placing inside 3D geometries. Figure 12. The line-space intersections (green) and the centroid (red). Figure 12. The line-space intersections (green) and the centroid (red). The results of the first case study (Tables 1 and 2) show the best performance of the first method, The resu which l pr ts of the fi oduces rs the t case most study basic (rTables 1 esults. and 2) show the best perf However, the results of orma the nce of the fi first method rst method, can be used as which prod input uces the most for more basic advanced results. indoor However, the navigation resul network ts of the first method extraction methods [ ca 33 n be ]. used While as i the npu time t share for more of adv SQL anced indoo query r navigation Appendix A.1 network extr execution time a is ction methods [3 low for the first3]. While the ti method, the me share of SQL execution time of the query A1 ex SQL ecution time query is low Appendix A.2 r for epr the first me esents the thod, the exec majority of the pr ution time ocessing of the SQL time of methods query A 2 and 3. 2 This is represents the majority caused by the of the processin SFCGAL g time ST_3DIntersection of me functionthods being 2 and 3. This is c computationally aused much by t more he SFCGAL demanding than ST_3DIn the tersection SFCGAL function being ST_3DIntersects comput function.ationa The lly much more FME processing t demand ime r ing t emains r han th oughly e S the FCGA same L for all ST_3DIntersec methods ts function and incr . T eases he FME proc linearly with essin the g time rema increase in in the s roughly the number of pr same ocessed for all metho spaces. The d di s ff an er d ence in increases line pr arly with the ocessing time for incre methodsase 2 in th and 3 e number o is minimal f as processed the query spaces. The d Appendix A.2 ifference in is the same pro for cessin both g methods. time for methods 2 Additional and queries 3 i ins minimal method 3 as the q that uery A2 calculate is the the sa appr me for b oximate oth methods. Addi medial axes and tional their queries connections are in method 3 basically that calculate th non-significant e ap in proxi terms m of ate me executidial on axes time and t as theyhei ar r connect e all ions executed are b in a less sicall than y non- 0.6 s for all significant cases. in te The rms of exec 3D ut cadastral ion t data ime as model they are that is all used ex in ecut this ed in case less t study han has 0.6 s an for all c additionalases. The 3D advantage when cadast it ra is l da pr ta model t ocessed hat using is use 3D d in this functions. case The study has an intersection additional geometry advantage calculated when it in is pro methods 2 cessed and 3 using using 3D functions. The ST_3DIntersection intersection geo function is a metry calculated in metho surface. The SFCGAL function ds 2 and 3 using ST_3DArea can ST be _3DInte used to rsection calculate the function area is of a s an urface. The S intersection F to CGA checkL func if the tion ST_3DA passage is rea can b navigable e for used to c a a particular lculate the ar locomotion ea of type. an intersection to check The results if the of the passage is navi second case gab study le (T for a able pa 3) rti pr cul ove ar locomoti that the pr on type. oposed methods can be efficiently The resu applied lts o to f the second c IFC data if it ase stud contains y (T the ab r le 3) prove equired that the propos entities (IfcSpace, ed methods c IfcDoor and an be e IfcStair). fficient Since ly the IFC applied to data IFC dat model a is if it cont standar ains t dized,he re any quired other ent IFC ities ( IfcS dataset pace that , IfcDoor and contains IfcS these tair). Si entities nce the IFC da can be pr ta ocessed the model is stan same dard way as ized, the an test y other IFC dataset that co dataset. Method 2 was nta not ins these enti used in this ties ca case n be study processed th as it aims to e same extract nodes way as at the test da the taset. connections Method 2 of spaces was (wher not used e they in t touch)his c and ase stud the y as it IfcSpace aims to extract nod geometries do not es touch at the each other. connections In of spaces combination (w with here they touch) an IfcDoor geometries, d the IfcSpace method 1 pr geom oducesetries do not similar r touch each ot esults as method her. In 2 in the first combinat case ion wit study. h IfcDoor The geometries, method query A8 reduces the err 1 pro ors in duces similar the resul navigation ts as method 2 network, but it r in the equires first ca the se use of the study. The query A8 red ST_3DIntersection uces the function, erro which rs in the nav significantly rigation network, but educes the query it requires the use of performance. the ST_3DIntersection function, which significantly reduces the query performance. In the cadastral data model, the stair spaces were modelled the same way as other spaces, which is not the case for the IFC data model. If the stairs are located inside the building, our approach successfully derived the vertical connections. Some IfcStair entities in the test dataset have no geometric contact with other IfcStair entities or any of the IfcDoor and IfcSpace entities, which caused some vertical connections to be left out. This can be seen in the upper part of Figure 9 and Figure 10, where a part of the navigation network is not connected to the rest of the network. This represents a disadvantage of the geometry-based methods, as there is no option to extract the connection if the geometries of the features are spatially disjoint. In these cases, the methods that are based on semantic information have the advantage. We compared our approach to the semantic-based approaches [30,31], which use the IfcRelSpaceBoundary entities to extract the navigation network from IFC dataset. The comparison can be done only for the IFC models, that also contain the IfcRelSpaceBoundary entity. We joined the RelatingSpace attribute from the IfcRelSpaceBoundary entity to IfcDoor, thus getting information about IfcDoor and IfcSpace relations that were used to derive the navigation network. The process was very fast, as the connections were found based on the attributes. However, we found ISPRS Int. J. Geo-Inf. 2020, 9, 417 11 of 15 In the cadastral data model, the stair spaces were modelled the same way as other spaces, which is not the case for the IFC data model. If the stairs are located inside the building, our approach successfully derived the vertical connections. Some IfcStair entities in the test dataset have no geometric contact with other IfcStair entities or any of the IfcDoor and IfcSpace entities, which caused some vertical connections to be left out. This can be seen in the upper part of Figures 9 and 10, where a part of the navigation network is not connected to the rest of the network. This represents a disadvantage of the geometry-based methods, as there is no option to extract the connection if the geometries of the features are spatially disjoint. In these cases, the methods that are based on semantic information have the advantage. We compared our approach to the semantic-based approaches [30,31], which use the IfcRelSpaceBoundary entities to extract the navigation network from IFC dataset. The comparison can be done only for the IFC models, that also contain the IfcRelSpaceBoundary entity. We joined the RelatingSpace attribute from the IfcRelSpaceBoundary entity to IfcDoor, thus getting information about IfcDoor and IfcSpace relations that were used to derive the navigation network. The process ISPRS Int. J. Geo-Inf. 2020, 9, 417 11 of 15 was very fast, as the connections were found based on the attributes. However, we found that the IfcRelSpaceBoundary entities contain errors, which caused some wrong connections between spaces and that the IfcRelSpaceBoundary entities contain errors, which caused some wrong connections between doors, shown in Figure 13. It should be noted that the files available in 2020 were altered and corrected spaces and doors, shown in Figure 13. It should be noted that the files available in 2020 were altered for these errors. and corrected for these errors. Figure 13. Figure 13. Errors in Errors in IfcSpace and IfcSpace and IfcDoor (having equ IfcDoor (having al co equal lour) relat colour) r ions retrie elations r ved from the etrieved from the IfcRelSpaceBoundary IfcRelSpaceBoundary entity. entity. The results of the case studies show that the quality of the extracted navigation network depends The results of the case studies show that the quality of the extracted navigation network depends on on several factors. The first is the data model of the input data, which determines how the input several factors. The first is the data model of the input data, which determines how the input data is data is structured and modelled. Consequentially, it determines how well the input data supports structured and modelled. Consequentially, it determines how well the input data supports the the geometry-based navigation network extraction. In the case studies, the 3D cadastral dataset geometry-based navigation network extraction. In the case studies, the 3D cadastral dataset causes causes no issues with the extraction of navigation network for stairs, while for the IFC dataset, some no issues with the extraction of navigation network for stairs, while for the IFC dataset, some vertical vertical connections cannot be extracted based on SFCGAL geometry functions. As the paper is connections cannot be extracted based on SFCGAL geometry functions. As the paper is focused on focused on the usage of SFCGAL functions, the quality of the extracted navigation network with the the usage of SFCGAL functions, the quality of the extracted navigation network with the proposed proposed methods is limited by the limited capabilities of those functions. For the third method, methods is limited by the limited capabilities of those functions. For the third method, we use the we use the ST_ApproximateMedialAxis function for the extraction of the navigation network inside each ST_ApproximateMedialAxis function for the extraction of the navigation network inside each room. room. For larger spaces, it produces the network that is less appropriate for indoor path planning. For larger spaces, it produces the network that is less appropriate for indoor path planning. More More advanced solutions exist, which, on the other hand, rely on having the room connectivity advanced solutions exist, which, on the other hand, rely on having the room connectivity available available [33]. The options to integrate these solutions with the proposed methods need to be explored [33]. The options to integrate these solutions with the proposed methods need to be explored in the in the future to automatically obtain more advanced navigation networks, suitable for indoor path future to automatically obtain more advanced navigation networks, suitable for indoor path planning. Among the most important factors that affect the quality of the results are the errors in the planning. Among the most important factors that affect the quality of the results are the errors in the input data. We demonstrated how the errors in the IfcRelSpaceBoundary entity affect the semantic-based input data. We demonstrated how the errors in the IfcRelSpaceBoundary entity affect the semantic-navigation network extraction. The presented geometry-based approach is sensitive to geometry errors. based navigation network extraction. The presented geometry-based approach is sensitive to geometry errors. In the 3D cadastral data model, the connected indoor spaces should be in a touching relationship. In the process of 3D modelling or processing the input data, gaps between spaces may occur. These gaps cause the methods to fail in the extraction of connections. The possible solution for small gaps is rounding the coordinates, but it can make the gap even larger in some cases (for instance, 0.149 is rounded to 0.1, and 0.151 is rounded to 0.2). The solution would be to have a flexible tolerance setting for geometry processing, but this is not available for the SFCGAL functions. IFC entities can have multiple geometry representations. Due to the complexity of “Body” representation, less complex “Box” geometry representation is used in our case study for IfcDoor and IfcStair entities. Using these less complex geometries can cause some non-existent connections to be identified by our proposed methods. As stated before, the proposed methods can be applied to any IFC file that contains IfcSpace, IfcDoor and IfcStair entities, but one has to be aware of the presented limitations that can affect the quality of the extracted navigation network. Future research is needed to investigate the complementary use of the proposed geometry-based methods together with semantic-based approaches that are not affected by the invalid and too complex geometries. ISPRS Int. J. Geo-Inf. 2020, 9, 417 12 of 15 In the 3D cadastral data model, the connected indoor spaces should be in a touching relationship. In the process of 3D modelling or processing the input data, gaps between spaces may occur. These gaps cause the methods to fail in the extraction of connections. The possible solution for small gaps is rounding the coordinates, but it can make the gap even larger in some cases (for instance, 0.149 is rounded to 0.1, and 0.151 is rounded to 0.2). The solution would be to have a flexible tolerance setting for geometry processing, but this is not available for the SFCGAL functions. IFC entities can have multiple geometry representations. Due to the complexity of “Body” representation, less complex “Box” geometry representation is used in our case study for IfcDoor and IfcStair entities. Using these less complex geometries can cause some non-existent connections to be identified by our proposed methods. As stated before, the proposed methods can be applied to any IFC file that contains IfcSpace, IfcDoor and IfcStair entities, but one has to be aware of the presented limitations that can affect the quality of the extracted navigation network. Future research is needed to investigate the complementary use of the proposed geometry-based methods together with semantic-based approaches that are not affected by the invalid and too complex geometries. 5. Conclusions The presented methodology for geometry-based indoor navigation network extraction is developed using SFCGAL functions that support 3D geometries. This allows the original 3D geometries of building models to be used in a process without dividing the building models into separate floors. The developed methodology enables the full automation of the process from input data import to the final result in the form of a navigation network. Three methods are developed to demonstrate the possibilities of using SFCGAL functions for geometry-based navigation network extraction. The methods differ in the complexity of the output navigation network. To achieve the flexibility in terms of input data and to provide access to SFCGAL functions, the FME spatial ETL software, supporting a wide range of data formats, was integrated with the PostGIS database, providing access to SFCGAL functions. Future research can investigate the possibilities to use alternative software to implement the proposed methods. A wide range of supported formats and flexibility of the FME software allows the proposed methods to be applied on various datasets that provide 3D indoor information besides the case study data (for instance CityGML LOD 4 models or DWG files), which also needs to be investigated in the future. The pre-processing of geometries is crucial for efficient implementation of the proposed methods. The deaggregation of various input geometry types and surface triangulation in FME proved to be an efficient approach for obtaining valid solid geometries in PostGIS that can be analysed using SFCGAL functions. SQL queries were used to process 3D geometries and extract connectivity information. Besides the navigation network extraction, this integration can be used to perform various 3D analyses of spatial data. The case studies show that the proposed methods are efficient enough to enable processing of larger datasets. The case study with a 3D cadastral dataset aims to emphasize the multipurpose role of a 3D cadastral system and data in the future. While 3D geometries of indoor spaces significantly help with clarifying the legal situation in the building, they can, if properly structured, also support many new applications, including indoor navigation. The second case study showed some deficiencies of the presented methodology when applied to IFC data. Although most of the connections were identified and properly modelled in 3D, some connections were not found due to the methodology being purely geometry-based and fully automated. Additionally, the proposed methodology can be used to validate the IfcRelSpaceBoundary entities. In the test dataset, they provided some physically impossible connections that were discovered using the proposed methodology. Author Contributions: Conceptualization, investigation, writing—review and editing, Jernej Tekavec and Anka Lisec; methodology, formal analysis, software, validation, visualization, writing—original draft preparation, Jernej Tekavec; supervision, Anka Lisec. All authors have read and agreed to the published version of the manuscript. Funding: The authors acknowledge the financial support from the Slovenian Research Agency (research core funding No. P2-0406 Earth observation and geoinformatics). ISPRS Int. J. Geo-Inf. 2020, 9, 417 13 of 15 Conflicts of Interest: The authors declare no conflict of interest. Appendix A. Appendix A.1. SQL Query for Selection of Identifiers of Geometries that Intersect in 3D Space—Method 1 SELECT a.sn as start, b.sn as end FROM spaces a, spaces b WHERE ST_3DIntersects(a.geom, b.geom) AND a.sn! = b.sn Appendix A.2. SQL Query for Selection of Identifiers and Intersection Geometry of Geometries that Intersect in 3D Space—Method 2 SELECT a.sn as start, b.sn as end, ST_3DIntersection(a.geom, b.geom) FROM spaces a, spaces b WHERE ST_3DIntersects(a.geom, b.geom) AND a.sn! = b.sn Appendix A.3. SQL Query that Generates an Approximate Medial Axis for Each Space SELECT sn, ST_ApproximateMedialAxis(geom) as geom FROM spaces2d Appendix A.4. SQL Query that Generates Points from the Approximate Medial Axis for Each Space SELECT sn, (ST_DumpPoints (ST_ApproximateMedialAxis(geom)).geom as geom FROM spaces2d Appendix A.5. SQL Query that Finds Closest Points on Edges of the Connected Networks SELECT a.id, b.sn, ST_ClosestPoint (b.geom, a.geom) FROM connections a INNER JOIN (SELECT pn, ST_ApproximateMedialAxis (geom) as geom FROM spaces2d) b ON a.start = b.sn OR a.end = b.sn Appendix A.6. SQL Query for Selection of Intersecting Doors and Spaces SELECT a.id as start, b.id as end FROM doors a JOIN spaces b ON ST_3Dintersects (a.geom, b.geom) Appendix A.7. SQL Query for Selection of Intersecting Doors, Spaces and Stairs SELECT a.id as start, b.id as end FROM doors a JOIN spaces b ON ST_3Dintersects (a.geom, b.geom) UNION ALL SELECT a.id as start, b.id as end FROM stairs a JOIN spaces b ON ST_3Dintersects (a.geom, b.geom) UNION ALL SELECT a.id as start, b.id as end FROM stairs a JOIN stairs b ON ST_3Dintersects (a.geom, b.geom) AND a.id < b.id Appendix A.8. SQL Queries for Selection of Intersecting Doors, Spaces and Stairs with Added Intersection Geometry for the Door–Space Intersections SELECT a.id as start, b.id as end, ST_3DIntersection(a.geom, b.geom) FROM doors a JOIN spaces b ON ST_3Dintersects (a.geom, b.geom) SELECT a.id as start, b.id as end FROM stairs a JOIN spaces b ON ST_3Dintersects (a.geom, b.geom) SELECT a.id as start, b.id as end FROM stairs a JOIN stairs b ON ST_3Dintersects (a.geom, b.geom) ISPRS Int. J. Geo-Inf. 2020, 9, 417 14 of 15 AND a.gid < b.gid Appendix A.9. SQL Query for Selection of Spaces which do not Contain the Corresponding Centroid SELECT a.sn FROM spaces a, centroids b WHERE a.sn = b.sn AND ST_IsEmpty (ST_3DIntersection(ST_MakeSolid (a.geom), b.geom)) Appendix A.10. SQL Query that Selects the Intersecting Points of Lines and Spaces for Centroid Placement in 3D SELECT a.sn, ST_3DIntersection (a.geom, b.geom) FROM spaces a, lines b WHERE a.sn = b.sn References 1. Biljecki, F.; Stoter, J.; Ledoux, H.; Zlatanova, S.; Çöltekin, A. Applications of 3D City Models: State of the Art Review. ISPRS Int. J. Geo-Inf. 2015, 4, 2842–2889. [CrossRef] 2. Gunduz, M.; Isikdag, U.; Basaraner, M. A review of recent research in Indoor modelling & mapping. ISPRS–Int. Arch. Photogramm. Remote. Sens. Spat. Inf. Sci. 2016, 41, 289–294. [CrossRef] 3. Liu, X.; Wang, X.; Wright, G.; Cheng, J.; Li, X.; Liu, R. A State-of-the-Art Review on the Integration of Building Information Modeling (BIM) and Geographic Information System (GIS). ISPRS Int. J. Geo-Inf. 2017, 6, 53. [CrossRef] 4. Rahman, A.A. Advances in 3D Geoinformation; Springer International Publishing: Cham, Switzerland, 2017. [CrossRef] 5. Brena, R.F.; García-Vázquez, J.P.; Galván-Tejada, C.E.; Muñoz-Rodriguez, D.; Vargas-Rosales, C.; Fangmeyer, J. Evolution of Indoor Positioning Technologies: A Survey. J. Sens. 2017, 6, 359. [CrossRef] 6. Paul, N.; Borrmann, A. Geometrical and topological approaches in building information modelling. J. Inf. Technol. Constr. 2009, 14, 705–723. 7. Gröger, G.; Plümer, L. How to achieve consistency for 3D city models. GeoInformatica 2011, 15, 137–165. [CrossRef] 8. Zhao, Z.; Stoter, J.; Ledoux, H. A Framework for the Automatic Geometric Repair of CityGML Models. In Cartography from Pole to Pole, Lecture Notes in Geoinformation and Cartography; Buchroithner, M., Prechtel, N., Burghardt, D., Eds.; Springer: Berlin/Heidelberg, Germany, 2014. [CrossRef] 9. Deng, Y.; Cheng, J.C.P.; Anumba, C. Mapping between BIM and 3D GIS in different levels of detail using schema mediation and instance comparison. Autom. Constr. 2016, 67, 1–21. [CrossRef] 10. Ohori, K.A.; Diakité, A.; Krijnen, T.; Ledoux, H.; Stoter, J. Processing BIM and GIS Models in Practice: Experiences and Recommendations from a GeoBIM Project in The Netherlands. ISPRS Int. J. Geo-Inf. 2018, 7, 23. [CrossRef] 11. Knoth, L.; Scholz, J.; Strobl, J.; Mittlböck, M.; Vockner, B.; Atzl, C.; Rajabifard, A.; Atazadeh, B. Cross-Domain Building Models—A Step towards Interoperability. ISPRS Int. J. Geo-Inf. 2018, 7, 363. [CrossRef] 12. Diakité, A.A.; Zlatanova, S. Valid Space Description in BIM for 3D Indoor Navigation. Int. J. 3-D Inf. Modeling 2016, 5, 1–17. [CrossRef] 13. Kang, H.-K.; Li, K.-J. A Standard Indoor Spatial Data Model—OGC IndoorGML and Implementation Approaches. ISPRS Int. J. Geo-Inf. 2017, 6, 116. [CrossRef] 14. Xiong, Q.; Zhu, Q.; Du, Z.; Zlatanova, S.; Zhang, Y.; Zhou, Y.; Li, Y. Free multi-floor indoor space extraction from complex 3D building models. Earth Sci. Inform. 2017, 10, 69–83. [CrossRef] 15. Yang, L.; Worboys, M. Generation of navigation graphs for indoor space. Int. J. Geogr. Inf. Sci. 2015, 29, 1737–1756. [CrossRef] 16. Diakité, A.A.; Zlatanova, S. Extraction of the 3D Free Space From Building Models for Indoor Navigation. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2016, 4, 241–248. [CrossRef] 17. Lin, Z.; Xu, Z.; Hu, D.; Hu, Q.; Li, W. Hybrid Spatial Data Model for Indoor Space: Combined Topology and Grid. ISPRS Int. J. Geo-Inf. 2017, 6, 343. [CrossRef] 18. Mortari, F.; Clementini, E.; Zlatanova, S.; Liu, L. An indoor navigation model and its network extraction. Appl. Geomat. 2019, 11, 413–427. [CrossRef] ISPRS Int. J. Geo-Inf. 2020, 9, 417 15 of 15 19. Thomsen, C. Extract-Transform-Load. In Encyclopedia of Big Data Technologies; Sakr, S., Zomaya, A.Y., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 728–733. [CrossRef] 20. SFGGAL. Wrapper around CGAL Library for 2D and 3D Operations Based on OGC Standards. 2019. Available online: http://www.sfcgal.org (accessed on 5 March 2020). 21. Open Geospatial Consortium. City Geography Markup Language (CityGML) Encoding Standard, Version 2.0. 2012. Available online: http://www.opengeospatial.org/standards/citygml (accessed on 5 March 2020). 22. Open Geospatial Consortium. IndoorGML, Version 1.02. 2014. Available online: http://www.opengeospatial. org/standards/indoorgml (accessed on 5 March 2020). 23. ISO. Industry Foundation Classes (IFC) for Data Sharing in the Construction and Facility Management Industries; International standard ISO 16739-1:2018; International Organization For Standardization: Geneva, Switzerland, 2018. 24. Tekavec, J.; Lisec, A. Cadastral data as a source for 3D indoor modelling. Land Use Policy 2020, 104322. [CrossRef] 25. Aien, A.; Rajabifard, A.; Kalantari, M.; Williamson, I. Review and Assessment of Current Cadastral Data Models for 3D Cadastral Applications. In Advances in 3D Geoinformation, Lecture Notes in Geoinformation and Cartography; Rahman, A.A., Ed.; Springer International Publishing: Cham, Switzerland, 2017; pp. 423–442. [CrossRef] 26. Alattas, A.; Zlatanova, S.; van Oosterom, P.; Chatzinikolaou, E.; Lemmen, C.; Li, K.-J. Supporting Indoor Navigation Using Access Rights to Spaces Based on Combined Use of IndoorGML and LADM Models. ISPRS Int. J. Geo-Inf. 2017, 6, 384. [CrossRef] 27. Lin, Y.H.; Liu, Y.S.; Gao, G.; Han, X.G.; Lai, C.Y.; Gu, M. The IFC-based path planning for 3D indoor spaces. Adv. Eng. Inform. 2013, 27, 189–205. [CrossRef] 28. Xu, M.; Wei, S.; Zlatanova, S.; Zhang, R. BIM-Based Indoor Path Planning Considering Obstacles. ISPRS Ann. Photogramm. Remote. Sens. Spat. Inf. Sci 2017, 4, 417–423. [CrossRef] 29. Staats, B.R.; Diakité, A.A.; Voûte, R.L.; Zlatanova, S. Detection of doors in a voxel model, derived from a point cloud and its scanner trajectory, to improve the segmentation of the walkable space. Int. J. Urban Sci. 2018, 23, 369–390. [CrossRef] 30. Teo, T.-A.; Cho, K.-H. BIM-oriented indoor network model for indoor and outdoor combined route planning. Adv. Eng. Inform. 2016, 30, 268–282. [CrossRef] 31. Khan, A.A.; Donaubauer, A.; Kolbe, T.H. A multi-step transformation process for automatically generating indoor routing graphs from existing semantic 3D building models. In Proceedings of the 9th 3D GeoInfo Conference, Dubai, United Arab Emirates, 11–13 November 2014. 32. Lewandowicz, E.; Lisowski, P.; Flisek, P. A Modified Methodology for Generating Indoor Navigation Models. ISPRS Int. J. Geo-Inf. 2019, 8, 60. [CrossRef] 33. Boguslawski, P.; Mahdjoubi, L.; Zverovich, V.; Fadli, F. Automated construction of variable density navigable networks in a 3D indoor environment for emergency response. Autom. Constr. 2016, 72, 115–128. [CrossRef] 34. Liu, L.; Zlatanova, S. Towards a 3D Network Model for Indoor Navigation. In Urban and Regional Data Management; Zlatanova, S., Ledoux, H., Fendel, E.M., Rumor, M., Eds.; CRCpress/Taylor and Francis Group: London, UK, 2012. [CrossRef] 35. Tang, S.J.; Zhu, Q.; Wang, W.W.; Zhang, Y.T. Automatic topology derivation from IFC building model for in-door intelligent navigation. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. –ISPRS Arch. 2015, 40, 7–11. [CrossRef] 36. BIMcollab, Examples and Templates. Available online: https://www.bimcollab.com/en/Support/Support/ Downloads/Examples-templates (accessed on 5 March 2020). © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Document Outline Abstract Introduction 3D cadastral data models Materials and methods Data model design Data model concept Results DBMS implementation Data model integration Integration with IFC Integration with CityGML Integration with IndoorGML Discussion Conclusions Disclosure statement Notes on contributors ORCID References Cadastral data as a source for 3D indoor modelling Introduction Literature review Research methodology and materials Framework design Data preparation and digitization phase Transformation phase Database data storage 2D database storage 3D database storage 3D data visualization Structuration phase Discussion Data assessment Findings Conclusions Acknowledgements References Introduction Materials and Methods Procedural Modelling of Buildings Transforming the Generated Building Models Georeferencing Using the Link to the Existing Cadastral Data Results Discussion Conclusions SQL Query to Select the Data for the Table Building_Simulated SQL Query to Select the Data for the Table Room_Simulated Modified A1 SQL Query (Modifications in Bold Text) to Select the Data for the Table Building_Simulated Modified A2 SQL Query (Modifications in Bold Text) to Select the Data for the Table Room_Simulated Modified A3 SQL Query (Modifications in Bold Text) to Select the Data for the Table Building_Simulated SQL Query that Checks for the Intersection of 3D Geometries in Building_Simulated Table SQL Query that Checks for the Overlap of 3D Geometries in Room_Simulated Table SQL Query that Checks for the Intersection of Room Geometries with Building Exterior Geometry References Introduction Methodology Results Navigation Network from 3D Cadastral Data Navigation Network from IFC Data Discussion Conclusions SQL Query for Selection of Identifiers of Geometries that Intersect in 3D Space—Method 1 SQL Query for Selection of Identifiers and Intersection Geometry of Geometries that Intersect in 3D Space—Method 2 SQL Query that Generates an Approximate Medial Axis for Each Space SQL Query that Generates Points from the Approximate Medial Axis for Each Space SQL Query that Finds Closest Points on Edges of the Connected Networks SQL Query for Selection of Intersecting Doors and Spaces SQL Query for Selection of Intersecting Doors, Spaces and Stairs SQL Queries for Selection of Intersecting Doors, Spaces and Stairs with Added Intersection Geometry for the Door–Space Intersections SQL Query for Selection of Spaces which do not Contain the Corresponding Centroid SQL Query that Selects the Intersecting Points of Lines and Spaces for Centroid Placement in 3D References