63 Les/Wood, Vol. 73, No. 2, December 2024 ASSESSING CLIMATE-GROWTH RELATIONSHIPS WITH DAILY AND MONTHLY OBSERVATIONAL AND GRIDDED METEOROLOGICAL DATA PRIMERJAVA KORELACIJ ŠIRIN BRANIK Z DNEVNIMI IN MESEČNIMI IZMERJENIMI IN MODELIRANIMI METEOROLOŠKIMI PODATKI Nina Škrk Dolar 1* , Katarina Čufar 1 , Jernej Jevšenak ² UDK članka: UDK 630*111:811.4 Received / Prispelo: 4.11.2024 Original scientific article / Izvirni znanstveni članek Accepted / Sprejeto: 19.11.2024 . Abstract / Izvleček Abstract: We compared climate-growth relationships by correlating tree-ring variation with daily and monthly meteorological data obtained from the stations of the Slovenian Environment Agency (ARSO) and modelled data from the SLOCLIM database. Tree-ring width series for analyses were obtained from previously collected European beech (Fagus sylvatica) tree-ring data from 30 sites all over Slovenia. Climate-growth correlations were calculated to evaluate whether daily meteorological data exhibits stronger correlations than monthly data. We also compared the maximum correlation coefficients using meteorological station data and gridded SLOCLIM data. The analysis was conducted using the dendroTools R package, incorporating data on daily and monthly average air temperatures and precipitation sums from the period 1960–2018. Our findings revealed significantly higher maximum correlation coefficients for daily data compared to monthly data, underscoring the importance of using daily data, particularly for precipitation. However, no significant difference was observed between maximum correlation coefficients using the meteorological station and modelled data, and the difference did not change significantly with increasing altitude. Keywords: observational data, gridded data, tree rings, correlation analysis, dendroclimatology Izvleček: V raziskavi smo primerjali korelacije med širinami branik in dnevnimi oziroma mesečnimi meteorološkimi podatki, pridobljenimi iz meteoroloških postaj (ARSO) ali iz modelirane baze SLOCLIM. Uporabili smo podatke o dnevnih in mesečnih povprečnih temperaturah zraka in vsotah padavin za obdobje 1960–2018. V analize smo vključili 30 kronologij širin branik navadne bukve (Fagus sylvatica) iz celotne Slovenije. Raziskali smo tudi, kako na korelacije vpliva uporaba podatkov iz meteoroloških postaj ali iz baze SLOCLIM. Naše ugotovitve so pokazale značilno višje maksimalne korelacijske koeficiente, ko smo uporabili dnevne meteorološke podatke, kot če smo uporabili mesečne. Glede na to je priporočljiva uporaba dnevnih podatkov v dendroklimatoloških analizah, zlasti pri padavinah. Pri primerjavi korelacij s podatki iz meteoroloških postaj in modeliranimi podatki nismo ugotovili statistično značilnih razlik. Razlike med uporabo dnevnih in mesečnih podatkov ter podatkov iz dveh baz se z nadmorsko višino rastišč bukve niso značilno spreminjale. Ključne besede: meteorološki podatki, modelirani podatki, branike, korelacijska analiza, dendroklimatologija 1 INTRODUCTION 1 UVOD The use of meteorological data to explain var- iations in tree-ring widths has long been a corner- stone of dendrochronology, with traditional anal- yses relying on monthly data. However, over the past decade, authors have begun to emphasize the importance of daily data (Beck et al., 2013; Liang et al., 2013; Pritzkow et al., 2014), which has led to the increasing use of daily meteorological data- sets (Castagneri et al., 2015; Kaczka et al., 2017). Daily data better captures weather hazards which last several days such as heatwaves, heavy rainfall events or sudden frost, which significantly affect the formation of tree rings. These effects can re- main obscured in monthly averages. This is espe- Vol. 73, No. 2, 63-74 DOI: https://doi.org/10.26614/les-wood.2024.v73n02a06 1 Univerza v Ljubljani, Biotehniška fakulteta, Oddelek za lesarstvo, Jamnikarjeva 101, 1000 Ljubljana, Slovenija ² Gozdarski inštitut Slovenije, Oddelek za načrtovanje in monitoring gozdov in krajine, Večna pot 2, 1000 Ljubljana, Slovenija * e-mail: nina.skrk@bf.uni-lj.si 64 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki cially relevant for precipitation data, which has an unpredictable occurrence pattern with low auto- correlation. Daily meteorological data also have a stronger relation to phenological phases (Kaczka et al., 2017), as the growth of a tree is not confined to the first and last day of a calendar month (Na- gavciuc et al., 2019). Recently, a study by Jevšenak (2019) examined the question of the importance of using daily data in tree-ring analysis, and report- ed that the correlation coefficients calculated us- ing the daily data compared to monthly data are on average by 0.076 higher for precipitation and 0.060 for air temperature data. However, these dif- ferences were generally not statistically significant, although still important with regard to capturing stronger climate signals. Gridded meteorological series with daily res- olution are less common and not available in all regions worldwide. However, high resolution grid- ded climate datasets have already been developed regionally, e.g. for Spain (Serrano-Notivoli et al., 2017, 2019), Norway (Lussana et al., 2018), Af- rica (Chaney et al., 2014) and certain large areas, e.g. E-OBS, ERA5, CHELSA, and the Berkeley Earth climate dataset. E-OBS is a commonly used daily gridded land-only observational dataset covering Europe, with a spatial resolution of 0.1° x 0.1° and 0.25° x 0.25° (Haylock et al., 2008). ERA5 is a glob- al climate reanalysis dataset providing hourly esti- mates of atmospheric, land-surface and sea-state parameters from 1940, with a spatial resolution of 0.25° x 0.25° (or 0.1° for ERA5-Land) (Coperni- cus Climate Change Service, 2023). CHELSA is also global downscaled climate dataset with climate layers for various time periods and variables and a spatial resolution of 1 km (Karger et al., 2017). The Berkeley Earth climate dataset provides high-reso- lution land and ocean time series data and gridded temperature data from 1850 (Rohde & Hausfather, 2019). These datasets provide meteorological data for evenly spaced locations across a defined area, overcoming the limitations of station-based data- sets, which are sometimes incomplete and have low spatial resolution. In Slovenia, the Slovenian Environment Agency (ARSO) provides nationwide daily meteorological data collected from meteorological stations, with the earliest records dating back to 1850 (Nadbath, 2015). The distribution of these stations is uneven, with a higher density found in urban areas and at lower altitudes, while more forested regions have limited coverage. In recent years, there has been a significant decline in the number of traditional meteorological stations with observers, while the number of automatic meteorological stations has increased (Nadbath, 2015). All the data is being regularly validated and is publicly available. Al- though such data is of great value, there are still some downsides. For example, the data provid- ed by the stations is of varying length and often partly incomplete. Moreover, many stations have been relocated in the past, resulting in altitudinal variations and corresponding climate discrepan- cies. However, these issues have been resolved through the homogenization of temperature and precipitation data, which can be obtained upon request. For dendrochronological studies, access to a comprehensive and consistent dataset is es- sential to ensure accurate climate reconstructions or climate-related dendroecological investigations. The SLOCLIM – Slovenian Modelled Climate Da- tabase was created (Škrk et al., 2021) to address these challenges, using available measurement data from ARSO, incorporating latitude, longitude, altitude and distance from the coast. Generalized linear mixed models (GLMMs) and generalized lin- ear models (GLMs) were applied in the calculation process. The dataset has a spatial resolution of 1 x 1 km and contains daily data on maximum and minimum air temperatures, as well as precipitation amounts for the period 1950-2018. It offers mod- elled local meteorological data, ensuring continu- ous coverage without missing values. The aim of this study was to evaluate wheth- er daily meteorological data better explains the climate-growth relationship compared to month- ly data, as daily data offers more precise weather information and is less constrained by temporal averaging. Our first hypothesis was that daily me- teorological data will yield significantly stronger cli- mate-growth correlations than monthly data, and differences will be greater for precipitation data. To test this, we employed selected sites from a Slove- nian tree-ring database of European beech (Fagus sylvatica) (Čufar et al., 2008a; Dolar et al., 2023), the predominant tree species in Slovenia, compris- ing approximately 33% of the country’s wood stock (Skudnik et al., 2021). The extensive coverage of 65 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Assessing climate-growth relationships with daily and monthly observational and gridded meteorological data tree-ring data from beech across Slovenia also pro- vides a robust foundation for our analysis. Additionally, we sought to compare data de- rived from traditional meteorological stations with that obtained from a gridded dataset, i.e. SLOCLIM, which feature higher spatial resolution and more comprehensive data. We hypothesized that grid- ded meteorological data for a given site explain the influence of weather conditions on the varia- tion of tree-ring widths better than the commonly used data from meteorological stations (Hypothe- sis 2). Notably, there is a lack of studies that have systematically compared data from meteorological stations with gridded datasets in the context of dendrochronological research. This comparison is critical for understanding how different data sourc- es influence dendrochronological analysis. We further hypothesized that at higher al- titudes the differences in maximum correlation coefficients between gridded and observational meteorological data, as well as between daily and monthly data, would be more pronounced (Hy- pothesis 3). This is attributed to the shorter grow- ing season at higher altitudes, making daily data more significant than monthly data. Additionally, meteorological stations are predominantly located at lower altitudes, with fewer stations at higher al- titudes, and we thus expected that the altitudinal correction of meteorological data should reveal stronger climate-growth effects. 2 MATERIALS AND METHODS 2 MATERIALI IN METODE 2.1 TREE-RING CHRONOLOGIES 2.1 KRONOLOGIJE ŠIRIN BRANIK Tree-ring data of selected European beech (Fa- gus sylvatica) sites in Slovenia from the collection of the Department of Wood Science and Technology, Biotechnical Faculty, University of Ljubljana (Čufar et al., 2008a; Dolar et al., 2023) were used in this study. The data contains raw tree-ring measure- ments from 30 sites across Slovenia with altitudes ranging between 230 and 1330 m a.s.l. (Figure 1, Table 1). Sites are located in three different climatic zones: Subcontinental, Subalpine, and Sub-Medi- terranean. For each site, we calculated first-order autocorrelation (AR1), Gleichläufigkeit coefficient (%GLK) and mean interseries correlation (rbar) (Ta- ble 2). AR1 represents the impact of growth from the previous to the current year (Fritts, 1976). High positive values close to 1 indicate strong positive Figure 1. Selected European beech (Fagus sylvatica) sites for tree-ring analyses in Slovenia with respective altitudes in metres above sea level. Slika 1. Izbrane lokacije s kronologijami širin branik navadne bukve (Fagus sylvatica) in njihove nadmorske višine. 66 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki Table 1. Basic site information for chronologies: altitude (m a.s.l.) (ALT), latitude (LAT), longitude (LON), ARSO nearest meteorological station (ARSO MS), mean annual air temperature (TAVG), annual precipita- tion sum (PCP). ARSO – data from Slovenian Environment Agency; SLOCLIM (Škrk et al., 2021). Meteoro- logical data is calculated for the period 1960-2018. Preglednica 1. Osnovni podatki o rastiščih: nadmorska višina (m) (ALT), zemljepisna širina (LAT), zemljepi- sna dolžina (LON), kronologiji najbližja ARSO meteorološka postaja (ARSO MS), povprečna letna tempe- ratura zraka (TAVG), letna količina padavin (PCP). ARSO – podatki Agencije Republike Slovenije za okolje; SLOCLIM (Škrk et al., 2021). Meteorološki podatki so izračunani za obdobje 1960–2018. Code / Oznaka ALT LAT LON ARSO MS TAVG (ARSO) [°C] PCP (ARSO) [mm] TAVG (SLOCLIM) [°C] PCP (SLOCLIM) [mm] CEA 450 46.1 15.5 BIZELJSKO 10.4 1031 10.2 955 CEB 450 46.1 15.4 MALKOVEC 10.4 1126 9.5 1033 CRO 1000 45.7 15.0 KOČEVJE 8.8 1502 8.5 1331 DOB 1000 46.4 14.3 BRNIK–LETA- LIŠČE 9.0 1344 9.0 1515 DRA 750 45.6 14.7 BABNO POLJE 6.5 1688 7.9 1569 GOR 450 45.8 15.3 MALKOVEC 10.4 1126 7.9 1293 KAF 950 46.5 14.1 KREDARICA -1.1 2025 6.8 1656 KLI 531 45.6 15.0 KOČEVJE 8.8 1502 9.2 1373 KRE 568 45.6 14.8 KOČEVJE 8.8 1502 8.3 1524 MOK 400 45.9 15.2 MALKOVEC 10.4 1126 9.6 1068 PA2 380 46.0 14.7 LJUBLJANA– BEŽIGRAD 10.6 1390 9.6 1215 POS 640 45.8 14.2 POSTOJNA 9.1 1557 9.6 1449 SNE 1100 45.6 14.4 BABNO POLJE 6.5 1688 7.3 1709 TOA 355 46.2 13.7 KREDARICA -1.1 2025 11.0 1991 TOB 821 46.2 13.8 KREDARICA -1.1 2025 8.2 2015 TOC 1328 46.2 13.8 KREDARICA -1.1 2025 7.0 2010 VAN 1000 46.2 14.2 VOJSKO 6.6 2379 7.9 1879 KOZ 380 45.6 13.9 GODNJE 11.3 1411 11.1 1167 LAZ 650 45.7 14.1 POSTOJNA 9.1 1557 9.3 1382 POT 690 45.7 14.0 POSTOJNA 9.1 1557 9.2 1401 POV 500 45.7 13.9 GODNJE 11.3 1411 10.7 1236 TON 590 45.6 14.1 POSTOJNA 9.1 1557 9.8 1269 VR1 570 45.8 14.0 POSTOJNA 9.1 1557 10.0 1353 VR2 615 45.8 14.0 POSTOJNA 9.1 1557 9.5 1383 VR3 690 45.7 14.1 POSTOJNA 9.1 1557 9.4 1395 VR4 638 45.7 14.1 POSTOJNA 9.1 1557 9.3 1382 VR5 650 45.7 14.1 POSTOJNA 9.1 1557 8.9 1404 VR6 515 45.8 13.9 GODNJE 11.3 1411 10.3 1398 ZT1 900 46.2 14.8 KRVAVEC 3.6 1476 6.8 1207 BOR 230 45.5 13.8 GODNJE 11.3 1411 12.0 1148 67 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Assessing climate-growth relationships with daily and monthly observational and gridded meteorological data autocorrelation, reflecting stronger lag effects re- lated to previous tree-ring growth. GLK is a non- parametric measure of growth similarity when comparing two tree-ring series (Speer, 2010). GLK can reach a maximum of 100, indicating a complete agreement between two series, while values above 60 are usually considered to indicate good agree- ment (Geijer et al., 2024). Rbar represents the aver- age correlation between individual tree-ring chro- nologies within each site (Cook & Kairiukstis, 2013), where higher values represent a stronger under- lying common signal. The raw tree-ring data was detrended using the detrend() function from the dplR package, using a spline with a 50% frequen- cy cutoff response at 32 years (Popa et al., 2024). We used pre-whitening to remove biological trends and temporal autocorrelation from each tree-ring width series. Finally, we built 30 site chronologies by robust bi-weight averaging of individual de- trended tree-ring width series, used in subsequent analyses. 2.2 METEOROLOGICAL DATA 2.2 METEOROLOŠKI PODATKI For each site, daily and monthly meteorolog- ical data was extracted based on the shortest dis- tance between the site and the location of mete- orological stations or grid points. For the modelled meteorological data, the SLOCLIM database was used (Škrk et al., 2021) for the period 1960-2018, extracting daily maximum (TMAX) and minimum air temperature (TMIN), and daily precipitation (PCP). The same daily data (TMAX, TMIN, PCP) were ex- tracted from the Slovenian Environment Agency’s observational database (ARSO, 2024), considering only the nearest meteorological stations that have a complete dataset for the period from 1960 to 2018. The period from 1960 to 2018 was chosen because it offers comprehensive meteorological data coverage from both ARSO and SLOCLIM. The mean air temperature (T) was calculated as the av- erage of the maximum and minimum temperature. The differences in monthly values between ARSO and SLOCLIM datasets for both temperature and precipitation were also computed. As shown in Figure 2, the average distance from the chronology site to the nearest meteorological station provided by the Slovenian Environment Agency (ARSO) was 15044 m (± 6405.2 m), whereas Table 2. Basic characteristics of the tree-ring width chronologies of European beech: first-order au- tocorrelation (AR1), Gleichläufigkeit coefficient (%GLK), mean interseries correlation (rbar). Preglednica 2. Osnovni podatki o izbranih kronolo- gijah navadne bukve: avtokorelacija prve stopnje (AR1), koeficient ujemanja Gleichläufigkeit (% GLK), drseča korelacija med zaporedji širin branik krono- logije (rbar). Code / Oznaka Number of trees / Število dreves First year / Prvo leto Last year / Končno leto %GLK rbar AR1 BOR 13 1939 2022 0.75 0.43 0.62 CEA 5 1840 2001 0.69 0.45 0.82 CEB 4 1883 2001 0.66 0.30 0.74 CRO 6 1831 2004 0.69 0.47 0.70 DOB 12 1840 2013 0.63 0.33 0.72 DRA 7 1752 2004 0.67 0.38 0.75 GOR 11 1830 2005 0.65 0.33 0.73 KAF 7 1859 2010 0.63 0.30 0.83 KLI 6 1845 2004 0.69 0.48 0.62 KOZ 13 1935 2022 0.69 0.28 0.68 KRE 7 1775 2004 0.61 0.29 0.80 LAZ 17 1921 2020 0.66 0.45 0.50 MOK 27 1854 2007 0.64 0.34 0.65 PA2 7 1911 2020 0.65 0.33 0.68 POS 8 1840 2007 0.63 0.24 0.74 POT 18 1886 2020 0.66 0.44 0.72 POV 13 1919 2022 0.69 0.37 0.67 SNE 24 1844 2008 0.60 0.25 0.72 TOA 7 1880 2001 0.70 0.29 0.76 TOB 4 1924 2001 0.77 0.43 0.74 TOC 10 1731 2001 0.66 0.32 0.74 TON 7 1911 2021 0.62 0.19 0.73 VAN 16 1927 2016 0.58 0.17 0.80 VR1 7 1914 2014 0.71 0.47 0.51 VR2 4 1928 2014 0.74 0.47 0.72 VR3 5 1927 2014 0.66 0.36 0.77 VR4 4 1889 2014 0.64 0.18 0.76 VR5 5 1872 2015 0.66 0.32 0.70 VR6 4 1910 2014 0.67 0.39 0.75 ZT1 10 1896 2021 0.68 0.43 0.68 68 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki the average distance to the modelled climate data from SLOCLIM was only 409 m (± 141.6 m). This indicates the often-discussed inadequate spatial coverage of meteorological datasets in contrast to gridded datasets, underscoring their limited abili- ty to accurately represent local climate conditions (e.g. Škrk et al., 2021). The average difference between the altitude of a chronology site and the altitude of the near- est meteorological station was 370 m (± 559.1 m), while the average difference with regard to the altitude of SLOCLIM point was 103 m (± 122.2 m) (Figure 3). 2.3 STATISTICAL ANALYSES 2.3 STATISTIČNA ANALIZA Climate-growth relationships were calculated using the daily_response() function for daily mete- orological data and monthly_response() for month- ly data from the dendroTools R package (Jevšenak & Levanič, 2018). Both functions work by sliding moving windows with variable lengths and calcu- lating the correlation coefficient between the ag- gregated meteorological data of interest and the selected tree-ring chronology. We performed these calculations for the growing season, spanning from March to September to cover the entire period of cambial production, which for European beech in Slovenia mainly starts in April and ends in August (Čufar et al., 2008b; Prislan et al., 2013, 2019). This time frame was chosen because it is presumed that climate has the greatest influence on growth dur- ing this period. For each site, the highest absolute Pearson correlation coefficient was extracted and used for the comparison, calculated separately for mean air temperatures and precipitation sums, and for meteorological observational ARSO and gridded SLOCLIM datasets. Pairwise statistical significance was assessed using the Wilcoxon-rank sum test (Wilcoxon, 1992). Figure 2. The boxplots of distances between site chronologies and the nearest meteorological sta- tion (ARSO) or gridded point (SLOCLIM). The red dots represent the average distance. Slika 2. Okvir z ročaji za razdalje med lokacijami kro- nologij in najbližjo meteorološko postajo (ARSO) ali točko na mreži modeliranih podatkov (SLOCLIM). Rdeči piki predstavljata povprečno razdaljo. Figure 3. Difference between the altitude of the chronology site and the altitude of the ARSO mete- orological station or gridded point (SLOCLIM). The red dots represent the average distance. Slika 3. Okvir z ročaji za razlike med nadmorskimi višinami lokacij kronologij in nadmorskimi višinami najbližjih meteoroloških postaj (ARSO) ali točke na mreži modeliranih podatkov (SLOCLIM). Rdeči piki predstavljata povprečno razliko. 69 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Assessing climate-growth relationships with daily and monthly observational and gridded meteorological data To test whether meteorological daily data exhibit significantly higher correlations with var- iations in tree-ring widths than monthly data, we compared daily and monthly data correlations with tree-ring widths for both meteorological stations (ARSO) and the gridded dataset (SLOCLIM), with separate calculations for precipitation and air tem- perature (Hypothesis 1). To test whether modelled meteorological data for a given site better explain the influence of climate conditions on tree-ring width variation than the commonly used meteor- ological station data, we compared ARSO and SLO- CLIM data for daily and monthly air temperature and precipitation (Hypothesis 2). Furthermore, this study aimed to investigate whether altitude has an influence on the magnitude of the difference in the maximum correlation coefficient between data from meteorological stations and gridded datasets. We also sought to determine if altitude impacts the magnitude of the difference in maximum correla- tion coefficients between monthly and daily data (Hypothesis 3). To address the altitude effects, we first calculated differences between daily and monthly correlations with tree-ring data and also differences between SLOCLIM and ARSO correla- tions with tree-ring data (separately for tempera- ture and precipitation variables), and afterwards fitted linear models where these differences are regressed as a function of altitude. 3 RESULTS AND DISCUSSION 3 REZULTATI IN DISKUSIJA When comparing climate-growth relationships using daily and monthly meteorological data, the results show significantly higher maximum correla- tion coefficients between tree-ring data and mete- orological data, if daily data of SLOCLIM or ARSO is used compared to monthly data for both the tem- perature and precipitation variables (Figure 4). The mean absolute maximum correlation coefficient with daily temperature was 0.30 for both ARSO and SLOCLIM (Figure 4). For daily precipitation, it was 0.44 for ARSO and 0.43 for SLOCLIM. For the monthly data, the absolute maximum correlation coefficient for temperature was 0.22 for ARSO and 0.21 for SLOCLIM, while for precipitation it was 0.34 for both ARSO and SLOCLIM. The correlations for temperature data were on average lower than the correlations in a study by Jevšenak (2019), where they were 0.47 for daily data and 0.41 for monthly data. Similarly, they were lower for precip- itation data. However, the data in Jevšenak (2019) mainly included conifers (82%), which are known to be generally more sensitive to the climate. The advantage of using daily data applies especially for precipitation, which was also confirmed in the aforementioned study. This is due to the higher au- tocorrelation of temperatures compared to precip- itation, and therefore the relative position of time windows for temperature is less important. The stronger correlation between daily meteorological data and tree-ring proxies, compared to monthly data, was also confirmed in a study by Nagavciuc et al. (2019). The comparative analysis revealed no signif- icant differences between daily and monthly data for SLOCLIM and ARSO data for either temperature or precipitation variables (Figure 4). Although the SLOCLIM database was developed on the basis of data from ARSO meteorological stations across Slo- venia (considering also altitude, latitude, longitude and distance to the coast), the result remains sur- prising. Given SLOCLIM’s higher spatial resolution, it was expected to more accurately reflect local climate conditions and their influence on tree-ring variations. Furthermore, the analysis of monthly temper- ature and precipitation differences between the ARSO and SLOCLIM databases revealed that the medians of the differences are consistently close to zero throughout the year (Figure 5). Nevertheless, in certain months (July to October), ARSO recorded lower temperature values than SLOCLIM in some years, suggesting occasional discrepancies during these months which could explain the lack of signif- icant differences between ARSO and SLOCLIM me- teorological data in climate-growth analysis. Linear regression was applied to describe the statistical relationship between altitude and differ- ences in absolute maximum climate-growth corre- lations between gridded data and meteorological station data, as well as between daily and monthly data. The results indicate that none of the depend- ent variables are significantly influenced by altitude (Table 3, Figure 6). 70 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki Figure 4. Boxplots of absolute maximum cor- relation coefficients, calculated between tree-ring data and daily or monthly meteoro- logical data, calculated separately for gridded data (SLOCLIM) and data from meteorolog- ical stations (ARSO). Horizontal lines indicate statistical significance of pairwise compari- sons. Significance lev- els: NS not significant at p >0.05; significant: *(0.01 < p ≤ 0.05); **(0.001 < p ≤ 0.01); ***(p ≤ 0.001). Slika 4. Okvirji z ročaji za prikaz vrednosti absolutnih maksimalnih korelacijskih koeficientov, izračunani med širinami branik in dnevnimi oziroma mesečnimi meteorološkimi podatki, ki so pridobljeni iz modeli- rane baze SLOCLIM ali iz meteoroloških postaj (ARSO). Ravne črte predstavljajo statistično značilnost pri parnih primerjavah. Stopnje značilnosti: NS ni značilno pri p >0.05; značilno: *(0.01 < p ≤ 0.05); **(0.001 < p ≤ 0.01); ***(p ≤ 0.001). Figure 5. Boxplots of monthly dif- ferences between ARSO and SLO- CLIM datasets for precipitation (PCP) and temperature (T). The difference is calculated as ARSO – SLOCLIM for the period 1960- 2018. Slika 5. Okvirji z ročaji za prikaz razlik v mesečnih vrednostih med meteorološkimi podatki SLOCLIM ali ARSO za padavine in povprečne temperature. Razlika je izračuna- na kot ARSO – SLOCLIM v obdobju 1960–2018. 71 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Assessing climate-growth relationships with daily and monthly observational and gridded meteorological data 4 CONCLUSIONS 4 ZAKLJUČEK In this study we compared the climate-growth relationship of European beech from selected sites in Slovenia, based on newly compiled tree-ring chronologies (with the recommended detrending procedure) and meteorological data from the me- teorological stations of the Slovenian Environment Agency (ARSO) and gridded data from the SLOCLIM database. Our results show that daily meteorolog- ical data provided a significantly higher maximum correlation coefficients with tree-ring data com- Figure 6. Difference in absolute maximum correla- tion coefficient between tree-ring data and mete- orological data when observational (ARSO) versus gridded data (SLOCLIM) is used for temperature (T) and precipitation (PCP) regarding altitude (A). Difference in maximum correlation coefficient between tree-ring data and meteorological data when daily versus monthly data is used from ARSO and SLOCLIM regarding altitude (B). Altitude (x-ax- is) refers to the elevation of the specific site. Slika 6. Razlike v maksimalnih korelacijskih koe- ficientih med širinami branik in meteorološkimi podatki, glede na to, ali uporabimo podatke iz me- teoroloških postaj (ARSO) ali modelirane podatke (SLOCLIM) pri temperaturah (T) in padavinah (PCP) glede na nadmorsko višino (A), ter ali uporabimo dnevne ali mesečne meteorološke podatke (B) gle- de na nadmorsko višino. Nadmorska višina (x os) se nanaša na nadmorsko višino posameznega rastišča. Table 3. The results of the linear regression analyses for the relationship between altitude (independent variable) and the difference (Diff) in absolute maximum correlation coefficient between tree-ring data and meteorological data when observational data (ARSO) versus gridded data (SLOCLIM) is used for tempera- ture (T) and precipitation (PCP), as well as difference in daily and monthly meteorological data from ARSO or SLOCLIM (dependent variables). The table presents the beta coefficient (β), standard error (SE), t-value, p-value, and the coefficient of determination (R²). Preglednica 3. Rezultati linearne regresije, kjer je neodvisna spremenljivka nadmorska višina, odvisne spremenljivke pa so razlike v maksimalnih korelacijskih koeficientih med širinami branik in dnevnimi ozi- roma mesečnimi meteorološkimi podatki ter podatki iz meteoroloških postaj (ARSO) ali baze SLOCLIM za temperature (T) in padavine (PCP). Beta – koeficient β, standardna napaka (SE), t-vrednost, p-vrednost in koeficient determinacije (R²). Dependent Variable / Odvisna spremenljivka Beta (β) SE t-value p-value R² Diff between ARSO and SLOCLIM in daily T 0.000 0.000 0.418 0.679 0.006 Diff between ARSO and SLOCLIM in daily PCP 0.000 0.000 0.543 0.591 0.010 Diff between ARSO and SLOCLIM in monthly T 0.000 0.000 0.339 0.738 0.004 Diff between ARSO and SLOCLIM in monthly PCP 0.000 0.000 0.160 0.874 0.001 Diff between daily and monthly data in ARSO T 0.000 0.000 -1.156 0.258 0.046 Diff between daily and monthly data in SLOCLIM T 0.000 0.000 -1.105 0.279 0.042 Diff between daily and monthly data in ARSO PCP 0.000 0.000 0.768 0.449 0.021 Diff between daily and monthly data in SLOCLIM PCP 0.000 0.000 0.352 0.728 0.004 72 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki pared to monthly data, which underlines the advis- ability of using daily data (if available) over monthly data in dendrochronological analyses. This advan- tage is particularly pronounced for precipitation, and agrees with results reported by other studies. Hypothesis 1 is thus supported. Using these particular tree-ring data and the statistical approach, no statistically significant dif- ference was found in correlations between tree- ring data and the meteorological data from me- teorological stations and the gridded data from SLOCLIM, with a high spatial resolution. This con- firms that the observations from ARSO, which were also used to interpolate SLOCLIM database, adequately reflect the actual climate variability in Slovenia as already discussed by Škrk et al. (2021). Boxplots of the differences in monthly meteorolog- ical data between the ARSO and SLOCLIM databas- es revealed that both datasets report similar values for precipitation and temperature, indicating that the datasets are largely consistent. We thus reject Hypothesis 2. The advantage of using daily data over month- ly data applies to both lower and higher altitudes. However, the difference in absolute maximum cli- mate-growth correlations between daily or month- ly data does not significantly vary with altitude. Furthermore, the difference between gridded and observational meteorological data remains fairly constant across different altitudes. As such, Hy- pothesis 3 is not supported. These results suggest the importance of fur- ther studies with a larger tree-ring network, possi- bly also including other tree species to gain a more comprehensive understanding of the benefits of daily and gridded meteorological data, particular- ly in regions with sparse meteorological station coverage, such as forested areas and higher alti- tudes, and in cases with many gaps in meteorolog- ical data. High spatial resolution of meteorological data is also crucial for the calibration of process models. When interpreting the climate-growth re- lationships, it should also be considered that the tree-ring width integrates the effect of climatic and other ecological influences, and that these reac- tions recorded in the tree-ring width cannot simply be described by the selected climate data alone. 5 SUMMARY 5 POVZETEK V dendrokronoloških študijah se za analizo vpliva podnebnih dejavnikov na variiranje širin branik tradicionalno uporabljajo mesečni meteo- rološki podatki. V mesečnih povprečjih pa pogosto ostanejo prezrti ekstremni vremenski dogodki, ki pomembno vplivajo na rast dreves in se v zadnjem času zaradi podnebnih sprememb pojavljajo pogo- steje. Zato se v zadnjih letih vedno bolj poudarja pomen uporabe dnevnih meteoroloških podatkov (Beck et al., 2013; Jevšenak, 2019; Liang et al., 2013; Pritzkow et al., 2014), ki lažje zajamejo tudi različne fenološke faze, ki jih beležimo v dnevnih intervalih. Ker meteorološke postaje, ki beležijo dnevne podatke, običajno niso gosto in enakomerno raz- porejene po prostoru, so bile razvite različne in- terpolirane in/ali modelirane podatkovne baze, ki nudijo tudi dnevne meteorološke podatke z visoko prostorsko ločljivostjo, tako na evropski (E-OBS) kot tudi svetovni ravni (CHELSA, ERA5). V Sloveniji ima- mo v okviru Agencije Republike Slovenije za okolje (ARSO) dobro razvito mrežo meteoroloških postaj. Število klasičnih postaj z opazovalci v zadnjih letih sicer upada, veča pa se število samodejnih postaj (Nadbath, 2015). Gostota teh postaj je manjša na gozdnatih in višje ležečih predelih, lokacije postaj pa so bile v preteklosti tudi večkrat spremenjene. Da bi zagotovili meteorološke podatke z visoko ča- sovno in prostorsko ločljivostjo, je bila razvita mo- delirana baza SLOCLIM (Škrk et al., 2021), ki smo jo uporabili tudi v tej raziskavi. Glavni cilj raziskave je bil preučiti, ali dnevni podatki bolje pojasnijo vpliv podnebnih razmer na variiranje širin branik kot doslej običajno uporablje- ni mesečni podatki (hipoteza 1). Poleg tega smo že- leli preveriti tudi, ali se korelacijski koeficienti med širinami branik in meteorološkimi podatki razlikuje- jo, če uporabimo podatke iz meteoroloških postaj ali iz baze SLOCLIM (hipoteza 2). Zanimalo nas je še, ali se te razlike značilno spreminjajo z nadmorsko višino (hipoteza 3). Uporabili smo 30 na novo sestavljenih krono- logij navadne bukve (Fagus sylvatica L.) iz celotne Slovenije z razponom nadmorskih višin med 230 in 1330 metri. Izračunali smo maksimalne absolu- tne korelacijske koeficiente med širinami branik in dnevnimi oziroma mesečnimi meteorološkimi po- 73 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Assessing climate-growth relationships with daily and monthly observational and gridded meteorological data datki iz meteoroloških postaj ARSO ali baze SLOC- LIM. Meteorološka postaja ARSO ali točka na mreži SLOCLIM je bila izbrana glede na oddaljenost od lokacije kronologije, pri čemer smo upoštevali naj- krajšo razdaljo. Pri analizah smo uporabili knjižnico dendroTools (Jevšenak & Levanič, 2018) s funkcija- ma daily_response() za dnevne meteorološke po- datke in monthly_response() za mesečne podatke. Analize so vključevale povprečne temperature zra- ka ter količino padavin za obdobje od 1960 do 2018. Rezultati so pokazali, da so maksimalni ko- relacijski koeficienti med širinami branik in me- teorološkimi podatki statistično značilno višji, če uporabimo dnevne, kot pa če uporabimo mesečne podatke. To potrjuje smiselnost uporabe dnevnih podatkov v dendroklimatoloških analizah. Korela- cijski koeficienti so bili višji tako pri temperaturah kot pri padavinah. Uporabo dnevnih podatkov še posebej priporočamo pri korelacijah širin branik s padavinami, kar je pokazal tudi Jevšenak (2019). Temperaturni podatki imajo namreč veliko stopnjo avtokorelacije, ki je padavine nimajo. Hipotezo 1 smo torej potrdili. V primerjavi z raziskavo, kjer so bili vključeni predvsem iglavci (Jevšenak, 2019), so bili maksimalni korelacijski koeficienti v naši študiji z bukvijo nekoliko nižji, verjeto zato, ker so iglavci na splošno občutljivejši na podnebne razmere. Čeprav je bila povprečna oddaljenost meteoro- loške postaje ARSO od lokacije s kronologijo približ- no 15 km, točke SLOCLIM, ki temeljijo na podatkih več okoliških postaj, pa le okoli 409 m, rezultati niso potrdili statistično značilnih razlik v maksimalnih korelacijskih koeficientih med širinami branik in meteorološkimi podatki iz dveh različnih baz podat- kov. Tudi analize razlik v mesečnih meteoroloških podatkih so pokazale, da ni večjih razlik med baza- ma tako pri padavinah kot temperaturah. Hipotezo 2 smo zato zavrnili. Z višanjem nadmorske višine se razlike med ko- relacijami s širinami branik glede na to, ali smo upo- rabili dnevne ali mesečne podatke, tako pri tempe- raturah kot tudi padavinah niso statistično značilno spreminjale. Prav tako se z višanjem nadmorske vi- šine niso značilno spreminjale razlike v korelacijah s kronologijami med bazama. Predvidevamo, da je med razlogi za to manjše število proučenih krono- logij z višjih nadmorskih višin, kjer bi pomanjkanje meteoroloških postaj lahko privedlo do večjih raz- lik. Hipotezo 3 smo prav tako zavrnili. Naši rezultati so potrdili pomembnost izbire načina izračuna korelacijskih koeficientov. Delno smo potrdili ugotovitve predhodnih študij, pokazali pa tudi, da so merjeni meteorološki podatki, čeprav oddaljeni od lokacij rastišč v Sloveniji, še vedno za- nesljiv vir za zajem variabilnosti podnebja kot de- javnika pri rasti bukve. Rezultati so pokazali nezna- čilen vpliv nadmorske višine na razlike v korelacijah. V prihodnjih raziskavah bi lahko analize razširili s širšim naborom kronologij bukve in drugih dreve- snih vrst ter tako pridobili še celovitejše razume- vanje pomena uporabe meteoroloških podatkov z visoko prostorsko in časovno ločljivostjo, še pose- bej na območjih, kjer je meteoroloških postaj malo, npr. v gozdnatih in višje ležečih predelih. Kljub temu je potrebno poudariti, da na širino branik vplivajo tudi ekološki in drugi individualni dejavniki in zato variacij med širinami branik ni možno pojasniti iz- ključno z izbranimi meteorološkimi podatki. ACKNOWLEDGEMENT ZAHVALA The research was financially supported by the Slovenian Research and Innovation Agency ARIS: Programme and Research Group “Wood and ligno- cellulose composites (P4-0015)” (NŠD, KČ), the young researchers’ programme (NŠD), and Pro- gramme and Research Group “Forest Biology, Ecol- ogy and Technology (P4-0107)” (JJ). REFERENCES VIRI Beck, W., Sanders, T. G. M., & Pofahl, U. (2013). CLIMTREG: detecting temporal changes in climate–growth reactions–a computer program using intra-annual daily and yearly moving time in- tervals of variable width. Dendrochronologia, 31(3), 232–241. Castagneri, D., Petit, G., & Carrer, M. (2015). Divergent climate re- sponse on hydraulic-related xylem anatomical traits of Picea abies along a 900-m altitudinal gradient. Tree Physiology, 35(12), 1378–1387. DOI: https://doi.org/10.1093/treephys/ tpv085 Chaney, N. W., Sheffield, J., Villarini, G., & Wood, E. F. (2014). De- velopment of a high-resolution gridded daily meteorological dataset over sub-Saharan Africa: Spatial analysis of trends in climate extremes. Journal of Climate, 27(15), 5815–5835. DOI: https://doi.org/10.1175/JCLI-D-13-00423.1 Cook, E. R., & Kairiukstis, L. A. (2013). Methods of dendrochronology: applications in the environmental sciences. Springer Science & Business Media. 74 Les/Wood, Vol. 73, No. 2, December 2024 Škrk Dolar, N., Čufar, K., & Jevšenak, J.: Primerjava korelacij širin branik z dnevnimi in mesečnimi izmerjenimi in modeliranimi meteorološkimi podatki Copernicus Climate Change Service, C. D. S. (2023). ERA5 hourly data on single levels from 1940 to present. URL: https://cds.climate. copernicus.eu/datasets/reanalysis-era5-single-levels?tab=o- verview (28.11.2024) Čufar, K., De Luis, M., Horvat, E., & Prislan, P . (2008a). Main patterns of variability in beech tree-ring chronologies from different sites in Slovenia and their relation to climate. Zbornik Gozdar- stva in Lesarstva, 87, 123–134. Čufar, K., Prislan, P ., & Gričar, J. (2008b). Cambial activity and wood formation in beech (Fagus sylvatica) during the 2006 growth season. Wood Research, 53, 1–11. Dolar, N. Š., Castillo, E. M. del, Serrano-Notivoli, R., Arrillaga, M. de L., Novak, K., Merela, M., & Čufar, K. (2023). Spatial and tem- poral variation of Fagus sylvatica growth in marginal areas un- der progressive climate change. Dendrochronologia, 81. DOI: https://doi.org/10.1016/j.dendro.2023.126135 Fritts, H. (1976). Tree rings and climate. Academic Press. Geijer, H., Ndongozi, F., & Edvardsson, J. (2024). Dendrochronology with a medical X-ray photon counting computed tomography scanner. Dendrochronologia, 86, 126233. DOI: https://doi. org/10.1016/j.dendro.2024.126233 Haylock, M. R., Hofstra, N., Klein Tank, A. M. G., Klok, E. J., Jones, P . D., & New, M. (2008). A European daily high-resolution gridded data set of surface temperature and precipitation for 1950- 2006. Journal of Geophysical Research Atmospheres, 113(20). DOI: https://doi.org/10.1029/2008JD010201 Jevšenak, J. (2019). Daily climate data reveal stronger climate-growth relationships for an extended European tree-ring network. Qu- aternary Science Reviews, 221. DOI: https://doi.org/10.1016/j. quascirev.2019.105868 Jevšenak, J., & Levanič, T. (2018). DendroTools: R package for studying linear and nonlinear responses between tree-rings and daily environmental data. Dendrochronologia, 48, 32–39. DOI: https://doi.org/10.1016/j.dendro.2018.01.005 Kaczka, R. J., Janecka, K., Hulist, A., & Spyt, B. (2017). Linking the growth/climate response of daily resolution with annual ring formation of Norway spruce in the Tatra Mountains. Trace, 15, 13–22. Karger, D. N., Conrad, O., Böhner, J., Kawohl, T., Kreft, H., Soria-Auza, R. W., Zimmermann, N. E., Linder, H. P ., & Kessler, M. (2017). Climatologies at high resolution for the earth’s land surface areas. Scientific Data, 4(1), 1–20. Liang, W., Heinrich, I., Simard, S., Helle, G., Liñán, I. D., & Heinken, T. (2013). Climate signals derived from cell anatomy of scots pine in NE Germany. Tree Physiology, 33(8), 833–844. DOI: https:// doi.org/10.1093/treephys/tpt059 Lussana, C., Tveito, O. E., & Uboldi, F. (2018). Three-dimensional spa- tial interpolation of 2 m temperature over Norway. Quarterly Journal of the Royal Meteorological Society, 144(711), 344– 364. DOI: https://doi.org/10.1002/qj.3208 Nadbath, M. (2015). Podnebna spremenljivost Slovenije v obdobju 1961-2011. Meteorološka opazovanja I. Ministrstvo za okolje in prostor, Agencija RS za okolje. Nagavciuc, V., Roibu, C. C., Ionita, M., Mursa, A., Cotos, M. G., & Popa, I. (2019). Different climate response of three tree ring proxies of Pinus sylvestris from the Eastern Carpathians, Ro- mania. Dendrochronologia, 54(October 2018), 56–63. DOI: https://doi.org/10.1016/j.dendro.2019.02.007 Popa, A., Jevšenak, J., Popa, I., Badea, O., & Buras, A. (2024). In pursuit of change: Divergent temporal shifts in climate sensi- tivity of Norway spruce along an elevational and continen- tality gradient in the Carpathians. Agricultural and Forest Meteorology, 358(July). DOI: https://doi.org/10.1016/j.agrfor- met.2024.110243 Prislan, P ., Gričar, J., Čufar, K., de Luis, M., Merela, M., & Rossi, S. (2019). Growing season and radial growth predicted for Fagus sylvatica under climate change. Climatic Change, 153(1–2), 181–197. DOI: https://doi.org/10.1007/s10584-019-02374-0 Prislan, P ., Gričar, J., de Luis, M., Smith, K. T., & Čufar, K. (2013). Phe- nological variation in xylem and phloem formation in Fagus sylvatica from two contrasting sites. Agricultural and Forest Meteorology, 180, 142–151. DOI: https://doi.org/10.1016/j. agrformet.2013.06.001 Pritzkow, C., Heinrich, I., Grudd, H., & Helle, G. (2014). Relationship between wood anatomy, tree-ring widths and wood density of Pinus sylvestris L. and climate at high latitudes in northern Sweden. Dendrochronologia, 32(4), 295–302. DOI: https://doi. org/10.1016/j.dendro.2014.07.003 Rohde, R., & Hausfather, Z. (2019). Berkeley Earth Combined Land and Ocean Temperature Field, Jan 1850-Nov 2019 [Data set]. Zenodo. DOI: https://doi.org/https://doi.org/10.5281/zeno- do.3634713 Serrano-Notivoli, R., Beguería, S., & de Luis, M. (2019). STEAD: A high-resolution daily gridded temperature dataset for Spain. Earth System Science Data Discussions, 1–27. DOI: https://doi. org/10.5194/essd-2019-52 Serrano-Notivoli, R., Beguería, S., Saz, M. Á., Longares, L. A., & de Luis, M. (2017). SPREAD: A high-resolution daily gridded preci- pitation dataset for Spain. Earth System Science Data Discussi- ons, 1–33. DOI: https://doi.org/10.5194/essd-2017-35 Škrk, N., Serrano-Notivoli, R., Čufar, K., Merela, M., Črepinšek, Z., Kajfež Bogataj, L., & de Luis, M. (2021). SLOCLIM: a high-reso- lution daily gridded precipitation and temperature dataset for Slovenia. Earth System Science Data, 13(7), 3577–3592. DOI: https://doi.org/10.5194/essd-13-3577-2021 Skudnik, M., Grah, A., Guček, M., Hladnik, D., Jevšenak, J., Kovač, M., Kušar, G., Mali, B., Pintar, A. M., Pisek, R., Planinšek, Š., Polja- nec, A., & Simončič, P . (2021). Stanje in spremembe slovenskih gozdov med letoma 2000 in 2018: rezultati velikoprostorskega monitoringa gozdov in gozdnih ekosistemov. Gozdarski inšti- tut Slovenije, Silva Slovenica. DOI: https://doi.org/10.20315/ SFS.181 Speer, J. H. (2010). Fundamentals of Tree-Ring Research. University of Arizona Press. Wilcoxon, F. (1992). Individual comparisons by ranking methods. In Breakthroughs in statistics: Methodology and distribution (pp. 196–202). Springer.