Confrontation of the Forest Soil Condition Level I database with the Soil Geographical Database of Europe

Confrontation of the Forest Soil Condition Level I database w ith the Soil Geographical Database of Europe Torsten Wiedemann, Jaco Klap, Dominique Langouche, Eric Van Ranst Forest Soil Co-ordinating Centre Ghent, May 2001 . Confrontation of the Forest Soil Condition Level I database w ith the Soil Geographical Database of Europe Torsten Wiedemann Jaco Klap Dominique Langouche Eric Van Ranst Forest Soil Co- ordinating Centre Ghent, May 2001 Prepared with the support of the Flemish Community and the European Commission (Flemish Community Project Nr. B&G/5/1998, EC - Project nr. 98.60 BL.003.00) Index I ndex List of figures ............................................................................................................................. 2 List of tables............................................................................................................................... 2 I ntroduction ............................................................................................................................... 3 1. Structure and comparability of FSCDB and SGDBE........................................................... 7 1.1. Structure of the SGDBE..................................................................................................... 7 1.2. Structure of he FSCDB...................................................................................................... 7 1.3. Comparability of the two databases ................................................................................... 8 1.3.1. Positional comparability .............................................................................................. 8 1.3.2. Thematic comparability .............................................................................................. 8 1.4. Methods for the detection of corresponding STU's for each Leve l I forest soil plot ...............11 1.4.1. Buffering ..................................................................................................................14 1.4.2. Simple approach.......................................................................................................16 1.4.3. Normal approach – direct and buffered ......................................................................17 1.4.4. Advanced approach ..................................................................................................17 2. Results of comparison betw een FSCDB Level I forest soil plots and SGDBE STU's...... 19 2.1. Overall correspondence according to FAO soil code and land use........................................19 Simple approach ...................................................................................................................19 Normal direct approach .........................................................................................................20 2.1.3. Normal buffered approach .........................................................................................21 2.1.4. Advanced approach ..................................................................................................23 2.1.5 Comparison of different approaches................................................................................24 2.2. Correspondence according to information about parent material ........................................25 2.2.1. Overall correspondence.............................................................................................25 2.2.2. Normal buffered approach .........................................................................................27 2.2.3. Advanced approach: .................................................................................................29 The soil plots with parent material information and corresponding STU's, as assigned by the advanced approach is given in Figure 18. ................................................................................29 2.2.4. Correspondence per country......................................................................................30 2.3. 3. Correspondance according to information about texture ....................................................31 Conclusions........................................................................................................................ 33 List of abbreviations ................................................................................................................ 39 References................................................................................................................................ 39 Annex........................................................................................................................................ 41 1. Translation of FAO74 code into FAO90code: ................................................................... 43 2. Synchronisation of parent material codes in both databases:....................................... 47 1 Index List of figures Figure 1: Schematic presentation of the confrontation procedure ................................................ 5 Figure 2: Structure of the SGDBE............................................................................................. 7 Figure 3: Overlay of SGDBE with map of land use in Belgian Flanders: there are no forests in the neighbourhood of soil plots 1702 and 1802, according to the SGDBE. ......................................10 Figure 4: One-to-one overlay of FSCD soil plots on SGDBE......................................................11 Figure 5: Result of one-to-one map overlay .............................................................................12 Figure 6: Distances from soil plots without corresponding SMU to nearest SMU.........................12 Figure 7: Percentage of soilpots within 40x40 km grid cells where FAO groups in both databases are identical and where land use is ‘forest’ ..............................................................................13 Figure 8: Confrontation of buffered FSCD soil plot with SGDBE.................................................14 Figure 9: Results of confrontation of buffered FSCD soil plots with SGDBE in a case study area in Finland …………………………………………………………………………………………………… 15 Figure 10: Map of soil plots with corresponding STU's, assigned by simple approach..................19 Figure 11: Map of soil plots with corresponding STU's, assigned by normal direct approach.........20 Figure 12: Map of soil plots with corresponding STU's, assigned by normal buffered approach.....21 Figure 13: Map of soil plots with corresponding STU's, assigned by advanced approach..............23 Figure 14: Percentage of selected soil plots vs. class of correspondence ....................................24 Figure 15: Histogram of the effect of application of buffered approach vs. direct approach ............25 Figure 16: Correspondence of parent material group vs. soil map properties for different approaches ...........................................................................................................................26 Figure 17: Map of soil plots with parent material information and corresponding STU's, assigned by normal buffered approach ......................................................................................................27 Figure 18: Map of soil plots with parent material information and corresponding STU's, assigned by advanced approach ...............................................................................................................29 Figure 19: Histogram of correlation between level of correspondence and percentage of corresponding parent material groups .....................................................................................30 Figure 20: Correspondence of textural class vs. soil map properties for direct and buffered overlay ...............................................................................................................................31 Figure 21: Map of FSCDB level I soil plots with/without corresponding STU’s according to final approach...............................................................................................................................34 Figure 22: Histogram of the final result: Percentage of soil plots with corresponding STU per country ……………………………………………………………………………………………………35 List of tables Table 1: Countries covered by level I soil plots and by SGDBE ........................................................ 4 Table 2: Number of Level I soil plots with attribute information.......................................................... 4 Table 3: Comparison of the attributes of FSCD and SGDBE ............................................................ 8 Table 4: Resulting information for each FSCDB plot after map overlay with SGDBE .........................16 Table 5: Ranking codes for the degree of correspondence between the two databases: simple approach...............................................................................................................................16 Table 6: Ranking codes for the degree of correspondence between the two databases: normal approach...............................................................................................................................17 Table 7: Ranking codes for the degree of correspondence between the two databases: advanced approach...............................................................................................................................18 Table 8: Assigned ranking codes for different levels of correspondence between FSCDB and SGDBE . ......................................................................................................................................28 Table 9: Correlation between level of correspondence and percentage of corresponding parent material groups......................................................................................................................28 Table 10: Assigned ranking codes for different levels of correspondence between FSCDB and SGDBE: final approach..........................................................................................................33 Annex 1: Translation of FAO code from 1974 into FAO code from 1990 Annex 2: Translation of SGDBE parent material code into FSCD parent material code FSCD-code ………...................................................................................................................................49 2 Introduction I ntroduction The study presented here was carried out as a part of the project mandate of the Forest Co-ordinating Centre (FSCC) of June 1998 till May 2001. The activities of FSCC are framed in the I nternational Cooperative Programme on Assessment and Monitoring of Air Pollution Effect on Forests (I CPForests)(UN/ ECE) and the European Union Scheme on the Protection of Forests against Atmospheric Pollution (EU). FSCC was established at the Laboratory for Soil Science at Ghent University in 1992, and started its activities in November 1993 for a period of four years. During this first mandate, FSCC co-ordinated the national soil inventories that were carried out in the frame of I CP Forests on forest soil plots located on a 16 x 16-km2 grid covering Europe, the so-called level I plots. The data generated from this survey were stored by FSCC in what is called the Forest Soil Condition Database (FSCDB). The study of these data (also by FSCC) resulted in the publication of the Forest Soil Condition Report (EC, UN/ ECE and Ministry of the Flemish Community, 1997). I n 1997, FSCC was mandated to continue its co-ordinating activities, and accept some new projects amongst which the confrontation of the Forest Soil Condition Database and the Soil Geographical Database of Europe (SGDBE). Data on parent material and texture, are presently unavailable for 60% , respectively 35% of the level I soil plots. The aim of the underlying study was to investigate the possibility of upgrading the Forest Soil Condition Database and obtain additional data on parent material and texture by making use of the Soil Geographical Database of Europe and the attribute data linked to it respectively. I n 1985, a soil map covering the entire European Community was published at a scale of 1: 1,000,000 (CEC, 1985). I t was digitised in 1986 in order to establish a comprehensive geographical database. The map displays more than 300 Soil Mapping Units (SMU’s) which relate to more than 16,000 soil polygons. Each SMU is an association of dominant and subdominant Soil Typological Units (STU’s). Version 3.2.8.1 of the Soils Geographical Database extends the map to other European countries. I n a first phase the Central European countries (Poland, Czech Rep., Slovak Rep., Eastern Germany, Bulgaria, Hungary and Romania) were included. The ‘Soil and GI S’ working group of the European Soils Bureau started in 1995 with the second phase of extension and included other Eastern European and Scandinavian countries. The SGDBE is now based on a soil map covering 42 countries (Table 1) at a scale of 1: 1.000.000. This version was available in October 1998. Countries covered by SGDBE ( bold : covered by FSCDB Level I soil plots) ALBANI A AUSTRI A BELGI UM BOSNI A HERZEGOVI NA BELARUS BULGARI A DENMARK I RI SH REPUBLI C ESTONI A CZECH REPUBLI C FI NLAND FAEROE I SLANDS FRANCE GUERNSEY GERMANY GREECE CROATI A HUNGARY I SLE OF MAN 3 Introduction I TALY JERSEY LATVI A LI THUANI A SLOVAKI A LUXEMBOURG MOLDOVA Former Yugoslav Republic of Macedonia MALTA MONTENEGRO NETHERLANDS NORW AY POLAND PORTUGAL ROMANI A RUSSI A SLOVENI A SPAI N SERBI A SW EDEN SWI TZERLAND UKRAI NE UNI TED KI NGDOM Table 1 Countries covered by level I soil plots and by SGDBE The STU’s in version 3.2.8.1 of the database are be characterised by: − 1974 and 1990 FAO Legend soil name; − texture (dominant and secondary, surface and subsurface textural class); − phase (used to indicate indurated layers, shallow depth, stoniness, alkalinity, salinity); − slope (dominant and secondary); − altitude (minimum and maximum); − parent material (dominant and secondary); − land use (dominant and secondary); − depth class of an obstacle to roots; − presence of an impermeable layer; − dominant annual average soil water regime class; − information on the use of an agricultural water management system. An overview of the number Level I plots for which attribute data from the Soil Geographical Database of Europe may be obtained is presented in table 1. Number of Level I soil plots Total Within area covered by With information about With information about With information about Table 2 SGDBE FAO code parent material texture 5289 5269 4993 2120 3418 % of level I soil plots 100 99 94 40 65 Number of Level I soil plots with attribute information The SGDBE consists of 1650 soil mapping units (SMU's) which relate to 28946 soil polygons. Each SMU is described by soil typological units (STU's), stored in the database. There are 5306 different STU's. 4 Introduction Procedure Basically, the confrontation of this database (SGDBE) with the forest soil condition database consists of the following steps (Figure 1) a) I dentification of SMU’s with at least one STU having ‘forest’ as dominant or secondary land use. I t is assumed that the data linked to STU’s not under forest cannot provide relevant information for the Level I plots; b) Localisation of the soil condition plots within the SMU’s identified under a) and identification of the corresponding STU; c) Matching the FAO soil name - the only common mandatory parameter - in both databases. The matching procedure may have three outcomes: 1. the soil observation plot has a FAO classification name which is identical to the FAO soil name of the STU; 2. the FAO classification name of the soil plot differs from the STU soil name, but the attributed FAO soil names are similar; 3. the FAO classification name of the soil plot differs from the STU soil name, and the names are dissimilar. I t has to be kept in mind that a 1: 1,000,000 map can only list the major soil types in each mapping unit. Associated soils which cover less than 10% of the SMU area are usually not reported (Breuning-Madsen and Jones, 1995). The selection of the observation plots in the soil condition survey, on the other hand, was based on a systematic grid, ensuring statistical representativeness at Eur opean scale but disregarding matters of representativeness of the plot location within the surrounding area. I n outcome 1 the information attributed to the STU can be assumed to be also valid for the forest soil condition plot, although this validity has to be verified. This would provide valuable additional information on parameters that are currently unavailable in the forest soil condition database. I f the matched soil names differ from each other, the similarity of the attributed soil names is evaluated. I n the situation of a dissimilar pair (outcome 3), the soil observed at the plot is not representative for the SMU unless an error has occurred in one of the databases. I n either case, information attributed to the STU cannot be linked to the forest soil condition plot. d) Verification of the matching results. The validity of the STU attribute information for the matched plots can be verified by comparing it to information collected at the plot itself. I ndeed, part of the attribute information, e.g. texture, parent material is available for the forest soil condition plots in several countries. This procedure not only verifies the occurrence of contradictions in the databases, but may determine under which matching condi tions STU attribute information can be linked to the forest soil condition plots. Overlay of FSCD (plots) with SDE (SMUs-STU’s) plot located in SMU lacking STU with landuse ‘Forest’ plot located in SMU having 1 or more STUs with landuse ‘Forest’ identification of corresponding STU FAO names identical FAO names similar STU attribute info valid for plot plot ? FAO names dissimilar STU attribute info invalid for plot plot Figure 1 Schematic presentation of the confrontation procedure 5 Introduction Since the positional accuracy of a map at a scale of 1: 1,000,000 is low, methods have to be developed to make the two databases (FSCDB and SGDBE) comparable: I n chapter 1 the structure and the comparability of the two databases will be discussed, as well as the different possible approaches that can be followed to compare both databases. I n chapter 2 the results of different methods for the selection of corresponding STU's for each Level I forest soil plot are discussed. Chapter 3 contain the conclusions for the possibilities to extract additional information from the SGDBE. 6 Structure and Comparability of FSCDB and SGDBE 1. Structure and comparability of FSCDB and SGDBE 1.1. Structure of the SGDBE The base units of this database are 5306 soil typological units (STU’s). These units are described by attributes specifying the nature and properties of the soil (table 3). These attributes or variables were estimated over large areas by expert judgement (based on national or regional maps at a more detailed scale - 1: 25,000 or 1: 50,000 - rather than measured at the spot itself. Since the map scale is 1: 1,000,000 it is not feasible to delineate the STU's as polygons on the map. Therefore they are grouped into Soil Mapping Units (SMU's) to form soil associations within landscapes, as shown in Figure 2. The original information (STU) is represented by the dotted line, the information kept in the SGDBE (SMU) by the full line. The SMU's are represented by polygons on a digital map with an overall positional accuracy of (estimated) 500 - 5000 m (corresponding with 0,5 5 mm at scale 1: 1,000,000). The minimum polygon area has been set to 9 ha (corresponding with 0,3 x 0,3 mm at scale 1: 1,000,000). 1 2 SMU STU Attributes 1 1 1 2 2 3 3 1 2 5 1 3 4 2 a… d… g… a… e… f… d… 3 Figure 2 Structure of the SGDBE Due to the nature of estimated data and the small scale of the map the thematic and positional accuracy of the SGDBE is low. As such the SGDBE does not contain the location of the STU's occurring in each SMU, but only the percentages of surface by which each SMU is covered with a certain STU. 1.2. Structure of he FSCDB The FSCDB contains at present information about 5289 soil plots, which has been gathered locally at the cross-sections of a 16x16 km grid, covering 28 European countries (Table 1). The database is based on 5 main tables: a) The soil plot information: each soil plot is described by its geographical co-ordinates, the soil type (according to the FAO codes of 1990), altitude and other general observations. b) Mandatory soil information: measured contents of N, P, K, Ca, Mg, organic material, CaCO3 and pH. c) Optional soil information: measured contents of Na, Al, Fe, Cr, Ni, Mn, Zn, Cu, Pb and cation exchange. d) Parent material: parent material is described for 2120 soil plots. e) Voluntary soil information: texture, bulk density and coarse fragments. The geographical accuracy of each soil plot is +/- 30m, since each co-ordinate is measured with a precision of +/- 1’’. 7 Structure and Comparability of FSCDB and SGDBE 1.3. Comparability of the tw o databases 1.3.1. Positional comparability While the Level I forest soil plots are described with a geographical precision of 1 second (corresponding with + / - 30m), the accuracy of the SGDBE is very low and varies between + / - 500 m to + / 5000 m. A direct overlay of the level I soil plots leads to situations where the soil plot is situated on the soil map far from its actual position. Differences between SGDBE and FSCDB information will result from this posi tional error. 1.3.2. Thematic comparability By comparing the list of attributes of the STU’s of the SGDBE with the attributes (mandatory, optional and voluntary parameters) of the FSCD soil plots we get the following table, where field “ FSCD-Soil plots” contains all FSCD parameters which can be compared with the attributes of the SGDBE-STU’s. A “ ?” indicates FSCD-parameters which might be extracted from the FSCD indirectly. Table 3 Comparison of the attributes of FSCD and SGDBE Attributes of the SGDBE – STU’s AGLI M1, 2 AREA CFL DT IL MAT1, 2 NB-POLYS NB-SMU NB-STU ROO SLOPE1, 2 SMU SOI L SOI L90 TEXT1, 2 TD1, 2 USE1, 2 Comparable attributes in FSCD- Soil plots ? ? PMAT Soil unit Texture02 Texture2x Forest WM1 WM2 WM3 WR ZMI N/ MAX Altitude Description of attribute codes ( 1 = dominant, 2 = secondary) Limitation to agricultural use (e.g. stony, lithic, flooded...) SMU, STU Global confidence level of the Soil Typological Unit attributes Depth class of textural change Presence of impermeable layer Parent material Number of polygons making SMU or containing STU Number of SMUs containing STU Number of STUs composing SMU Depth class of an obstacle to roots Slope classes I D (-7 ...) FAO 1974 FAO 1990 (and derived information: 1 st , 2 nd level, soilgroup) Surface textural class Sub-surface textural class Land use (2 = poplars, 5 = forest, coppice, 9 = Bush, macchia, 20 = Dehesa (extensive agricultural-pasture system in forest parks in spain)) Presence of water management system in agricultural land (Yes/ No) Purpose of water management system (drainage, irrigation...) Type of water management system Dominant annual average soil water regime class of the soil profile (depth and # of months) Altitude above sea-level (meters) 8 Structure and Comparability of FSCDB and SGDBE Five attributes have been detected as common in both databases: the soil type, according to the FAO-code of 1990 the parent material the texture between 0 – 20 cm the altitude land use - Before both datasets could be compared, these 5 attributes had to be synchr onized: 1.1. Soil type 4993 (94% of 5289) FSCDB level I soil plots are described by FAO 90 codes. 276 soil plots, mainly situated in Sweden en I taly, are not described by their FAO-soil name. 20 soil plots, situated on the Canaries and Azores are situated out of the map area of the SGDBE. For the description of most of the STU’s (58% , corresponding with 66% of the mapping area) only FAO-codes of 1974 have been used to describe soil properties in the SGDBE. I n the FSCDB however, soil classification names were given according to the FAO 1990 - classification system. Therefore, the soil type in the SGDBE had first to be translated into FAO-codes of 1990,. Therefore the full FAO-names of 1974 and 1990 have been compared with each other. I n cases where the FAO-name of 1974 did not occur in the FAO legend of 1990 the most similar soil type has been assigned. Third level properties of the FAO-code of 1974 have been neglected: e.g. “Bcc” (Calcaro-Chromic Cambisols) has been translated as “ Bc” (Chromic Cambisols) into “ CMx” . The full list of FAO 74 vs. FAO 90 codes is given in annex 1. 1.2. Parent material Parent material has been described for 2120 soil plots (40% ) and for almost all STU’s. The original codes used in both databases were not comparable and had been translated into 9 groups of parent material: 1: 2: 3: 4: 5: 6: 7: 8: 9: Crystalline rocks Volcanic rocks Sandstones Limestones Sands Loam Clays Organic Deposits The list of SGDBE parent material codes and groups vs. FSCDB codes and groups is given in annex 2. 1.3. Texture Texture has been described for 3418 (65% ) of the FSCD soil plots and for almost all STU’s. The texture in the SGDBE has been measured or estimated for the upper 20 cm of the soil. Since the layers in the FSCD have not been analyzed homogeneously for the same depth, the textural class for a 0-20 cm standard-layer has been calculated for each FSCD soil plot where information about texture was given (Klap et al., 2001). The texture classes are identical for both databases. 9 Structure and Comparability of FSCDB and SGDBE 1.4. Altitude Altitude has been described for all FSCD soil plots and for almost all SGDBE STU’s. I n the FSCD altitude has been classified into 50 m - intervals. I n the SGDBE for most of the STU’s (more than 99% ) a Zmin and a Zmax value has been given with a difference of more than 50 m. 43 % of the STU’s are described by a difference between Zmin and Zmax of more than 250 m. Due to this large interval this attribute appeared not to be suitable for the comparison of the two databases. 1.5. Land use I n the SGDBE ‘forest’ is described by the STU attribute “ land use” (primary and secondary), where the categories “ poplars” , “ forest” , “ coppice” , “ bush” , “ macchia” and “ dehesa” have been distincted. At least for the Flanders region it has been proven that (perhaps due to the high degree of generalization of the map) the soil map of Europe lacks detailed information about forests. Figure 3 illustrates this by an overlay of the soil map of Europe with the map of land use in Flanders. This figure also illustrates the low degree of positional accuracy of the SGDBE, if one compares the outlines of the two maps. Figure 3 Overlay of SGDBE with map of land use in Flanders (Belgium): there are no forests in the neighbourhood of soil plots 1702 and 1802, according to the SGDBE. 10 Structure and Comparability of FSCDB and SGDBE 1.4. Methods for the detection of corresponding STU's for each Level I forest soil plot By performing a map overlay of the FSCD soil plots with the SMU’s of the SGDBE, we get pairs of soil plots and SMU’s and each pair can be related to the corresponding STU’s. Figure 4 One-to-one overlay of FSCD soil plots on SGDBE. 11 Structure and Comparability of FSCDB and SGDBE Percentage of plots with same soil type as on soil map of Europe AND land use = "forest" 100,00 90,00 80,00 70,00 60,00 50,00 40,00 30,00 20,00 10,00 0,00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 29 30 Figure 5 Result of one-to-one map overlay (x-axis: Country-code: See list of abbreviations) Figure 4 and Figure 5 illustrate the result of a direct (one-to-one) overlay. Especially in regions with smaller, fragmented forests, such as southern Sweden, the Flemish region of Belgium, Northern Germany, Netherlands, England, Hungary etc. only few soil plots match with the SGDBE. I n countries with random coastlines (such as Norway and Croatia) and with lots of waterbodies (e.g. Finland), lots of soil plots are not situated within a SMU but outside the map area or within waterbodies, cities or other SMU’s wit hout soilproperties, according to the SGDBE. The distances of these soil plots to the nearest SMU illustrate the degree of geographical misplacement, which results from the low positional accuracy of the SGDBE (Figure 6 ). 35 30 25 20 15 10 5 47 50 42 50 37 50 32 50 27 50 22 50 17 50 12 50 75 0 0 25 0 number of plots without SMU Distance of Level 1 plots from nearest SMU meters Figure 6 Distances from soil plots without corresponding SMU to nearest SMU Figure 7 shows the percentage of soil plots where within a distance of 5 km at least one STU can be found with corresponding FAO soilgroup. The correspondence is low especially in areas where forests are fragmented. These areas are situated in Norway, Hungary, U.K., Greece, I reland and the Netherlands, where more than 40 % of all soil plots do not correspond with matching STU’s. 12 Structure and Comparability of FSCDB and SGDBE An exception on this rule is Poland, where the correspondence between the two databases is high, although the soil plots are situated in an area with fragmented forests. Figure 7 Percentage of soilpots within 40x40 km grid cells where FAO groups in both databases are identical and where land use is ‘fo rest’ 13 Structure and Comparability of FSCDB and SGDBE The results of these one-to-one overlays show that the methods for the detection of correspon ding STU’s have to respond to the low thematic and geographic precision of the SGDBE. Different approaches will be discussed in what follows. 1.4.1. Buffering By creating a buffer of 5000 m around each soil plot and performing an overlay/ intersect operation between the buffered soil plots and the soil map of Europe, we get better results than in the case of a one-to-one map-overlay, by avoiding the problems that are induced by the differences between the geographical accuracies of both databases. Figure 8 shows schematically to which STU a FSCDB-plot may be linked after creating such a buffer. 1 a d b 2 102 c Plotid SMU 102 2 102 2 STU d e Attributes x… y… z… y… 102 102 g f x… y… 3 3 x… r … f e g 3 Figure 8 Confrontation of buffered FSCD soil plot with SGDBE Figure 9 illustrates this operation for a case study area in Finland. 14 Structure and Comparability of FSCDB and SGDBE Figure 9 Results of confrontation of buffered FSCD soil plots with SGDBE in a case study area in Finland (bold FAO-codes are soil properties of FSCD soil plots, pie-diagrams illustrate the content of STU’s (according to their area within SMU’s)) FAO-code “ CMd” is associated with “PZh”: “ HSf “: “Gle”: “PZg”: ‘forest’ ‘forest’ ‘forest’ ‘forest’ ‘forest’ as secondary land use-category as primary luc. as secondary luc. as secondary luc. as primary luc. 15 Structure and Comparability of FSCDB and SGDBE 1.4.2. Simple approach Corresponding STU’s can be selected by the following queries: a) b) c) Select the largest STU with primary land use = ‘forest’; if there is none: Select the largest STU with secondary land use = ‘forest’, if there is none: Select the largest STU. The result contains the following information for all FSCDB plots: Country Nr Linkid Soilid Soil unit Pmatco Pmat gr STU (* ) Stuarea (* * ) P No_of_STUs Use1, 2 Pmat1 (* * * ) Pmatg1 (* * * * ) Pmat2 (* * * ) Pmatg2 (* * * * ) Soil unit_s Text 1 Slope1 Aglim1, 2 Mat1, 2 Dt Td1 Roo Il Wr Table 4 FSCDB-country code FSCD-linknr FSCD-linkid FSCD-soil plot-id, missing values = “-1” FSCD-code, missing values = “-1” FSCD-code (parent material), missing values = “-1” ( → Pmat1, 2) FSCD-code (parent material group 1-9), missing values = “-1“ (→ Pmatg1, 2) SGDBE-code for Soil Typological Unit, missing value = “-2” The share of the buffer area of the STU (% ), missing value = “-2” Hierarchical value, according to queries Number of STU’s with the same value of “ P” Primary/ secondary land use-code, according to SGDBE, missing values = “-1” Primary parent material-code, SGDBE, missing values = “-2” Primary parent material group-code, SGDBE, missing values = “-2” Secondary parent material-code, SGDBE, missing values = “-2” Secondary parent material group-code, SGDBE, missing values = “-2” Soil unit-code, according to SGDBE, missing values = “-2” Texture-code, according to SGDBE, missing values = “-1” missing values = “-1” missing values = “-1” SGDBE-code for parent material, missing values = “-1” missing values = “-1” missing values = “-1” missing values = “-1” missing values = “-1” missing values = “-1” Resulting information for each FSCDB plot after map overlay with SGDBE We define a ranking code “ P” which describes the correspondence between the two databases.I n this case the values for “ P” are described as follows: 3 2 1 Table 5 The largest STU with primary land use = ‘forest’ The largest STU with secondary land use = ‘forest’ The largest STU Ranking codes for the degree of correspondence between the two databases: simple approach *: **: ***: ****: STU’s have also been assigned to plots out of the mapping area of the European Soil map, when they were within a distance of < 5000m. values of “ 0” are values where the share was less than 0.00001 % (corrsponding with 785 m² ) The codes of the SGDBE had to be translated into FSCD-codes. Not all parent materialcodes could be translated. Codes which were not translated were assigned a ‘missing value’ All SGDBE pmat-codes have been translated into FSCD pmat-group-codes, according to table in Annex 2. 16 Structure and Comparability of FSCDB and SGDBE Since the selection of corresponding STU’s in this simple approach is independent from the FSCDB soil attributes, it will be used as a reference to calculate the validity of the more advanced approaches. 1.4.3. Normal approach – direct and buffered The order in which STU's are selected is very important for the result. The 'normal' approach is an approach where STU's with identical full FAO names in both databases are selected prior to STU's with similar FAO names, even if the identical full FAO names are not correlated with 'forest' as primary land use category. This approach is based on the assumption that valuable information can only be extracted from STU’s where FAO codes are corresponding with the FAO codes of the FSCDB soil plots. I f two or more STU's are described by the same FAO name and if they are correlated with the same soil plot, 'forest' as primary or secondary land use category will be handled as an indicator that the correspondence between the two databases might be higher than in cases where land use is not described as 'forest'. FAO codes are thought to be similar if the soilgroups are identical (e.g. Calcaric Cambisol vs. Gleyic Cambisol). STU’s are selected if they respond to the following queries: Table 6 Ranking codes for the degree of correspondence between the two databases: normal approach Ranking code “ P” for correspondence Queries 8 7 6 5 4 3 2 1 0 Are FAO codes identical AND is primary land use = ‘forest’ ? I f not: next query Are FAO codes identical AND is secondary land use = ‘forest’ ? I f not: next query Are FAO codes identical ? I f not: next query Are FAO codes similar AND is primary land use = ‘forest’ ? I f not: next query Are FAO codes similar AND is secondary land use = ‘forest’ ? I f not: next query Are FAO codes similar ? I f not: next query I s primary land use = 'forest' ? I f not: next query I s secondary land use = 'forest' ? I f not: next query I f none of the queries can be answered positively I f the ranking codes are identical for different STU’s and if these STU’s correspond with the same soil plot, only the largest STU has been selected. The normal approach will be applied for the direct and buffered overlay method as well to get information about the impact of the low geographical accuracy of the SGDBE. 1.4.4. Advanced approach The advanced approach takes the information about parent material and secondary soil properties into account as well: STU's with corresponding parent material codes are selected prior to STU's with identical FAO codes if both attributes do not correspond simultaneously. 17 Structure and Comparability of FSCDB and SGDBE STU’s are selected if they respond to the following queries: Table 7 Ranking codes for the degree of correspondence between the two databases: advanced approach RC1 Queries2 RC1 Queries2 RC1 Queries2 30 27 24 21 18 15 12 9 6 3 Forest 1, FAO, pmat Forest1, FAO, pmatg Forest1, pmatg Forest1, soilpr, pmatg Forest1, pmat Forest1, pmatg Forest1, FAO Forest1, soilgroup Forest1, soilpr Forest1, maxarea 29 26 23 20 17 14 11 8 5 2 Forest 2, FAO, pmat Forest2, FAO, pmatg Forest2, FAOg, pmatg Forest2, soilpr, pmatg Forest2, pmat Forest2, pmatg Forest2, FAO Forest2, soilgroup Forest2, soilpr Forest2, maxarea 28 25 24 19 16 13 10 7 4 1 No No No No No No No No No No 1 forest, FAO, pmat forest, FAO, pmatg forest, FAOg, pmatg forest, soilpr, pmatg forest, pmat forest, pmatg forest, FAO forest, soilgroup forest, soilpr forest, maxarea The numbers are the ranking codes. Only the STU’s with the maximum ranking code are selected and if the ranking codes are identical for different STU’s and if these STU’s correspond with the same soil plot, only the largest STU has been selected . 2 Forest1, 2: Primary/ secondary land use is ‘forest’, according to the SGDBE. FAO: FAO code of 1990 is the same in both databases. FAOg : FAO group names are the same in both databases. Pmat: Parent material-code is the same in both databases. Pmatg: Parent material group-code is the same in both databases. Soilpr: Maximum number of corresponding soil properties in both databases. Soilgroup: Soil group-code is the same in both databases. Maxarea: STU with the largest share within buffer This approach can be used as a reference to be compared with the “ normal” approach to get an idea about the amount of STU’s which have been wrongly selected, according to the attribute “ parent material” . This approach assigns a higher priority to the corresponding soil properties rather than the corresponding land use attributes of the SGDBE. An STU with identical full FAO code and with land use ≠ ‘forest’ will be selected even if there is another STU, corresponding with the same soil plot, where the FAO group is identical and with land use = ‘forest’. Thus this approach can also be used to get an estimation about the importance of ‘forest’ as land use categ ory. I n what follows we will compare the results for the different approaches according to the attributes "parent material" and "texture": Only those soil plots will be selected, where information for these attributes is available to perform a simulation of STU selection for those FSCDB soil plots where no additional information about texture and parent material is available. The correspondence between the attribute information of the two databases will be calculated for the different approaches, so that conclusions can be made for an optimised selection procedure. 18 Results of comparison between FSCDB Level I forest soil plots and SGDBE 2. Results of comparison betw een FSCDB Level I forest soil plots and SGDBE STU's 2 .1 . Overall correspondence according to FAO soil code and land use 2.1.1. Simple approach Figure 10 Map of soil plots with corresponding STU's, assigned by simple approach 19 Results of comparison between FSCDB Level I forest soil plots and SGDBE 0 illustrates that with the simple approach for only 41% of all soil plots corresponding STU's are found where at least the FAO name is similar (the group name is identical). The map shows large regional differences. Low correspondences between both databases occur in e.g. Switzerland, France and Flanders. High correspondences occur in Finland, Slovak Republic, Sweden, Wallonie, … The reason for these differences is the homogeneity/ heterogeneity of the STU's. The more STU's with different FAO name and land use properties are corresponding with the same soil plot, the lower the chance becomes that the largest STU corresponds with the properties of the soil plot. Higher correspondences will be found in areas with more homogeneous STU's. Another reason for higher correspondences might be dependence between both data sources, in cases where information for the FSCDB soil plots might have been derived from the SGDBE. 2.1.2. Normal direct approach Figure 11 Map of soil plots with corresponding STU's, assigned by normal direct approach 20 Results of comparison between FSCDB Level I forest soil plots and SGDBE With the normal direct approach, illustrated by Figure 11, 76% of all soil plots with information about FAO names , corresponding STU's where at least the FAO name is similar are found. The reasons for high or low correspondence are the same as for the normal buffered approach (see below) but here an additional source for low correspondence is the lack of positional accuracy. 2.1.3. Normal buffered approach Figure 12 presents the degree of correspondence when the normal buffered approach is used. Figure 12 Map of soil plots with corresponding STU's, assigned by normal buffered approach 21 Results of comparison between FSCDB Level I forest soil plots and SGDBE When the soil type is used as the prime criterion for correspondence, and when the low positional accuracy is taken into account, for 88% of all soil plots with information about FAO names, corresponding STU's where at least the FAO name is similar are found. The reason for high/ low correspondences is the homogeneity/ heterogeneity of the landscapes. Areas with low correspondences between both maps are characteristic for landscapes which are too fragmented to surface on a map with scale 1: 1,000,000. The correspondence improves when landscapes are less fragmented. The map generalisation is the main cause for a lack of correspondence. As mentioned before, high correspondences may also be attributed to dependence between the data sources of both databases, in cases where information for the FSCDB soil plots might have been derived from the same base maps as for the SGDBE. 22 Results of comparison between FSCDB Level I forest soil plots and SGDBE 2.1.4. Advanced approach The result of the advanced approach, in which more detailed soil information is taken into account, is illustrated by Figure 13. Figure 13 Map of soil plots with corresponding STU's, assigned by advanced approach 23 Results of comparison between FSCDB Level I forest soil plots and SGDBE Taking information about parent material and secondary soil properties into account as an additional check for similarity leads to a somewhat lower percentage of correspondence than is the case when these data are not used (normal buffered approach). For 85 % of all soil plots with information about the FAO-classification, corresponding STU's where at least the FAO name is similar are found. 2.1.5 Comparison of different approaches Figure 14 shows the percentage of all soil plots with FAO name information where at least FAO group names are corresponding in both databases for the different approaches and for different degrees of correspondence. % of selected Level I soilplots for different selection procedures vs. class of correspondence with SGDBE Simple Normal/direct Normal/buffered Advanced 100 90 80 70 60 50 40 30 20 10 0 forest1/fao forest1/faogr forest2/fao forest2/faogr forest0/fao forest0/faogr total Figure 14 Percentage of selected soil plots vs. class of correspondence ("forest 1/2" = primary/secondary land use, "forest 0" = no forest “fao” = full FAO name is corresponding, “faogr” = only FAO group name is corresponding) By applying the simple approach, for not more than 2044 of all 4993 soil plots (41% ), STU's can be selected where at least the soil group is corresponding. This percentage increases to 76% and 89% for the direct and buffered approach respectively. This is a measurement for the overall heterogeneity of soil plots and STU's, within the same buffer: I f all soil plots and STU's within the same buffer would have the same basic soil properties (i.e. identical or similar FAO-codes), there would be no difference between the simple and the normal buf fered approach. By searching STU's within a buffer of 5000 m, instead of performing a direct overlay, corresponding STU's can be selected for 622 additional soil plots for which no corresponding STU has been found by performing a direct overlay. This advantage of the buffered approach is illustrated by Figure 15. 24 Results of comparison between FSCDB Level I forest soil plots and SGDBE weighted change of corresponding soilplots: ((direct - buffered)/direct)*100 relative change of number of corresponding soilplots: Buffered vs. Direct overlay 100 90 80 70 60 50 40 30 20 10 0 SW BE FI CZ EL PO AU SR SL BU PO NO ES RO DL DK HU FR NL IR CR ET LA UK LI CH LX Figure 15 Histogram of the effect of application of buffered approach vs. direct approach I n some countries the buffered approach leads to an increase of selected soil plots with corresponding STU's of more than 100% (Switzerland, Lithuania, …). This does not mean that the reliability of this approach is higher than the direct approach, since searching within a buffer leads automatically to a better 'fit' because the number of selectable STU's increases. For each soil plot a corresponding STU can be found, where FAO names are identical and where primary land use is 'forest', as long as the buffer distance is chosen wide enough. As we will see in chapter 2.2 the 'fit' according to additional attributes (parent material) is lower for the buffered approach than for the direct approach, which is an indicator that the buffer range has been chosen too wide. For the advanced approach the percentage of corresponding STU's, according to FAO code properties, is lower than for the normal buffered approach, since for the advanced approach STU's with corresponding parent material codes have been selected prior to STU's with corresponding similar FAO code properties. The differences between the advanced and the normal buffered approach are an indicator for the simultaneous correspondence of FAO code properties and parent material properties: For 170 of all soil plots (that is 8% percent of all soil plots with parent material information) no STU can be found where at least the FAO group names are corresponding and parent material groups are corresponding at the same time. The simultaneous correspondence of the two attributes "soil name" and "parent material" is low in several countries, e.g. France, Norway, Austria,… 2.2. Correspondence according to information about parent material 2.2.1. Overall correspondence Since the full parent material codes could not be translated into comparable codes for both databases, only parent material groups have been used for further analysis. Within each group of correspondence, according to FAO names and land use, the soil plots with available parent material information have been selected. The share of corresponding information in both databases has been calculated for the different approaches. The presence of 'forest' as secondary land use category is no guarantee for better results. This can be concluded from the fact that the correspondences between both databases are even higher within the cluster of selected STU's where 25 Results of comparison between FSCDB Level I forest soil plots and SGDBE land use is not 'forest' than in cases where secondary land use is 'forest', if the FAO code is identical in both databases. There is a significant difference in correspondence according to parent material information between the group of soil plots with corresponding STU's with primary land use = 'forest' and STU's where primary land use is not 'forest'. % of corresponding parent material groups for different selection procedures vs. class of correspondence with SGDBE Normal/direct Normal/buffered Advanced 100 75 50 25 0 forest1/fao forest1/faogr forest2/fao forest2/faogr forest0/fao forest0/faogr total Figure 16 Correspondence of parent material group vs. soil map properties for different approaches("forest 1/2" = primary/secondary land use, "forest 0" = no forest, “fao” = full FAO name is corresponding, “faogr” = only FAO group name is corresponding As Figure 16 shows, the correspondence is somewhat higher for the normal direct approach than for the normal buffered approach. This can be explained by the fact that by searching STU's within a distance of 5000m STU's with identical FAO codes are selected prior to STU's with similar FAO codes, even if the distance to the STU with identical FAO code is higher so that there is a risk for the selection of the "wrong" STU. This effect is compensated by the amount of corresponding STU's for soil plots for which no STU has been selected by applying the direct approach. 26 Results of comparison between FSCDB Level I forest soil plots and SGDBE 2.2.2. Normal buffered approach The result of the normal bufferd approach is illustrated by Figure 17. Figure 17 Map of soil plots with parent material information and corresponding STU's, assigned by normal buffered approach 27 Results of comparison between FSCDB Level I forest soil plots and SGDBE The highest correspondences according to parent material groups have been observed within the group of soil plots with corresponding STU's where the full FAO code is identical and where primary land use is 'forest'. The second highest correspondences have been observed within the group where the FAO names are similar and where primary land use is 'forest' .The third (and last) group with high correspondences, according to the attribute 'parent material', is the group of plots with identical full FAO names in both databases, where primary land use is 'not forest'. By assigning a ranking code to each level of correspondence in a way that high ranking codes correspond with high levels of correspondence (Table 8), an overall correlation of 0.9610 has been calculated between the percentage of corresponding parent material groups and the levels of correspo ndence, according to soil names and land use (Table 9 ). Table 8 Assigned ranking codes for different levels of correspondence between FSCDB and SGDBE Ranking code Correspondence between Soil plot properties and STU properties 6 5 4 3 2 1 Full FAO name and primary land use is 'forest' FAO group name and primary land use is 'forest' Full FAO name only FAO group name only Primary land use is 'forest' None Table 9 Correlation between level of correspondence and percentage of corresponding parent material groups level of correspo ndence Total number of soil plots with parent material information Number of soil plots with corresponding parent material groups Percentage of corresponding parent material groups Ranking code none forest Fao group Fao Faogroup/ f1 32 325 295 355 464 13 152 139 231 340 40,625 46,76923 47,11864 65,07042 73,27586 1 2 3 4 5 Fao/ f1 649 484 74,57627 6 The calculated correlation is significant for p < 0.005. For 64 % (1359)of all soil plots corresponding STU's with corresponding parent material codes are found. 28 Results of comparison between FSCDB Level I forest soil plots and SGDBE 2.2.3. Advanced approach: The soil plots with parent material information and corresponding STU's, as assigned by the advanced approach is given in Figure 18. Figure 18 Map of soil plots with parent material information and corresponding STU's, assigned by advanced approach 29 Results of comparison between FSCDB Level I forest soil plots and SGDBE 1671 of 2120 soil plots with parent material information (79 % ), where parent material group is identical in both databases, have been selected. This value of correspondence is considered to be the maximum possible degree of correlation between the two databases according to parent material group. The difference of correspondence between the normal and the advanced approach is 15% . This means that for 15 % of all soil plots with parent material information (318 plots) no STU can be found where at least the FAO group name and the parent material group names are corresponding simultaneously. 2.2.4. Correspondence per country We might expect that there is a significant correlation between the level of correspondence according to FAO names and land use properties and the percentage of corresponding parent material groups in all countries where parent material information is available. Figure 19 shows that this expectation cannot be confirmed in reality: a significant correlation (> 0.729 at p < 0.05) has been observed for five countries only: Belgium, Denmark, Spain, Hungary and Greece, containing 20% of all soil plots with available parent material information. This means that a high correspondence between both databases, according to FAO names and land use is not correlated with high correspondence, according to the attribute 'parent material group', in all countries. 1.00 0.75 0.50 0.25 0.00 IR UK PO NL DL CH AU LX SR NO FR ET EL HU ES DK BE total -0.25 -0.50 -0.75 -1.00 Figure 19 Histogram of correlation between level of correspondence and percentage of corresponding parent material groups 30 Results of comparison between FSCDB Level I forest soil plots and SGDBE 2.3. Correspondance according to information about texture As shown in Figure 20, there is no correspondence between the level of correspondence, according to FAO names and land use, and information about texture between both databases. Therefore we will not use the attribute texture to validate the correlation between both databases according to this attribute. The overall correspondence is arbitrary: Even a "blind" overlay (selection of STU's at random and independent from their location) would result in a correspondence of 40% . We can only conclude that the information of both databases is likely not to be comparable for all attributes and that we have to be careful before we think about deriving information from the SGDBE and assigning it to FSCDB level I pots. 100 75 50 25 tota l FA On am e/f ore st1 FA On am e/f ore st2 sim ila r/fo res t1 0 sim ila r/fo res t2 % of level I plots with corresponding STU % of corresponding textural classes vs. class of correspondance between FSCDB and SGDBE - direct overlay 100 75 50 25 tot al FA On am e/f ore st1 FA On am e/f ore st2 sim ila r/fo re st1 0 sim ila r/fo re st2 % of level I plots with corresponding STU % of corresponding textural classes vs. class of correspondance between FSCDB and SGDBE - buffered overlay Figure 20 Correspondence of textural class vs. soil map properties for direct and buffered overlay “similar/forest2”: FAO code is similar AND secondary land use is ‘forest’ “similar/forest1”: FAO code is similar AND primary land use is ‘forest’ “FAOname/forest2”: FAO code is similar AND secondary land use is ‘forest’ “FAOname/forest1”: FAO code is similar AND primary land use is ‘forest’ 31 Results of comparison between FSCDB Level I forest soil plots and SGDBE 32 Conclusions 3. Conclusions Based on the results, discussed above, the corresponding STU's should be selected as follows: (1) For each soil plot corresponding SMU's have to be detected within a certain distance, according to the geographical map accuracy of the SGDBE. We handle a distance of 5000m. (2) For each STU, corresponding with the selected SMU's, the size of it's area within the search radius has to be calculated: Percentage of STU within SMU x Area of SMU within bufferdistance. (3) Corresponding STU's have to be selected by the following queries: - I s the FAO code identical in both databases and is primary land use = 'forest' ? I f not: I s the FAO code similar in both databases and is primary land use = 'forest' ? I f not: I s the FAO code identical in both databases ? (4) I f none of the 3 queries mentioned above can be answered positively, a meaningful correspondence cannot be expected, according to additional attributes which might be selected from the SGDBE. I n the final approach STU’s are selected, if they respond to the following queries: Table 10 Assigned ranking codes for different levels of correspondence between FSCDB and SGDBE: final approach Ranking code “ P” for correspondence 6 5 4 3 2 1 Queries Are FAO codes identical AND is primary land use = ‘forest’ ? I f not: next query Are FAO codes similar AND is primary land use = ‘forest’ ? I f not: next query Are FAO codes identical ? I f not: next query Are FAO codes similar ? I f not: next query I s primary land use = 'forest' ? I f not: next query I f none of the queries can be answered positively The result of this approach is represented in Figure 21. 33 Conclusions Figure 21 Map of FSCDB level I soil plots with/without corresponding STU’s according to final approach 34 Conclusions Only for those soil plots, for which corresponding STU's with a ranking code of 4 or higher are found, one can expect to be able to extract valuable information from the SGDBE. % of corresponding parent material groups for different selection procedures vs. class of correspondence with SGDBE 100 fao 75 fao group/f1 fao/f1 50 25 0 Figure 22 Histogram of the final result: Percentage of soil plots with corresponding STU per country ('fao' = full FAO name is corresponding 'fao group/f1' = FAO group name is corresponding and primary land use is 'forest' 'fao/f1' = full FAO name is corresponding and primary land use is 'forest') I t has been proven that a high degree of correspondence, according to FAO names and land use, does not necessarily induce a high degree of correspondence with other attributes. For 69% of all FSCDB soil plots with information about FAO names a corresponding STU can be detected where at least the full FAO name or the FAO group name and land use are corresponding. For these plots we do expect an overall 'fit' of about 72% according to information about parent material groups. The percentage of 'fit' within the group of plots matching with STU's with a lower degree of correspondence, according to FAO names and land use, is only 47% , even if the FAO group names are identical and as long as primary land use is not 'forest'. Whether this degree of 'fit' is valid for other attributes is unsure and it is at least not valid for information about te xture. We may conclude that the SGDBE can only be used as a rough reference system to get an idea about overall tendencies with huge differences in reliability between different countries and even between different areas within the same country. Suggestions for further research: The buffer distance within which corresponding STU's are searched has to be optimised, since the correspondence, according to parent material information, is higher for the direct overlay than for the buffered overlay, if a buffer distance of 5000 m has been maintained. Only the correspondence with parent material information has been examined. As the lack of correspondence according to texture information has shown, the validity of other attributes of selected STU's has to be examined. The reasons for differences in correspondence among the different countries have to be analysed: No distinction can be made between the data quality of the FSCDB and of the SGDBE respectively. High degrees of correspondence on the other hand might be the result of dependence between the two databases, in cases where the same data sources have been used for the gathering of information for both databases. 35 Conclusions 36 Summary Summary The following conclusions can be made for the comparison of the two databases, FSCDB and SGDBE: The comparability of the two data sets is limited by two factors: - the low positional accuracy of the SGDBE, naturally induced by the small map scale of 1:1,000,000, which leads to positional deviations of 500 m – 5000 m. - the high degree of generalisation: each single location of the SGDBE refers to different soil descriptions. Only for those soil plots, where the soil description, according to the FAO codes, is similar in both databases, reliable information can be extracted from the SGDBE. 3845 of the 5289 (73%) level I soil plots are potentially comparable with the SGDBE. Data on parent material and texture are presently unavailable for 60%, respectively 35% of the level I soil plots. The study shows that only information about parent material groups can be extracted from the SGDBE. The reliability of the derived information in this case is 72%. For 2377 of the 3169 level I soil plots (75%) without information about parent material, information about parent material groups can be extracted from the SGDBE. Information about texture can not be extracted from the SGDBE for the selected soil plots, since the reliability of the results is less than 50%. For other additional information, which could be extracted from the SGDBE, such as (e.g.) “depth class of textural change” or “presence of impermeable layer”, no reliability level could be calculated, since no reference data were available. The conclusion is that the SGDBE can only be used as a rough reference system to get an idea about overall tendencies of different soil attributes for those soil plots where these attributes are not described in the FSCDB. Huge differences in reliability of the derived information occurred between different attributes, between different countries and even between different areas within the same country. The most valuable result of the study is the development of a method which allows an overlay of geographical data sets or maps with relatively high positional accuracy (here: the locations of the level I soil plots) and maps with relatively low positional accuracy (here: the SGDBE). 37 Summary 38 Abbreviations, references List of abbreviations FSCC FSCDB SGDBE SMU STU 14 2 29 20 23 24 8 4 9 11 25 15 1 21 7 5 30 22 12 3 19 17 10 16 26 18 13 6 AU BE BU CH CR CZ DK DL EL ES ET FI FR HU IR IT LA LI LX NL NO PL PO RO SL SR SW UK Forest Soil Coordinating Centre Forest Soil Condition Database Soil Geographical Database of Europe Soil Mapping Unit Soil Typological Unit Austria Belgium Bulgaria Switzerland Croatia Czech Republic Denmark Germany Greece Spain Estonia Finland France Hungary I reland I taly Latvia Lithuania Luxembourg Netherlands Norw ay Poland Portugal Romania Slovenia Slovak Republic Sweden United Kingdom References Ball, C. (1998), Chi Square Tutorial, Georgetown (http: / / www.georgetown.edu/ cball/ webtools/ web_chi_tut.html ) Breuning- Madsen and Jones (1995), Soil Profile Analytical Database for the European Union. Danish J. Geogr., 95: 49-58 EC, UN/ ECE and Ministry of the Flemish Community (1997), Vanmechelen, L., R. Groenemans and E. Van Ranst. Forest Soil Condition in Europe. Results of a large soil survey. Technical Report, Brussels, Geneva. 260 p. European Soil Bureau (1996), Metadata – Data about the Soil Geographical Database of Europe, Joint Research Centre, I spra, I taly Klap, J. (2001), I ntegrated study of the influence of soil condition on forest vitality, FSCC, Gent, Belgium Wijvekate, M. L. (1972), Verklarende Statistiek, Utrecht/ Antwerpen 39 Abbreviations, references 40 Annexes Annex 41 Annexes 42 Annexes 1. Translation of FAO74 code into FAO90code: FAO74 A Af Ag Ah Ao Ap B Ba Bc Bcc Bch Bck Bd Bda Bdg Bds Be Bea Bec Bef Beg Bev Bg Bgc Bge Bgg Bgs Bh Bk Bkf Bkh Bkv Bv Bvc Bvk C Ch Chv Ck Ckb Ckc Ckcb Cl D Dd De Dg FAO90 AC ACf ACg ACu ACh ACp CM CMc CMx CMx CMx CMx CMd CMd CMd CMd CMe CMe CMe CMe CMe CMe CMg CMg CMg CMg CMg CMu CMc CMc CMc CMc CMv CMv CMv CH CHh CHh CHk CHk CHk CHk CHl PD PDd PDe PDg NAME74 no informations Acrisols Ferric Acrisols Gleyic Acrisols Humic Acrisols Orthic Acrisols Plinthic Acrisols Cambisols Calcaric Cambisols Chromic Cambisols Calcaro-Chromic Cam bisols Humo-Chromic Cambisols Calci-Chromic Cambisols Dystric Cambisols Ando-Dystric Cambisols Gleyo-Dystric Cambisols Spodo-Dystric Cambisols Eutric Cambisols Ando-Eutric Cambisols Calcaro-Eutric Cambisols Fluvi-Eutric Cambisols Gleyo-Eutric Cambisols Verti-Eutric Cambisols Gleyic Cambisols Calcaro-Gleyic Cambisols Eutri-Gleyic Cambisols Stagno-Gleyic Cambisols Spodo-Gleyic Cambisols Humic Cambisols Calcic Cambisols Fluvi-Calcic Cambisols Humo-Calcic Cambisols Verti-Calcic Cambisols Vertic Cambisols Calcaro-Vertic Cambisols Calci-Vertic Cambisols Chernozems Haplic Chernozems Verti-Haplic Chernozems Calcic Chernozems Vermi-Calcic Chernozems Calcaro-Calcic Chernozems Vermi-Calcaro-Calcic Chernozs Luvic Chernozems Podzoluvisols Dystric Podzoluvisols Eutric Podzoluvisols Gleyic Podzoluvisols NAME90 no informations Acrisols Ferric Acrisols Gleyic Acrisols Humic Acrisols Haplic Acrisols Plinthic Acrisols Cambisols Calcaric Cambisols Chromic Cam bisols Chromic Cam bisols Chromic Cam bisols Chromic Cam bisols Dystric Cambisols Dystric Cambisols Dystric Cambisols Dystric Cambisols Eutric Cambisols Eutric Cambisols Eutric Cambisols Eutric Cambisols Eutric Cambisols Eutric Cambisols Gleyic Cambisols Gleyic Cambisols Gleyic Cambisols Gleyic Cambisols Gleyic Cambisols Humic Cambisols Calcaric Cambisols Calcaric Cambisols Calcaric Cambisols Calcaric Cambisols Vertic Cambisols Vertic Cambisols Vertic Cambisols Chernozems Haplic Cher nozems Haplic Cher nozems Calcic Chernozems Calcic Chernozems Calcic Chernozems Calcic Chernozems Luvic Chernozems Dystric Podzoluvisols Dystric Podzoluvisols Eutric Podzoluvisols Gleyic Podzoluvisols 43 Annexes Dgd Dge Dgs E Ec Eo G Gc Gcf Gcs Gd Gds Ge Gef Geh Ges Gev Gfm Gh Ghf Ghh Ght Gm Gmc Gmf Gms Gs H Hc Hcb Hcf Hcg Hg Hgc Hgs Hgv Hh Hhv Hl Hlv Ho I Ic I ch Id Ie J Jc Jcf Jcg Jd Jdg Je Jef Jeg Jm PDg PDg PDg LPk LPk LPk GL GLk GLk GLk GLd GLd GLe GLe GLe GLe GLe GLu GLu GLu GLu GLu GLm GLu GLm GLu GLa PH PHc PHc PHc PHc PHg PHg PHg PHg PHh PHh PHl PHl GLk LPq LPq LPq LPq LPq FL FLc FLc FLc FLd FLd FLe FLe FLe FLm Dystric Gleyic Podzoluvisols Eutric Gleyic Podzoluvisols Stagno-Gleyic Podzoluvisols Rendzinas Cambic Rendzinas Orthic Rendzinas Gleysols Calcaric Gleysols Fluvi-Calcaric Gleysols Stagno-Calcaric Gleysols Dystric Gleysols Stagno-Dystric Gleysols Eutric Gleysols Fluvi-Eutric Gleysols Histo-Eutric Gleysols Stagno-Eutric Gleysols Verti-Eutric Gleysols Molli-Fluvic Gleysols Humic Gleysols Fluvi-Humic Gleysols Histo-Humic Gleysols Thioni-Humic Gleysols Mollic Gleysols Calcaro-Mollic Gleysols Fluvi-Mollic Gleysols Stagno-Mollic Gleysols Stagnic Gleysols Phaeozems Calcaric Phaeozems Vermi-Calcaric Phaeozems Fluvi-Calcaric Phaeozems Gleyo-Calcaric Phaeozems Gleyic Phaeozems Calcaro-Gleyic Phaeozems Stagno-Gleyic Phaeozems Verti-Gleyic Phaeozems Haplic Phaeozems Verti-Haplic Phaeozems Luvic Phaeozems Verti-Luvic Phaeozems Orthic Phaeozems Lithosols Calcaric Lithosols Humo-Calcaric Lithosols Dystric Lithosols Eutric Lithosols Fluvisols Calcaric Fluvisols Fluvi-Calcaric Fluvisols Gleyo-Calcaric Fluvisols Dystric Fluvisols Gleyo-Dystric Fluvisols Eutric Fluvisols Fluvi-Eutric Fluvisols Gleyo-Eutric Fluvisols Mollic Fluvisols Gleyic Podzoluvisols Gleyic Podzoluvisols Gleyic Podzoluvisols Rendzic Leptosols Rendzic Leptosols Rendzic Leptosols Gleysols Calcic Gleysols Calcic Gleysols Calcic Gleysols Dystric Gleysols Dystric Gleysols Eutric Gleysols Eutric Gleysols Eutric Gleysols Eutric Gleysols Eutric Gleysols Umbric Gleysols Umbric Gleysols Umbric Gleysols Umbric Gleysols Umbric Gleysols Mollic Gleysols Umbric Gleysols Mollic Gleysols Umbric Gleysols Andic Gleysols Phaeozems Calcaric Phaeozems Calcaric Phaeozems Calcaric Phaeozems Calcaric Phaeozems Gleyic Phaeozems Gleyic Phaeozems Gleyic Phaeozems Gleyic Phaeozems Haplic Phaeozems Haplic Phaeozems Luvic Phaeozems Luvic Phaeozems Calcic Gleysols Lithic Leptosols Lithic Leptosols Lithic Leptosols Lithic Leptosols Lithic Leptosols Fluvisols Calcaric Fluvisols Calcaric Fluvisols Calcaric Fluvisols Dystric Fluvisols Dystric Fluvisols Eutric Fluvisols Eutric Fluvisols Eutric Fluvisols Mollic Fluvisols 44 Annexes Jmg Jmv Jt Kk Kkb Kkv Kl Ko L La Lc Lcp Lcr Lcv Ldg Lf Lg Lga Lgp Lgs Lh Lk Lkc Lkcr Lkv Lo Lop Lp Lv Lvc Lvk Mo O Od Odp Oe Ox P Pf Pg Pgh Pgs Ph Phf Pl Plh Po Pof Poh Pp Q Qa Qc Qcc Qcd Qcg FLm FLm FLt KSk KSk KSk KSl KSh LV LVa LVx LVx LVx LVx LVh LVf LVg LVh LVg LVg LVh LVk LVk LVk LVk LVh LVh PTd LVv LVv LVv GRh HS HSf FLd HSf HSi PZ PZf PZg PZg PZg PZc PZc PZb PZb PZh PZh PZh PZh AR ARa ARb ARc ARb ARb Gleyo-Mollic Fluvisols Verti-Mollic Fluvisols Thionic Fluvisols Calcic Kastanozems Vermi-Calcic Kastanozems Verti-Calcic Kastanozems Luvic Kastanozems Orthic Kastanozems Luvisols Albic Luvisols Chromic Luvisols Plano-Chromic Luvisols Rhodo-Chromic Luvisols Verti-Chromic Luvisols Gleyo-Dystric Luvisols Ferric Luvisols Gleyic Luvisols Albo-Gleyic Luvisols Plano-Gleyic Luvisols Stagno-Gleyic Luvisols Humic Luvisols Calcic Luvisols Chromo-Calcic Luvisols Rhodo-Chromo-Calcic Luvisols Verti-Calcic Luvisols Orthic Luvisols Plano-Orthic Luvisols Plinthic Luvisols Vertic Luvisols Chromo-Vertic Luvisols Calci-Vertic Luvisols Orthic Greyzems Histosols Dystric Histosols Placi-Dystric Histosols Eutric Histosols Gelic Histosols Podzols Ferric Podzols Gleyic Podzols Histo-Gleyic Podzols Stagno-Gleyic Podzols Humic Podzols Ferro-Humic Podzols Leptic Podzols Humo-Leptic Podzols Orthic Podzols Ferro-Orthic Podzols Humo-Orthic Podzols Placic Podzols Arenosols Albic Arenosols Cambic Arenosols Calcaro-Cambic Arenosols Dystri-Cambic Arenosols Gleyo-Cambic Arenosols Mollic Fluvisols Mollic Fluvisols Thionic Fluvisols Calcic Kastanozems Calcic Kastanozems Calcic Kastanozems Luvic Kastanozems Haplic Kastanozems Luvisols Albic Luvisols Chromic Luvisols Chromic Luvisols Chromic Luvisols Chromic Luvisols Haplic Luvisols Ferric Luvisols Gleyic Luvisols Haplic Luvisols Gleyic Luvisols Gleyic Luvisols Haplic Luvisols Calcic Luvisols Calcic Luvisols Calcic Luvisols Calcic Luvisols Haplic Luvisols Haplic Luvisols Dystric Plinthosols Vertic Luvisols Vertic Luvisols Vertic Luvisols Haplic Greyzems Histosols Fibric Histosols Dystric Fluvisols Fibric Histosols Gelic Histosols Podzols Ferric Podzols Gleyic Podzols Gleyic Podzols Gleyic Podzols Carbic Podzols Carbic Podzols Cambic Podzols Cambic Podzols Haplic Podzols Haplic Podzols Haplic Podzols Haplic Podzols Arenosols Albic Arenosols Cambic Arenosols Calcaric Arenosols Cambic Arenosols Cambic Arenosols 45 Annexes Qcs Ql Qld Qlg R Rc Rd Rds Re Sg Sm So Sof Th Tm To Tv U Ud V Vc Vcc Vg Vp Vpc Vpg Vpn W Wd Wdv We Wev Xk Xl Xy Z Zg Zgf Zo Zt g p r ARb ARl ARl ARl RG RGc RGd RGd RGe SNg SNm SNh SNh ANu ANm ANh ANz LPu LPd VR VRe VRe VRe VRe VRe VRe VRe PL PLd PLd PLe PLe CLh LXa GYh SC SCg SCg SCh SCh g ATf r Spodo-Cambic Arenosols Luvic Arenosols Dystri-Luvic Arenosols Gleyo-Luvic Arenosols Regosols Calcaric Regosols Dystric Regosols Undefined codes Eutric Regosols Gleyic Solonetzs Mollic Solonetzs Orthic Solonetzs Fluvi-Orthic Solonetzs Humic Andosols Mollic Andosols Ochric Andosols Vitric Andosols Rankers Dystric Rankers Vertisols Chromic Vertisols Calcaro-Chromic Vertisols Gleyic Vertisols Pellic Vertisols Calcaro-Pellic Vertisols Gleyo-Pellic Vertisols Sodi-Pellic Vertisols Planosols Dystric Planosols Verti-Dystric Planosols Eutric Planosols Verti-Eutric Planosols Calcic Xerosols Luvic Xerosols Gypsic Xerosols Solonchaks Gleyic Solonchaks Fluvi-Gleyic Solonchaks Orthic Solonchaks Takyric Solonchaks Glaciers Plaggensols Rock Outcrops Cambic Arenosols Luvic Arenosols Luvic Arenosols Luvic Arenosols Regosols Calcaric Regosols Dystric Regosols Dystric Regosols Eutric Regosols Gleyic Solonetz Mollic Solonetz Haplic Solonetz Haplic Solonetz Umbric Andosols Mollic Andosols Haplic Andosols Vitric Andosols Umbric Leptosols Dystric Leptosols Vertisols Eutric Vertisols Eutric Vertisols Eutric Vertisols Eutric Vertisols Eutric Vertisols Eutric Vertisols Eutric Vertisols Planosols Dystric Planosols Dystric Planosols Eutric Planosols Eutric Planosols Haplic Calcisols Albic Lixisols Haplic Gypsisols Solonchaks Gleyic Solonchaks Gleyic Solonchaks Haplic Solonchaks Haplic Solonchaks Fimic Anthrosols 46 Annexes 2. Synchronisation of parent material codes in both databases: SGDBE-code 110 111 112 113 120 130 131 140 150 200 209 210 211 212 213 214 215 216 217 218 219 220 230 231 232 233 234 240 250 300 310 311 312 313 314 319 320 321 322 323 324 330 331 332 333 340 350 400 410 411 412 SGDBE_NAME River alluvium Old fluviatile deposit (Tertiary) Terraces Lacustrofluvial alluvium Estuarine/ Marine alluvium Glaciofluvial deposits Till Glaciofluvial drift Colluvium Calcareous rocks residuum from calcareous rocks Limestone Primary limestone (Carboniferous) Secondary limestone Tertiary limestone Ferrugineous limestone Hard limestone Soft limestone Marly limestone Chalky limestone Detrital limestone Secondary chalk Marl Secondary marl Tertiary marl Gypseous marl Schistose marl Gypsum Dolomite Clayey materials Old clayey sedimentary deposits Primary clay and sandstone Secondary clay Tert iary clay Pleistocene clay Residuum from old clayey sedimentary deposit Alluvial or glaciofluvial clay Tertiary alluvial clay Glacial clay (Tertiary and Quaternary) Gravelly clay Boulder clay Residual clay from calcareous rocks Clay-with-flints Siderolith formations Calcareous decalcification clay Claystone, mudstone Calcareous clay Sandy materials Old sandy sedimentary deposits Secondary sands Tertiary sands FSCD FSCD-CLASS -1 9 -1 9 -1 9 -1 9 -1 9 93 9 -1 9 -1 9 91 9 -1 4 -1 4 41 4 412 4 -1 4 -1 4 413 4 -1 4 -1 4 -1 4 -1 4 -1 4 -1 4 45 4 -1 4 -1 4 452 4 -1 4 491 4 43 4 -1 7 -1 7 -1 7 -1 7 -1 7 -1 7 -1 7 72 7 -1 7 -1 7 -1 7 723 7 722 7 73 7 731 7 732 7 332 3 711 7 -1 5 -1 5 -1 5 -1 5 47 Annexes 413 414 419 420 421 422 429 430 431 440 441 442 450 451 452 453 454 455 456 457 459 500 510 511 512 513 514 520 521 522 523 530 539 600 610 620 630 640 700 709 710 711 712 719 720 721 722 723 730 731 732 739 740 741 742 743 Flint sands Pleistocene sands Residuum from old sandy sedimentary deposits Alluvial or glaciofluvial sands Glacial sands Sandy gravelly materials Residuum from alluvial or glaciofluvial sand Eolian sands Locally sandcover Coastal sands (Dune sands) Shelly coastal sands Non calcareous coastal sands Sandstone Calcareous sandstone (Macigno) Ferrugineous sandstone (Old Red sandstone) Clayey sandstone Soft quartzy sandstone Hard quartzy sandstone Quartzite Schistose sandstone Residuum from sandstone Loamy materials Residual loam Old loam (Touyas) Stony loam Clay loam Sandy loam Eolian loam Loess Thin loess cover Sandy loess Siltstone Residuum from siltstone Detrital formations Arkose Breccia and Puddingstone Flysch and Molasse Ranas Crystalline rocks and migmatites Residuum from crystalline rocks and migmatit Acid crystalline rocks (and migmatites) Granite Diorite, Quartzodiorite Residuum from acid crystalline rocks Non acid crystalline rocks (and migmatites) Syenite Gabbro Serpentine Crystalline metamorphic rocks Gneiss Embrechites Residuum from crystalline metamorphic rocks Schists Micaschists Slates Shales -1 54 51 -1 -1 523 52 54 541 542 531 532 32 -1 322 323 321 321 13 -1 -1 -1 61 611 612 613 -1 62 621 622 623 331 -1 31 324 312 313 -1 1 -1 1 111 123 -1 -1 121 124 175 -1 14 -1 -1 15 152 163 33 5 5 5 5 5 5 5 5 5 5 5 5 3 4 3 3 3 3 1 3 3 6 6 6 6 6 6 6 6 6 6 3 3 3 3 3 3 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 48 Annexes 744 745 749 750 800 809 810 819 820 821 822 823 824 825 830 900 901 902 910 100 Calcschists Green schists Residuum from schists Other metamorphic rocks Volcanic rocks Residuum from volcanic rocks Acid volcanic rocks Residuum from acid volcanic rocks Basic volcanic rocks Phonolites Basalt Andesite Rhyolite Volcanic tuff Volcanic slag Other rocks Sedimentary rocks Sedimentary, metamorphic and eruptive rocks Organic materials Undifferentiated alluvial deposits (or glacial deposits) 156 154 -1 -1 -1 -1 -1 -1 -1 226 222 221 211 233 -1 -1 -1 -1 -1 -1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 8 9 FSCD-code “-1” indicates that no corresponding parent material-description could be found in the FSCD-parent material-list. 49