DIFFERENCES BETWEEN 100-METER AND 25-METER DIGITAL ELEVATION MODELS ACCORDING TO TYPES OF RELIEF IN SLOVENIA RAZLIKE MED STOMETRSKIM IN PETINDVAJSETMETRSKIM DIGITALNIM MODELOM VI[IN GLEDE NA TIPE RELIEFA V SLOVENIJI Mauro Hrvatin, Drago Perko Undulating low hills in the vicinity of Jeruzalem in Slovenske Gorice, northeastern Slovenia (photograph: Mimi Urbanc). Razgibano gri~evje v okolici Jeruzalema v Slovenskih goricah, severovzhodna Slovenija (fotografija: Mimi Urbanc). Differences between 100-meter and 25-meter digital elevation models according to types of relief in Slovenia UDC: 551.43:004.9(497.4) COBISS: 1.01 ABSTRACT: Studying relief and landscapes, we often employ digital elevation models. Their applicability is primarily linked to their accuracy. In this paper, we compare the older 100-meter digital elevation model of Slovenia and the more recent 25-meter digital elevation model. We assess applicability relative to differences in surface heights, surface slopes, and surface aspects for all of Slovenia and for four areas with different relief. We compare the frequency of distribution, arithmetic mean, standard deviation, range, and the coefficient of variation. The statistical analysis of the two digital elevation models of Slovenia indicates that the smallest differences exist relative to surface heights, there are larger differences relative to surface aspects, and the largest differences exist relative to surface slopes. KEYWORDS: relief, surface, digital elevation model, height, slope, aspect, Slovenia. The article was submitted for publication on January 27, 2005. ADDRESSES: Mauro Hrvatin, B. Sc. Anton Melik Geographical Institute Scientific Research Centre of the Slovenian Academy of Sciences and Arts Gosposka ulica 13, SI - 1000 Ljubljana, Slovenia E-mail: mauro@zrc-sazu.si Drago Perko, Ph.D. Anton Melik Geographical Institute Scientific Research Centre of the Slovenian Academy of Sciences and Arts Gosposka ulica 13, SI - 1000 Ljubljana, Slovenia E-mail: drago@zrc-sazu.si Contents 1 Introduction 2 Selected areas 3 Surface height 4 Surface slope 5 Surface aspect 6 Conclusion 10 11 15 15 16 22 23 7 References 1 Introduction Among all landscape elements, relief contributes the most to the external appearance of Slovenia's landscapes (Hrvatin and Perko 2002,2003). Given the rapid development of geographical information systems, its visual presentation or the digital elevation model is almost indispensable in modern geographical research. The applicability of digital elevation models for studying relief and landscape is primarily linked to their accuracy. Until the end of the 1990's, we in Slovenia had only 500-meter and 100-meter digital elevation models of the country's entire territory at our disposal. Both are based on the points of the square grid of the Gauß-Krüger coordinate system (Rihtar{i~ and Fras 1991). It is understandable that Slovene geographers mainly used the more accurate of the two models, the 100-meter digital elevation model, in their studies as did cartographers in creating thematic maps. The model is composed of data on altitude points that are 100 meters apart from north to south and from east to west, with the corners of the square cells having a 100-meter baseline, a 141-meter diagonal, and an area of one hectare (Perko 2001). The distance between the points, however, limits the model's applicability since relief features that are not substantially larger than 100 or 141 meters are completely lost or at least somewhat deformed. This failing is particularly significant in studies of landscapes with numerous small relief forms, such as those characteristic of the karst relief that covers almost one half of Slovenia. In 2000, we at the Scientific Research Center of the Slovenian Academy of Sciences and Arts created a 25-meter digital elevation model from radar pictures taken by the European Space Agency between 1995 and 1999 called the »InSAR DEM 25 radar interferometry digital elevation model« (Podobnikar and O{tir 1999; O{tir, Podobnikar, Stan~i~, and Mlinar 2000; Podobnikar 2002; Podobnikar 2005). The model is composed of data on altitude points that are 25 meters apart from north to south and from east to west, with the corners of the square cells having a 25-meter baseline, a 35-meter diagonal, and an area of 625 m2. Testing proved that the accuracy of the 100-meter digital elevation model is approximately 10 meters (3 meters for plains and 16 meters for mountains), while the accuracy of the 25-meter digital elevation model is approximately 5 meters (2 meters for plains and 14 meters for mountains); in both cases, the roughest errors in a few examples in the Alps exceeded 50 meters (Državna geodezija 1998, Digitalni modeli višin 2004). The ratio between the distance of the points in the 100-meter and the 25-meter digital elevation models is 4:1, and between the surface area of the square cells, 16:1. In this sense, the 25-meter digital elevation model is sixteen times more accurate that the 100-meter model. Of the three basic geometric properties of planes that we can establish in the framework of the geographical information system using a digital elevation model, that is, distance, inclination, and curvature relative to horizontal and vertical (Perko 2002), we selected the three properties or indicators for comparison that are the most frequently used in geography. These are: • surface height or the distance of surface relative to the horizontal, • surface slope or the inclination of surface relative to the horizontal, and • surface aspect or the inclination of surface relative to the vertical. Surface height is given in meters, surface slope in degrees from 0 for level surfaces to 90 for vertical surfaces, and surface aspect in degrees from 0 for north facing sites to 180 for south facing sites. We calculated these indicators for all of Slovenia and for four areas with different relief (rectangular sections) each nine kilometers long and six kilometers wide with a surface area of 54 km2. For calculating the indicators, we used the IDRISI program package (Eastman 1995). The established indicators from the 100-meter digital elevation model are based on 2,027,198 data items for all of Slovenia and on 5,400 data items for the individual areas, while those from the 25-meter digital elevation model are based on 32,436,693 data items for all of Slovenia and 86,400 data items for the individual areas. The areas (Figure 1), which we named after Mount Škrlatica in the Julian Alps, the Mirna River, a tributary of the Sava River, and the villages of Jeruzalem in Slovenske Gorice and Škocjan near the Škocjan Caves, are described briefly in Chapter 2. The following three chapters present the results of comparisons between the 25-meter and 100-meter digital elevation models of Slovenia and the selected areas relative to surface height, surface slope, and surface aspect. We compare the frequency of distribution, arithmetic mean, standard deviation, range (difference between highest and lowest values), and the coefficient of variation (ratio between standard deviation and arithmetic mean) and test the arithmetic means of surface heights, surface slopes, and surface aspects. Hereafter, »DEM-100« refers to the 100-meter digital elevation model and »DEM-25« to the 25-meter digital elevation model. 2 Selected areas The areas we selected for the comparison of DEM-100 and DEM-25 differ primarily relative to the morphological and genetic type of relief, the vertical and horizontal surface roughness, and the rock structure (Figure 1). The Mount Škrlatica area (Figure 2) lies in the northern part of the eastern Julian Alps and includes the Škrlatica and Martuljek mountains that rise above the tree line and substantially surpass 2,000 meters in altitude. To the north this imposing mountain group is bordered by the Sava River valley, to the west by the valley of the Velika Pišnica River, and on the east by the Vrata valley. The mountains are largely composed of carbonate rock, mostly thick layers of limestone, and the valleys are filled with glacial and river sediments. In the landscapes heavily transformed by glacial action, precipitous hillslopes that descend from peak ridges into deep cirques stand out (Mihelič 2003). Large vertical and small horizontal surface roughness is characteristic of the area (Perko 2002). The Mirna River area (Figure 3) covers the hills and to a smaller extent the low hills of the lower watershed of the Mirna River between the settlement of Tržišče and its confluence with the Sava River. In this section, the Mirna River cuts through the ridge that links the hills around Boštanj and Krško. The river carved a deep gorge, and its entrenched meanders are especially interesting. The fluviodenudational surface composed mostly of marl and dolomite rock is extremely dissected, the ridges are quite branched, and the hillslopes are smooth without any ravines (Topole 1998). Moderate vertical and horizontal surface roughness is characteristic of the area (Perko 2002). The Jeruzalem area (Figure 4) covers part of the low Ljutomersko-Ormoške hills in eastern Slovenske Gorice. Their backbone is the marl and sandstone ridge dividing the watersheds of the Ščavnica and Drava rivers. The divide ridge runs east to west, and numerous side spurs run mainly north and south (Kert 1991). In the impermeable and easily erodible rock, streams have carved numerous ravines and small valleys that are often moist due to the modest gradient. Landslides are periodically triggered on the steeper slopes (Belec 1968). Moderate vertical and large horizontal surface roughness is characteristic of the area (Perko 2002). The Škocjan area (Figure 5) is located in the southeastern part of the Divača karst region and covers the wider vicinity of the Škocjan Caves. This karst plateau is roughly level and boasts numerous dolines and deep collapse dolines that formed above the former and current underground courses of the Reka River. Many streams flow in from the northern foothills of the fysch Brkini region: several flow into the Reka River on the surface while others sink in the gables of blind valleys. The picturesque canyon and the large blind valley at the entrance to the Škocjan Caves were created by the Reka River (Mihevc 2001). Small vertical and moderate horizontal surface roughness is characteristic of the area (Perko 2002). Figure 4: Ridges of the Jeruzalem hills near Vinski Vrh (photograph: Mimi Urbanc). 3 Surface heights The curves of the frequency distribution of surface heights in DEM-25 and DEM-100 for both Slovenia and the selected areas overlap over almost their entire length. The differences in the average surface heights between DEM-25 and DEM-100 are only 0.56 m for Slovenia, 2.46 m for the Mount Škrlatica area, 1.76 m for the Mirna River area, 0.83 m for the Jeruzalem area, and 0.18 m for the Škocjan area. For the standard deviation of surface heights, the differences are 0.79 m for Slovenia, 1.64m for the Mount Škrlatica area, 1.45 m for the Mirna River area, 1.95 m for the Jeruzalem area, and 0.75 m for the Škocjan area. The differences in the coefficients of variation for surface heights between DEM-25 and DEM-100 are barely 0.08% (percentage point) for Slovenia, 0.15% for the Mount Škrlatica area, 0.29% for the Mirna River area, 0.73% for the Jeruzalem area, and 0.15% for the Škocjan area with ratios of 1.0012 or 0.12% for Slovenia, 1.0054 or 0.54% for the Mount Škrlatica area, 1.0110 or 1.10% for the Mirna River area, 1.0790 or 7.90% for the Jeruzalem area, and 1.0105 or 1.05% for the Škocj an area. It is understandable that the variability expressed by the coefficient of variation is higher for all of Slovenia than for the selected areas because the range or span of surface height for Slovenia is between 1 and 2,816 meters in DEM-100 and between 0.00 and 2,851.80 meters in DEM-25, while in the Jeruzalem area, where the range is the lowest, it is only between 208 and 332 meters in DEM-100 and between 204.50 and 346.20 meters in DEM-25. In all the cases, the differences are relatively small, and therefore the use of DEM-100 suffices for calculating basic statistical indicators such as arithmetic means, variance, standard deviation, and coefficients of variation for both the entire area of Slovenia and for smaller sections such as our selected areas. This is further demonstrated by a comparison of the average surface heights of DEM-100 and DEM-25 using the t-test, which indicated that for Slovenia and the selected areas there are no statistically significant differences given a 99% confidence level. We can also establish the differences between DEM-100 and DEM-25 by comparing their graphic representations of surface heights. For example, in the representation of the Jeruzalem area in DEM-100, we can only perceive the course and general formation of valleys and ridges (Figure 18) while DEM-25 reveals their detailed dissection by erosion gullies and ravines (Figure 19). 4 Surface slope The curves of the frequency distribution of surface slopes for all of Slovenia and the selected areas differ substantially in DEM-25 and DEM-100. For all of Slovenia, the DEM-100 curve is higher for surface slopes up to 12°, while the DEM-25 curve is higher for surface slopes from 13° on. In all of the selected areas, the DEM-100 curve is also higher for lower surface slopes and lower for higher surface slopes. This is understandable because the more accurate DEM-25 can show the variability of the surface more adequately and therefore the higher more extreme surface slopes as well. For example, for the Mirna River area the proportion of cells with a 12° surface slope is 5.78% in DEM-100 but only 3.69% in DEM-25, and the proportion of cells with a 24° surface slope is only 0.85% in DEM-100 although as high as 3.10% in DEM-25. In individual cases, the differences are multiple. The differences in average surface slopes between DEM-25 and DEM-100 are 3.06° for all of Slovenia, 2.81° for the Mount Škrlatica area, 5.31° for the Mirna River area, 3.38° for the Jeruzalem area, and 2.20° for the Škocjan area, and between the standard deviation of surface slopes, 1.83° for all of Slovenia, 1.38° for the Mount Škrlatica area, 2.11° for the Mirna River area, 2.08° for the Jeruzalem area, and 2.11° for the Škocjan area. The differences in the coefficients of variation of surface slopes in DEM-25 and DEM-100 are 6.00% (percentage points) for all of Slovenia, 1.05% for the Mount Škrlatica area, 4.91% for the Mirna River area, 3.62% for the Jeruzalem area, and 6.12% for the Škocjan area with ratios of 0.9319 or 6.81% for all of Slovenia, 1.0291 or 2.91% for the Mount [krlatica area, 0.9118 or 8.82% for the Mirna River area, 1.0706 or 7.06% for the Jeruzalem area, and 1.0857 or 8.57% for the [kocj an area. As with surface heights, the variability of surface slopes in all of Slovenia is higher than in the selected areas because the range of surface slopes for all of Slovenia is between 0.00° and 77.21° in DEM-100 and between 0.00° and 89.24° in DEM-25, while in the Jeruzalem area, where the range is the smallest, it is between 0.00° and 18.15° in DEM-100 and between 0.00° and 32.80° in DEM-25. In every case, the differences between the calculated statistical indicators are so large that even for all of Slovenia the use of DEM-25 instead of DEM-100 is recommended, and this recommendation applies even more for the smaller areas. This is further demonstrated by comparing the average surface slope of DEM-100 and DEM-25 using the t-test, which indicated that for Slovenia and the selected areas statistically significant differences exist given a 99% confidence level. The large differences between DEM-100 and DEM-25 that we established using statistical indicators are also confirmed by the graphical presentations of surface slopes. For example, the presentation of the Mirna River area on DEM-100 shows the formation of the landscape only approximately (Figure 14), while due to the greater differences between the largest and smallest slopes, smaller forms of fluviodenudational relief are clearly visible on DEM-25 (Figure 19), for example, river meanders and the larger scarps of the river terraces. 5 Surface aspect For all of Slovenia and the selected areas, the curves for the frequency of distribution of surface aspects on DEM-25 and DEM-100 differ more than the curves for the frequency of distribution of surface heights but considerably less than the curves for the frequency of distribution of surface slopes. The path of the surface aspect curve for DEM-25 is similar to the path of the surface aspect curve for DEM-100 but smoothed, indicating that the proportions of individual surface aspects on DEM-25 are more even distributed. This is true for all of Slovenia and for the selected areas. For example, for all of Slovenia the proportion of cells with the southernmost aspect of 180° is 2.53% for DEM-100 but only 1.50% for DEM-25. On both digital elevation models, the smallest proportion of cells have an exposition of 64°: 0.26% on DEM-100 and 0.40% on DEM-25. For DEM-100, the ratio between the two values is almost 10, while for DEM-25 it is less than 4. The differences in the average aspects on DEM-25 and DEM-100 are 2.11° for all of Slovenia, 1.38° for the Mount [krlatica area, 1.05° for the Mirna River area, 2.20° for the Jeruzalem area, and 3.87° for the [kocjan area, while the standard deviations of aspect are 1.24° for all of Slovenia, 0.18° for the Mount [krlatica area, 0.89° for the Mirna River area, 1.37° for the Jeruzalem area, and 0.23° for the [kocj an area. The differences in the coefficients of variation of aspect for DEM-25 and DEM-100 are 0.06% (percentage points) for all of Slovenia, 1.20% for the Mount [krlatica area, 1.68% for the Mirna River area, 2.44% for the Jeruzalem area, and 3.10% for the [kocjan area, while the ratios are 0.9990 or 0.10% for all of Slovenia, 0.9805 or 1.95% for the Mount [krlatica area, 0.9720 or 2.80% for the Mirna River area, 1.0540 or 5.40% for the Jeruzalem area, and 0.9507 or 4.93% for the [kocj an area. In contrast with surface heights and surface slopes, the variability of surface aspects on DEM-100 and DEM-25 - with the exceprion of the Jeruzalem area - is slightly larger in all the selected areas than for Slovenia as a whole, which is also understandable since the range of surface aspect for all of Slovenia and for all the selected areas is the same, between 0.00° and 180.00°. Although the differences between the calculated statistical indicators are smaller than with surface slopes, for both all of Slovenia and the smaller areas, the use of DEM-25 instead of DEM-100 is recommended, especially in hilly and dissected karst areas. Figure 10: Surface aspects of DEM-100 with values between 0.00° (lightest colour) and 180.00° (darkest colour) for the Mount [krlatica area. Figure 11: Surface aspects of DEM-25 with values between 0.00° (lightest colour) and 180.00° (darkest colour) for the Mount [krlatica area. Figure 18: Surface heights of DEM-100 with values between 208 m Figure 19: Surface heights of DEM-25 with values between 204.50 m (lightest colour) and 332 m (darkest colour) for the Jeruzalem area. (lightest colour) and 346.20 m (darkest colour) for the Jeruzalem area. Figure 20: Surface slopes of DEM-100 with values between 0.00° Figure 21: Surface slopes of DEM-25 with values between 0.00° (lightest colour) and 18.15° (darkest colour) for the Jeruzalem area. (lightest colour) and 32.80° (darkest colour) for the Jeruzalem area. Figure 24: Surface heights of DEM-100 with values between 334 m Figure 25: Surface heights of DEM-25 with values between 315.40 m (lightest colour) and 750 m (darkest colour) for the [kocjan area. (lightest colour) and 770.40 m (darkest colour) for the [kocjan area. Figure 26: Surface slopes of DEM-100 with values between 0.00° Figure 27: Surface slopes of DEM-25 with values between 0.00° (lightest colour) and 29.80° (darkest colour) for the [kocjan area. (lightest colour) and 52.82° (darkest colour) for the [kocjan area. 00 ^ LO_ CO co co co" CO C^ LO CN co LO 00 cm lo^ cm c^ co co lo" co i— co co c^ co co co cm LO "^r co, co, co_ co 1— 1— co" LO LO LO LO co co LO co co, co, co LO co 1— 1— LO LO co co LO co "" "" "" "" C^ C^ CO CO" LO LO co co co co, co, cO" co c^ c^ cm LO" co co co LO c^ cm co co LO cm LO c^ c^ c^ co c^ c^ c^ co >oo M co co co LO co co co LO co co cm co co cm LO co co co LO co co cm LO co co cm LO >cn M 6 Sklep Preučevanje reliefa in pokrajine s pomočjo digitalnih modelov višin je povezano predvsem z njihovo natančnostjo. V članku smo obravnavali stometrski in petindvajsetmetrski digitalni model višin Slovenije glede na razlike pri višinah, naklonih in ekspozicijah površja za Slovenijo in štiri reliefno različna območja. Primerjali smo pogostnostno porazdelitev, aritmetično sredino, standardni odklon, variacijski razmik in koeficient variacije ter testirali statistično pomembnost razlik aritmetičnih sredin. Pri primerjavi pogostnostnih porazdelitev smo ugotovili: • krivulji pogostnostne porazdelitve nadmorskih višin DMV-25 in DMV-100 se skoraj prek celotnega poteka prekrivata, • pri naklonih je krivulja DMV-100 pri manjših naklonih višja, pri večjih naklonih pa nižja od krivulje DMV-25, • pri ekspozicijah pa je potek krivulje DMV-25 podoben poteku krivulje DMV-100, a bolj umirjen, z manjšimi nihanji. Pri testiranju aritmetične sredine smo ugotovili: • razlika med povprečno nadmorsko višino DMV-25 in DMV-100 ni statistično pomembna niti pri Sloveniji niti pri izbranih območjih, • razlika med povprečnim naklonom DMV-25 in DMV-100 je statistično pomembna tako pri Sloveniji kot pri vseh izbranih območjih, • razlika med povprečno ekspozicijo DMV-25 in DMV-100 je statistično pomembna pri Sloveniji, Jeruzalemu in Škocjanu, pri Škrlatici in Mirni pa ne. Primerjava statističnih kazalcev je pokazala: • razlika med DMV-25 in DMV-100 je pri povprečni višini površja najmanjša pri Škocjanu in največja pri Škrlatici, pri povprečnem naklonu površja najmanjša pri Škocjanu in največja pri Mirni, pri povprečni ekspoziciji površja pa najmanjša pri Mirni in največja pri Škocjanu. • razlika med DMV-25 in DMV-100 je pri standardnem odklonu višine površja najmanjša pri Škocjanu in največja pri Jeruzalemu, pri standardnem odklonu naklona površja najmanjša pri Škrlatici in največja pri Mirni in Škocjanu, pri standardnem odklonu ekspozicije površja pa najmanjša pri Škrlatici in največja pri Jeruzalemu, • razlika med DMV-25 in DMV-100 je pri koeficientu variacije višine površja najmanjša pri Sloveniji in največja pri Jeruzalemu, pri koeficientu variacije naklona površja najmanjša pri Škrlatici in največja pri Škocjanu, pri koeficientu variacije ekspozicije površja pa najmanjša pri Sloveniji in največja pri Škocjanu. Glede na velikostno klasifikacijo reliefnih oblik, ki jo je opravil francoski geomorfolog Tricart (1965), lahko ugotovimo, da so na prikazih DMV-100 dobro vidne mezoreliefne in elementarne reliefne oblike, med katere na primer uvrščamo gorske hrbte, doline, kotline in podobno, manjše reliefne oblike, med katere spadajo pobočni erozijski žlebovi, rečne terase, večji morenski nasipi, vrtače ter nekatere antropogene oblike, na primer obrečni nasipi, izkopi kamnolomov in obdelovalne terase, pa se veliko jasneje kažejo na prikazih DMV-25. Splošna ocena je, da DMV-100 pri višinah površja lahko tudi pri manjših območjih pri temeljnih statističnih kazalcih nadomesti DMV-25, pri ekspozicijah površja v določenih primerih, pri naklonih površja pa le izjemoma. DMV-100 je glede na DMV-25 izrazito slab pri območjih z izrazito vodoravno razgibanostjo površja, kakršna je večina slovenskih gričevij v panonski in sredozemski Sloveniji in nizkih kraških planot v sredozemski in dinarski Sloveniji. Razlike med DMV-100 in DMV-25, ki smo jih ugotovili s statističnimi kazalci, potrjujejo tudi grafični prikazi višin, naklonov in ekspozicij površja (slike 6 do 29). 7 Literatura in viri Glej angleški del prispevka.