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