G i A V GEODETSKI VESTNIK | letn. / Vol. 601 št. / No. 2 V2/ 6(1/2 analiza kakovosti central geospatial ^ podatkov cestne database analysis of i infrastrukture the quality of road f kot dela centralne infrastructure data | geoprostorske | podatkovne zbirke Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic UDK: 711.7+004.65 Klasifikacija prispevka po COBISS.SI: 1.04 Prispelo: 7. 3. 2016 Sprejeto: 10. 6. 2016 DOI: 10.15292/geodetski-vestnik.2016.02.269-284 PROFESSIONAL ARTICLE Received: 7. 3. 2016 Accepted: 10. 6. 2016 _ IZVLEČEK V članku je opisan osnovni logični podatkovni model centralne baze podatkov (CGB). Model je uporabljen pri podatkovni strukturi listov digitalne topografske karte največjega merila — 1 : 25.000, ki so bili izdelani v Vojaškem geografskem inštitutu v Beogradu v Srbiji. Analiza kakovosti cestne infrastrukture v CGB je bila izdelana po standardu ISO 19157, kar je vodilo do nivojskega razvrščanja elementov inpodelementov kakovosti, ki so bili analizirani ločeno. Standard ISO 19157 določa postopek ocenjevanja kakovosti, ki prikazuje rezultate analize za vsak element. ABSTRACT _ This paper describes the basic logical data model from the central geospatial database (CGD) which is applicable to the structure of a digital topographic map of the largest scale — 1:25,000 made in the Military Geographical Institute (MGI), Belgrade, Serbia. CGD analysis of the road infrastructure quality, done according to the ISO 19157 standard, led to a layering of elements and sub-elements of quality which were analysed separately. ISO 19157 standard defines a procedure of quality assessment which depicts the results of the analysis for each element. KLJUČNE BESEDE KEY WORDS baza prostorskih podatkov, digitalna topografska karta, cestna infrastruktura, kakovost podatkov Geospatial database, digital topographic map, road infrastructure, data quality Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic', Miodrag Regodic' | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 269 | |60/2| GEODETSKI VESTNIK ^ 1 INTRODUCTION sg Spatial data represent a "reality model", a coherent and generalized display of a complex reality. Each 5 map or database is thus a model with a certain intent which contains a number of generalized elements i which are simplified, grouped or eliminated, all for the purpose of a clearer presentation and stimula-=■ tion of the communication process through information. Having an understanding of the data and •Ei its accuracy is of crucial importance both for the users and the publishers of geographic, topographic H and thematic maps. Their ability to be tested and quantified precisely is of crucial importance in this ^ field of work. Modern topographic cartography led to the established conventional solutions and standards in content display of topographic maps (TM) which raised their utility value in the country and worldwide (Petrovic, 2006; Zhang et al., 2013; Amovic et al., 2015). By following the development of standards in the process of collecting, organizing, processing and presenting of spatial data in the Military Geographic Institute (MGI), spatial data are organized in the central geospatial database (CGD). The main purpose of CGD is to be the basis for generating the whole scale series of digital topographic maps which are made in the MGI. The main part of the CGD content represents spatial data that define the digital topographic map at a scale of 1:25,000. By using methods of cartographic generalization we produce whole scale series of digital topographic maps. Topographic maps have a great advantage over some other collections of spatial data because of their visual effects, conciseness s and the simplicity of use and that is why the demands for a developed system of elaboration, assessment and reporting procedures of their quality are completely justified (Bozic et al., 2011; Radojcic et al., 2011). Road infrastructure represents a thematic layer of the central geospatial database. It is significant regarding the aspects of usage of different network analyses like: the establishment of the shortest route between certain locations, navigation in traffic, tracking the transport of goods, services and the population. In countries with a developed and stable economy, a clearly articulated road network data is usually established on the basis of updated topographic data and with the usage of updated and high quality spatial data and modern topographic technology. In less developed countries, there is an alternative and more financially acceptable strategy which is used for years for updating road network data by using the existing topographic maps, satellite images and other sources of spatial data which are available. (Siva Kumar, 2000; Ozah et al., 2008). Because of all the above mentioned, the main goal of this research is to analyse the quality of this thematic layer of CGD in accordance with ISO 19157 standard and to present the results of the analysis. The central geospatial database represents the basic topographic database in MGI. By using procedures of cartographic generalization and map design of CGD we produce whole scale series of digital topographic maps. Because of the mentioned, the main purpose of CGD is military and civilian mapping, various GIS analyses, navigation. The possibility of assessment and presentation of road infrastructure quality of CGD is in question. Due to diverse needs of users, the most frequent questions we encounter during the analysis of the road infrastructure quality are (Shekhar et al., 2002; Fisher et al., 2006, Goodchild et al., 2006): — Does the thematic layer contain all the elements of the database defined by the physical data model? — What is the value of road infrastructure positional accuracy for the needs of navigation? Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA 270 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | GEODETSKI VESTNIK | 60/2 | — Are all elements of thematic layer connected well enough to enable network analyses? — Are there any other geometries and irregular junctions? — When were the data collected and when were they updated? Several contributions were made for the needs of the research. An advanced conceptual frame for analysing the road infrastructure quality of CGD, which contains clear methodological and technological wholes, has been proposed. The methodological part forms an empirical frame for a selected set of limitations of a coherent evaluation of geospatial data, readability, keeping, entity level, specific/relational limitations, individual and group limitations, characteristics of classes composed of "enrichment", finding and date evaluation. In the implementation phase, the focus is on the evaluation of characteristics of classes which define the road infrastructure in CGD on the basis of this theory. 2 CENTRAL GEoSPATIAL DATABASE, DATA MoDEL AND A DEPICTioN oF THEMATIC LAYER oF road infrastructure The technological process of CGD creation is based on the mapping of the topographic map content by using methods of digital photogrammetric restoration, direct mapping from modern photogram-metric base maps such as orthophoto and its cartographic processing in CGD by using various alphanumeric data and the data acquired on the field. The process of vectorization is realized with a strict respect of a logical data model, bearing in mind possibilities and the work mode in the chosen software environment. SQL database and the software platform of the American firm ESRI, ArcGIS, which contains an entirely new approach in the process of creation of geospatial databases are chosen for the creation of the CGD. The selection of this software platform requires a completely new technology in all phases of work, but the existing cartographic solutions remain. With the consideration of these needs, the process of CGD creation encompasses the following phases (Sekulovic et al., 2011): — Creation of a logical data model, — Creation of a physical data model, — Creation of a model and generalization procedures, — Creation of symbols, — Quality analysis. In the process of creation of the logical data model of CGD specific geographical elements of TM are divided into thematic units. Particular cases of each of the content elements are defined by the system of layers and codes as unique indicators of belonging to the corresponding thematic unit, which is a closely defined specification of a particular class of objects. The main logical data model corresponds to the structure of a digital topographic map (DTM) of the largest scale 1:25,000 made in the Military Geographical Institute — Belgrade. Physical data model defines the appearance of the database, i.e., „the space" for the storage of data defined by the logical data model. Types of data, the way of storing as well as all the columns in which attributes of classes and specific objects are written are also defined in the process of creation of the physical data model (Table 1). Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 271 | 60/2 | GEODETSKI VESTNIK Table 1: Physical data model of the thematic layer - Roads. | DTK25.DBO.Putevi Alias Putevi Geometry:Polyline Dataset Type FeatureClass Average Number of Points: 0 Has M:No FeatureType Simple Has Z:Yes Grid Size:-6 Field Name Alias Name Model Name Type Precn. Scale Length Null OBJECTID OBJECTID OBJECTID OID 10 0 4 No SIFRA_txt SIFRA_txt SIFRA_txt String 0 0 4 Yes LAYER LAYER LAYER Integer 1O O 4 Yes SIFRA SIFRA SIFRA Integer 1O O 4 Yes NAZIV NAZIV NAZIV String O O 5O Yes VISINA VISINA VISINA Double 3S S S Yes OZNAKA OZNAKA OZNAKA String O O 1O Yes NAPOMENA NAPOMENA NAPOMENA String O O 255 Yes VEZA_1 VEZA_1 VEZA_1 String O O 5O Yes VEZA_2 VEZA_2 VEZA_2 Integer 1O O 4 Yes VEZA_3 VEZA_3 VEZA_3 Double 3S S S Yes VEZA_4 VEZA_4 VEZA_4 Blob 0 0 0 Yes DATUM DATUM DATUM Date O O 36 Yes NAZIV_ NAZIV_ NAZIV_ String 0 0 255 Yes ELEMENTA ELEMENTA ELEMENTA VISINA_OBJEKTA VISINA_OBJEKTA VISINA_OBJEKTA Double 3S S S Yes ANOTACIJA_1 ANOTACIJA_1 ANOTACIJA_1 Double 3S S S Yes ANOTACIJA_2 ANOTACIJA_2 ANOTACIJA_2 Double 3S S S Yes ANOTACIJA ANOTACIJA ANOTACIJA String 0 0 50 Yes ZNACAJ ZNACAJ ZNACAJ String O O 5O Yes TIP_ZASTORA TIP_ZASTORA TIP_ZASTORA String 0 0 50 Yes GlobalID GlobalID GlobalID Global ID O O 3S No RuleID RuleID RuleID Integer 1O O 4 Yes Override Override Override Blob O O O Yes BSP_PUTA BSP_PUTA BSP_PUTA String 0 0 10 Yes SHAPE SHAPE SHAPE Geometry O O 4 Yes FOTO FOTO FOTO Raster 0 0 4 Yes SHAPE.STLength() SHAPE.STLength() SHAPE.STLength() Double O O O No Physical data model precedes the hierarchical model and in a wider sense it defines the database domain, the coordinate system, i.e. a place for storing the main graphical elements, their attributes and the default relations (relationship class) among the complex (groups of classes of objects) and simple (classes of objects) elements that are closely defined. Default relations are automatically generated in the database Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA | 284 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 272-284 | GEODETSKI VESTNIK | 60/2 | and as a result the obtained themes are derived from the basic classes of objects. This can be explained on the example of conversion of labels into annotations where a new class of objects is automatically created. (e.g. "RoadsAnno" from the class of objects "Roads"). Default relations enable changes in the basic classes of objects to be automatically carried out in the derived classes of objects as well ("RoadsAnno"). The derived classes are physically present only in the database because their basic, essential parameters are either obtained from the default classes of objects or predefined by the software (Figure 1). BBB Figure 1: A picture of the representation of road infrastructure thematic layer. The definition of the appearance of CGD is a type of a physical data modelling where in the interface of the used software the sequence of the themes display is defined, i.e., the visual display of CGD is defined on a level that cannot be achieved through symbols. This phase represents the last step in the creation of CGD concerning the practical creation. 3 THE TEST EXAMPLE AND THE USED DATA The test area, which is analysed in this work, encompasses the road infrastructure thematic layer of a part of CGD of the Military Geographical Institute in the period of the testing shown on the Figure 2. A total number of 317,877 spatial objects analysed in the thematic layer of road infrastructure are shown in the Table 2. This includes a total of 4198 of point objects (labelled as "Objects on communication 1"), 259,991 of line objects (labelled as "Objects on communication 2"), 97 polygon objects (labelled as "Objects on communication 3") and a total of 53,591 objects with names presented in a form of annotations, geographical names related to roads. The marked rectangles on Figure 2 represent the tested sheets of digital topographic maps at scale 1:25,000, as part of CGD. Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 273 |60/2| GEODETSKI VESTNIK Figure 2: Test example for the evaluation of the road infrastructure quality. Table 2: The number of analysed spatial objects. Number Thematic layer title Total Number of elements Thematic layer of corresponding annotations Total number of annotations 1. Objects on communication 1 4198 Objects on comm._1_Anno 1105 2. Objects on communication 2 511 Objects on comm._2_Anno 264 3. Roads 259,480 Roads_Anno 52,171 4. Objects on communication 3 97 Objects on comm._3_Anno 51 4 ANALYISIS OF THE ROAD INFRASTRUCTURE QUALITY ACCORDING TO ISO QUALITY ELEMENTS We performed the analysis of the road infrastructure by using the classification of elements and subelements of spatial data quality defined by the ISO 19157 standard for the specified test area. According to ISO 19157 standard, the quality of spatial data is defined by a quantitative model through 5 elements: completeness, logical consistency, positional accuracy, time and thematic accuracy. Each element of quality is individually analysed and assessed (Drobnjak, et al., 2014; Talhofer, et al., 2011). The assessment procedure of road infrastructure data quality is carried out in accordance with the defined ISO 19157 standard procedure with steps displayed on Figure 3. (ISO 19157, 2011). Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA 274 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | GEODETSKI VESTNIK | 60/2 | Data Start evaluation process Specify i- Stepi: Specify data quality unit(s) -ï-_ If data quality is evaluated, the input dat is the actual data. If Metaquality is evaluated, the input data is the results from a data quality evaluation. Step 2: Specify data quality measures Step 3: Specify data quality evaluation procedures I Step 4: Determine the output of the data quality evaluation Product specification or user requirements Conformance quality level Evaluation finished Report data quality results (quantitative) J Figure 3: The procedure of quality assessment according to ISO 19157. 4.1 Completeness Completeness is an element of quality which describes the relationship between the spatial objects present in the data collection and objects which represent the abstractions of the modern world. It is usually described as the presence or the absence of spatial data, their attributes or relations. This is why completeness contains two main sub-elements: omission and commission of data (Servigne et al., 2006; Petrovič, 2006). Quality measurements for completeness which were analysed in this work are: the number of objects, the number of doubled objects and the number of missing objects of the road infrastructure, their attributes and relations. Physical data model defines the direct listing of annotations of height or depth (in meters) of banks and cuts if they are encountered on the road section, the foundation width, i.e. road width, as well as the base type of the road infrastructure (Figure 4). When it comes to bridges, additional data is listed concerning the construction material of the bridge, its transport capacity (in tonnes) and width of the bridge (in meters). In the process of making of the DTM within the CGD it happens that the restorers make mistakes and accidentally erase annotations while attributes remain Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 275 |60/2| GEODETSKI VESTNIK ^ in the attribute field and that is how we encounter an excess of attributes with regard to their relation ¡= which represents annotation. Figure 4: Defined annotations on the road infrastructure. The assessment of the excess, doubled or missing objects is realized by an internal method of quality assessment of CGD spatial data, by using ArcGis Data Reviewer extension. This extension contains tools which were used for the analysis of the total number of excess, doubled and missing objects and their relations and those are: the tool for the number of excess and missing elements (Table to Table Check) and the tools for the double geometry analysis (Duplicate Geometry Check). Table 3 shows the results of the quality analysis for completeness of the road infrastructure while the diagram (Graph 1) gives a graphical depiction of the total number of excess, doubled and missing objects in relation to the number of listed annotations for the test area. Table 3: Analysis of the road infrastructure completeness. Dataset layer Total number of objects in universe of discourse Data Quality Measure (data type value) Number of excess items (integer) Rate of excess items (%) Number of duplicate feature instances (integer) Number of missing items (integer) Rate of missing items (%) C O M M U N I C A T I O N Objects on communication 1 4198 4 0.10 3 7 0.17 Objects on communication 2 511 1 0.20 5 7 1.37 Objects on communication 3 97 0 0.00 1 1 1.03 Roads 259,480 13 0.01 25 5 0.00 Objects on comm._1_Anno 1105 7 0.63 10 12 1.09 Roads_Anno 52,171 14 0.03 18 10 0.02 Objects on comm._2_Anno 264 2 0.76 7 5 1.89 Objects on comm._3_Anno 51 1 1.96 2 1 1.96 Z 317,877 42 71 48 Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA 276 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | GEODETSKI VESTNIK |60/2 | Graph 1: Analysis of the road infrastructure completeness. 4.2 Logical consistency Logical consistency represents the degree of the data structure compatibility, its attributes and relations with logical rules. The mentioned structure can be conceptual, logical and physical. So the data collection is consistent on a logical level if it follows the structural characteristics of the selected data model and if it is consistent with the attributes and relations defined for that data (Devillers et al., 2006). Sub-elements of logical consistency are: — Conceptual consistency (consistency with the rules of conceptual scheme), — Domain consistency (consistency of data values with the allowed values), — Topological consistency (correctness ofthe explicitly encoded topological characteristics ofa dataset), — Format consistency (the degree of data compatibility with the physical data structure). ESESZB^HKv d -U*r Figure 5: An example of logical consistency - verification of topological rules Conceptual consistency, domain and format consistency are the three sub-elements of logical consistency quality which define the integrity of a spatial database. If errors occur in any of the mentioned sub-elements they have to be immediately removed because they diminish the integrity of the spatial Sinisa Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUCTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OFTHE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | I 277 | | 60/2 | GEODETSKI VESTNIK ^ database and can completely disable any work with spatial data from the database. Topological consistIs ency is analysed with the usage of topological rules. Those rules are depicted in Figure 5. | An internal, direct and complete method of quality assessment, done with Topology checks tool and ty ArcGis software extension - DataReviewer, was used for the analysis of quality of elements concerning g logical consistency. The results are depicted on Table 4 and on the diagram on graph 2. H Table 4: Analysis of the road infrastructure logical consistency. Dataset layer Total number of objects in universe of discourse Data Quality Measure (data type value) Must not have pseudonode (integer) Must not selfintersect (integer) Must not selfoverlap (integer) Must not overlap (integer) Must not intersect (integer) Must be single part (integer) C O M M U N I C A T I O N Objects on communication 1 4,198 4 5 3 7 4 2 Objects on communication 2 511 1 4 5 7 2 12 Objects on communication 3 97 5 9 13 6 11 14 Roads 259,480 76 47 25 36 67 78 Objects on comm._1_Anno 1,105 0 0 1 0 0 0 Roads_Anno 52,171 2 3 2 0 0 0 Objects on comm._2_Anno 264 2 1 1 0 1 1 Objects on comm._3_Anno 51 1 0 0 1 0 1 S 317,877 91 69 50 57 85 108 I 278 | Graph 2: Analysis of the road infrastructure logical consistency During the analysis of road infrastructure logical consistency, a total number of 317,877 elements were analysed. By means of topological rules a total of460 errors were found and they were automatically removed so that the thematic layer of road infrastructure could be used in other analyses. The relation between the number of analysed elements and the obtained topological errors showed a logical consistency of 99.86%. Sinisa Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | GEODETSKI VESTNIK | 60/2 | 4.3 Positional accuracy Knowledge of positional accuracy is of fundamental value both for the user of the map and for the producer. § Unlike other spatial data elements of quality, spatial accuracy consisting of a horizontal (planimetric) and 5 height accuracy can be completely tested and accurately quantified. The testing of the spatial accuracy is ^ done by comparing the specific points determined by reading the map with referent, far more accurate, coordinates of the same points obtained by land survey or retrieved from other accurate sources. The M main problem for the process of assessing spatial accuracy is the choice of measurements (i.e. the unit g for measurement and assessment of accuracy) as well as the appropriate set of points which represent a ^ certain sheet of topographic map and sheets that represent a whole map. (Drummond et al., 1995; Ozah et al., 2008; Govedarica et al., 2011; Amovic et al., 2015). In this analysis we used ArcGis software tool called PAAT (Positional accuracy assessment tool). While assessing the spatial accuracy this tool uses a starting point marked as a RMSE (Root Mean Squared Error). Error in RMSE is the second root mean value of the total number of square differences between the coordinates read from the map and the appropriate referent ("true") coordinates. Absolute horizontal accuracy represents an uncertainty of two dimensional positions (in relations to a horizontal datum) and is expressed in error tolerance range within the 90, 95 to 99% confidence level. Accuracy is expressed in the same units as those expressing coordinates in nature (meter) thus enabling the direct comparison of different products, regardless of differences in scale or resolution. PAAT tool can analyse ^ both horizontal and vertical components of positional accuracy. This paper only analyses the horizontal positional accuracy which is why the road infrastructure data is mapped with 2D base, digital orthophoto (ESRI, 2012). Relative horizontal accuracy represents uncertainty in difference of crossroad positions and the corresponding points collected from the referent base of digital orthophoto and it is expressed in error tolerance range within 90, 95 to 99% confidence level. The horizontal accuracy of the digital orthophoto was known and its maximum value is 0.8 m. For the tested example a total number of 6,000 characteristic points with clearly seen crossroads of different road categories was analysed and this can be seen on Figure 6. Figure 6: An example of logical consistency - verification of topological rules. Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 279 | 60/2 | GEODETSKI VESTNIK PAAT tool requires that the mean square errors should be computed by y and x axis accordingly (1): 11 n 1 n RMSEy = J1Z dy2 and RMSEx = 11 £ dx2 (1) Where: dy. = (y — y , ). is the difference of coordinates by axis y, J i v map J reference' i JJ *> J J dx. = (x — x . ). is the difference of coordinates by axis x, i map reference i (y , x ) are point coordinates measured on the map, map map (y , , x , ) are referent point coordinates ("true"). reference reference Then the equation for the root mean square error of horizontal position of point i is (2): RMSE, =y/ dy] + dx] (2) and the mean value of root mean square error of the horizontal position of the selected sample is (3): 1 n 1 n -- RMSEr = -Y dy2 +-Y dx2 = J RMSE2 + RMSE] (3) \n ! n ! v PAAT has the ability of the automatic testing and elimination of gross errors. A testing statistic called 3a threshold is used for this. If a specific positional error is greater than the value of 3ct, the program eliminates it, leaving the possibility to keep these points if we wish to do that (ESRI, 2012). In the procedure of quality assessment of positional accuracy, we used PAAT tool. The type of value used is RMSE while the unit of quality value is meter (m). While analysing horizontal positional accuracy we obtained the following results: — Absolute horizontal circular error with 90% confidence level is 4.69 m, relative horizontal circular error is 5.34 m, root mean square error in the range with 90% confidence level is 3.82 m while the standard deviation is 1.83. — Absolute horizontal circular error in the error range with 95% confidence level is 5.02 m, relative horizontal circular error is 5.87 m, root mean square error is 3.82 m while the standard deviation is 1.83. — Absolute horizontal circular error in the error range with 99% confidence level is 5.25 m, relative horizontal circular error is 6.71 m, root mean square error is 3.82 m while standard deviation is 1.83. 4.4 Thematic (semantic) accuracy Thematic accuracy represents a degree of consistency between quantitative attributes and the correctness of non-quantitative attributes, classification of objects and their relations. Sub-elements of thematic accuracy are (Harding, 2006): — Correctness of classification — comparison of classes assigned to objects or their attributes with the area of analysis; — Correctness of qualitative attributes; — Accuracy of quantitative attribute values. Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA | 284 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 280-284 | GEODETSKI VESTNIK Figure 7 shows an example of error done in thematic accuracy analysis. The shown error refers to a road being classified as a gravel road while the referent data of importance classified the road as a modern roadway (asphalt road). Figure 7: Analysis of thematic accuracy (asphalted - left, gravel road - right). The process of acquisition of thematic accuracy is sometimes similar to logical consistency if we consider the difference in conceptual modelling transformed to the attribute of class and vice versa. Similar to that, positional accuracy becomes a type of semantic accuracy when we treat the object location as a specific entity attribute. What is identified and used in assessment of thematic accuracy is the misclassification matrix which is a quadrant matrix with n columns and n lines where n marks the number of classes discussed: MCM (i,j) = (# class objects (i) classified as class (j)) (4) Its diagonal elements contain wrongly classified objects in the frame of one theme while non-diagonal elements contain number of errors of wrong classification where spatial objects are wrongly classified. Misclassification matrix of spatial objects represents one of the quantitative results of DTM thematic accuracy quality assessment. It is depicted in Table 5. With the usage of misclassification matrix, we get a coefficient of kappa statistics which is a good indicator of the choice of classification method consistency taking their randomness into account. Kappa coefficient (k) represents a coefficient which expresses a degree of compatibility between assigned classes by removing the misclassification and it is calculated based on the following formula (ISO 19157, 2011): N K = - ■z r=McM (i, ; )-z (z j-imcm (i, ; >z ;=imcm ( ;, i )) N2 •z ;= (z j mcm (i, j )-z ; mcm ( j , i )) (5) The high value of kappa coefficient (k = 0.9466) shows that the method of classification is good and highly efficient. Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | I 281 | | 60/2 | GEODETSKI VESTNIK Table 5: Misclassification Matrix of thematic road layer. Spatial Dataset Universe of discourse i w. it 13 1 5 JS S ! 2 -ü IS S 0 u « s V TS o ¡s "fl O Si o T3 « 0 t- t« ■s o £ CO -ü « 0 tS "s o 2 -ü « o t- "s u £ o ¿2 Highway 12 0 0 0 0 0 0 0 Semi-highway 0 32 2 7 0 12 0 0 Modern road 0 2 45 3 0 6 0 0 Modernized roadway 0 1 0 14 0 2 0 0 Gravel road 0 0 0 0 85 2 0 0 Better vehicular road 0 2 5 2 0 145 0 0 Vehicular road 0 0 0 0 0 0 3 0 Poorer vehicular road 0 0 00003 45 Sum 12 37 52 26 85 167 6 45 4.5 Temporal accuracy The date of data input or the date of its update is a very important factor for the user while assessing the data quality. Temporal accuracy refers to the date of acquisition, the type of change and the validity date of spatial data, i.e. temporal accuracy represents a degree of consistency between time attributes and time relations of objects. Manipulation of time data comes down to addition of time dimension to the data model and by that to all elements of the database, for example, with the help of one or more added "attributes" for each database entity, each attribute and each relation among objects. Sub-elements of temporal accuracy are (Devillers et al., 2006): — Accuracy oftime measurement (correctness oftime references — display oferrors in time measurement), — Temporal consistency (correctness of sequence of events if they exist), — Temporal validity (expiration date). For the assessment procedure of road infrastructure temporal accuracy, we used the creation date of the main cartographic source and that is, in this testing sample, a digital orthophoto with the date of acquisition from July 2009. Data quality value for the temporal accuracy of road infrastructure is date, while the unit of quality is year. The temporal aspect of the CGD is realized by including tools Enable Editor Tracking for monitoring changes of resulting spatial objects. The database can then generate special four fields that define the user who first mapped the specific object, the time when the listed building was mapped, as well as the user and time of the last change to the specified spatial object. The obtained temporal validity values for all objects of road infrastructure are in the range START_DATE = "2012-12-02", END_DATE = "2015-05-15". Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA | 284 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 282-284 | GEODETSKI VESTNIK | 60/2 | 5 CONCLUSION Spatial data have an increasing significance in the process of decision making, situation follow-up and planning in all segments of life and development. In these circumstances, the importance of knowledge of data quality is increasing as well as the need for quality indicators to be established and presented in a standardized way. While analysing the quality of CGD road infrastructure according to ISO standards we could conclude that it meets all international standards. Besides spatial accuracy, the goal of this research was to analyse and present the verification of geometrical connection of elements of road infrastructure in geospatial database, so the database would gain a required utility value. This is accomplished by creation of topological rules, but the correction of the content comes down to controlling the compliance with the rules. Topological rules refer to the elimination of double lines, connectivity of lines and points, control of the connected junctions, elimination of line breaks, compatibility control of joint border lines of polygonal element contours etc. Cartographic correction of CGD road infrastructure, compared to traditional cartographic correction has an additional control function of input attributes in the associated database. This additionally complicates the process of cartographic correction so it implies the correction of visual errors created in the previous process of mapping as well as the semantic consistency in respect to following the rules of the logical data model. In this form of correction, a special attention should be paid because even though some elements have an identical visual identity, they sometimes may have a completely different structure of the attribute in the geospatial database. Literature and references: Amovic, M., Vasiljevic, S., Sekulovic, D. (2015). Assessment of Positional Accuracy of Topographic Maps and Plans of Banja Luka (Bosnia and Herzegovina) of Different Editions. In FIG Working week 2015, From the Wisdom of the Ages to the Challenges of the Modern World, 17-21 May, Sofia, Bulgaria. Amovic, M., Vasiljevic, S., Sekulovic, D. (2015). Assessment of Positional Accuracy on the historical review of topographic maps and plans of Banja Luka Region (Bosnia and Herzegovina) of different editions. In 27th International Cartographic Conference, Maps Connecting the World, August 23-28, Rio de Janeiro, Brasil. Bozic B., Radojcic S. (2011). Horizontal accuracy of 1:50 000 digital topographic maps. Survey Review, 43 (319), 94-104. DOI: http://dx.doi.org/10.1179/00 3962611x12894696204740 Docan, D. C. (2013). Spatial data quality assessment in GIS. In 1st European Conference of Geodesy & Geomatics Engineering, GENG2013 Conference Proceedings, WSEAS Press: Antalya, Turkey. Devillers, R., Jeansoulin, R. (2006). Spatial Data Quality: Concepts, In R. Devillers, R. Jeansoulin (Eds.), Fundamentals of Spatial Data Quality. London, ISTE, pp. 31-42. DOI: http://dx.doi.org/10.1002/9780470612156.ch2 Drobnjak, S., Radojcic, S., Bozic, B. (2014). Primena ISO 19157 standarda u tehnoloskom procesu izrade digitalnih topografskih karata. Tehnika, 66 (4), 551-556. Drummond, J. (1995). Positional accuracy. In S. C. Guptill, J. L. Morrison (Eds.), Elements of Spatial Data Quality. Elsevier Science Ltd, pp. 31-58. DOI: http:// dx.doi.org/10.1016/b978-0-08-042432-3.50010-0 ESRI ArcGis 10.2.2, (2014). Positional Accuracy Assessment Tool Help, 380 New York Street, Redlands, CA 92373-8100, USA. Fisher, PF., Comber, A.J. and Wadsworth, R.A. (2006). Approaches to uncertainty in spatial data. In R. Devillers and R. Jeansoulin (Eds.), Fundamentals of Spatial Data Quality. London: ISTE, pp. 43-59. DOI: http://dx.doi. org/10.1002/9780470612156.ch3 Govedarica, M., Borisov, M. (2011). The Analysis of the Data Quality on Topographic Maps. Geodetski vestnik, 55 (4), 713-725. DOI: http://dx.doi.org/10.15292/ geodetski-vestnik.2011.04.713-725 Goodchild, M.F. (1993). Data Models and Data Quality: Problems and Prospects. In M. F.Goodchild, B. 0. Parks, L. T. Steyaert (Eds.), Environmental Modelling with GIS. Oxford, New York, pp. 94-103. Harding, J. (2006). Vector Data Quality: A Data Provider's Perspective, In R. Devillers, R. Jeansoulin (Eds.), Fundamentals of Spatial Data Quality. London: ISTE, pp. 141-159. DOI : http://dx.doi.org/10.1002/9780470612156.ch8 ISO (19157).The International Organization for Standardization, ISO 19157: Geographic information - Data Quality; Geneva. Ozah, A. P, Kufoniyi, O. (2008). Accuracy Assessment of Contour Interpolation from 1:50 000 Topographical Maps and SRTM Data for 1:25 000 Topographical Mapping. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXXVII (B7), 1347-1354. Petrovič, D. (2006). Ocena kakovosti državne topografske karte v merilu 1:50 000. Geodetski vestnik, 50 (2), 187-200. Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 | | 283 | 60/2 | GEODETSKI VESTNIK Radojčic, S., Borisov, M., Božič, B. (2011). Položajna tačnost geografskih informacija i savremeni standardi za njeno ocenjivanje. Glasnik Srpskog geografskog društva, 91 (2), 99-112. DOI: http://dx.doi.org/10.2298/gsgd1102099r Siva Kumar, R. (2000). Data integration and accuracy issues in digital topographic databases. International Archives ofPhotogrammetry and Remote Sensing, XXXIII (B4), 848-854. Sekulovic', D., Drobnjak, S. (2011). Primena savremenih tehnologija u procesu izrade geoprostorne baze podataka u rezoluciji 1:25 000 (GBP25). In Medunarodno -naučno stručni skup:„Arhitektura i urbanizam, Gradevinarstvo, Geodezija - Juče, Danas, Sutra", Zbornik radova (CD), Banja Luka, BiH, pp. 755-766. Servigne S., Lesage N., Libourel T. (2006). Quality Components, Standards, and Metadata. In R. Devillers, R. Jeansoulin (Eds.), Fundamentals of Spatial Data Quality. London, ISTE, pp. 179-208. DOI: http://dx.doi. org/10.1002/9780470612156.ch10 Shekhar S., Ma X., Xiong H., Wu W., Donath M., Cheng P. (2002). Digital Road Map Accuracy Evaluation: A Buffer-based Approach. In W. Wu (Ed.), Digital Road Map Accuracy Evaluation: A EuGer-based Approach. Department of computer Science, University of Minnesota. Talhofer, V., Hofmann, A., Hoskova-Mayerova, Š., Kubiček, P. (2011). Spatial analyses and spatial data quality. In Proceedings of the 14th AGILE Internationa Conference on Geographic Information Science-Advancing Geoinformatior Science for a Changing World, AGILE. Zhang, X., Stoter, J., Ai, T., Kraak, M. J., Molenaar, M. (2013). Automated evaluation of building alignments in generalized maps. International Journal of Geographica Information Science, 27 (8), 1550-1571. DOI: http://dx.doi.org/10.1080/136 58816.2012.758264 Drobnjak S., Sekulovic D., Amovic M., Gigovic L., Regodič M. (2016). Central geospatial database analysis of the quality of road infrastructure data. Geodetski vestnik, 60 (2): 270-285. DOI: 10.15292/geodetski-vestnik.2016.02.269-284 Assist. Siniša Drobnjak, M.Sc. Military Geographical institute Mije Kovačeviča 5 SRB-H000 Belgrade, Serbia e-mail: sdrobnjak8l@gmail.com Full prof. Dragoljub Sekulovic, Ph.D. University of Defence, Military Academy Pavla Jurišiča Šturma 33 SRB-II000 Belgrade, Serbia e-mail: sekulovicdr@yahoo.co.uk Assist. Mladen Amovic, M.Sc. University of Banja Luka, Faculty of Architecture, Civil Engineering and Geodesy Bulevar Vojvode Petra Bojoviča IA BiH-78000, Banja Luka, Bosna i Herzegovina e-mail: mladen.amovic@hotmail.com Assist.prof. Ljubomir Gigovic, Ph.D. University of Defence, Military Academy Pavla Jurisica Sturma 33 SRB-H000 Belgrade, Serbia e-mail: gigoviclj@gmail.com Assoc. prof. Miodrag Regodic, Ph.D. University of Defence, Military Academy Pavla Jurisica Sturma 33 SRB-H000 Belgrade, Serbia e-mail: mregodic62@gmail.com Siniša Drobnjak, Dragoljub Sekulovic, Mladen Amovic, Ljubomir Gigovic, Miodrag Regodic | ANALIZA KAKOVOSTI PODATKOV CESTNE INFRASTRUKTURE KOT DELA | 284 | CENTRALNE GEOPROSTORSKE PODATKOVNE ZBIRKE | CENTRAL GEOSPATIAL DATABASE ANALYSIS OF THE QUALITY OF ROAD INFRASTRUCTURE DATA | 269-284 |