Wind Erosion Susceptibility of European Soils

Abstract

The EU Thematic Strategy for Soil Protection identified soil degradation caused by erosion as one of the major threats to European soils. A thorough literature review revealed important gaps in research on soil erosion processes in Europe. This is particularly true for wind erosion processes. The current state of the art in erosion research lacks knowledge about where and when wind erosion occurs in Europe, and the intensity of erosion that poses a threat to agricultural productivity. To gain a better understanding of the geographical distribution of wind erosion processes in Europe, we propose an integrated mapping approach to estimate soil susceptibility to wind erosion. The wind-erodible fraction of soil (EF) is one of the key parameters for estimating the susceptibility of soil to wind erosion. It was computed for 18,730 geo-referenced topsoil samples (from the Land Use/Land Cover Area frame statistical Survey (LUCAS) dataset). Our predication of the spatial distribution of the EF and a soil surface crust index drew on a series of related but independent covariates, using a digital soil mapping approach (Cubist-rule-based model to calculate the regression, and Multilevel B-Splines to spatially interpolate the Cubist residuals). The spatial interpolation showed a good performance with an overall R2 of 0.89 (in fitting). We observed the spatial patterns of the soils' susceptibility to wind erosion, in line with the state of the art in the literature. We used regional observations in Lower Saxony and Hungary to ensure the applicability of our approach. These regional control areas showed encouraging results, and indicated that the proposed map may be suitable for national and regional investigations of spatial variability and analyses of soil susceptibility to wind erosion.

Full text

Wind erosion susceptibility of European soils
Pasquale Borrelli ⁎, Cristiano Ballabio, Panos Panagos, Luca Montanarella
European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via E. Fermi, 2749, I-21027 Ispra, VA, Italy
abstractarticle info
Article history:
Received 17 January 2014
Received in revised form 3 June 2014
Accepted 6 June 2014
Available online xxxx
Keywords:
Soil degradation
EU Thematic Strategy for Soil Protection
Wind-erodible fraction of soil
Digital soil mapping
The EU Thematic Strategy for Soil Protection identified soil degradation caused by erosion as one of the major
threats to European soils. A thorough literature review revealed important gaps in research on soil erosion pro-
cesses in Europe. This is particularly true for wind erosion processes. The current state of the art in erosion re-
search lacks knowledge about where and when wind erosion occurs in Europe, and the intensity of erosion
that poses a threat to agricultural productivity. To gain a better understanding of the geographical distribution
of wind erosion processes in Europe, we propose an integrated mapping approach to estimate soil susceptibility
to wind erosion. The wind-erodible fraction of soil (EF) is one of the key parameters for estimating the suscepti-
bility of soilto wind erosion. It was computed for 18,730 geo-referencedtopsoil samples (fromthe Land Use/Land
Cover Area frame statistical Survey (LUCAS) dataset). Our predication of the spatial distribution of the EF and a
soil surface crust index drew on a series of related but independent covariates, using a digital soil mapping ap-
proach (Cubist-rule-based model to calculate the regression, and Multilevel B-Splines to spatially interpolate
the Cubist residuals). The spatial interpolation showed a good performance with an overall R
2
of 0.89 (in fitting).
We observed the spatial patterns of the soils' susceptibility to wind erosion, in line with the state of the art in the
literature. We used regional observations in Lower Saxony and Hungary to ensure the applicability of our approach.
These regional control areas showed encouraging results, and indicated that the proposed map may be suitable for
national and regional investigations of spatial variability and analyses of soil susceptibility to wind erosion.
© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
1. Introduction
Wind erosion is a widespread phenomenon causing serious soil deg-
radation in arid and semi-arid regions (FAO, 1960; Wolfe and Nickling,
1993). In its more severe forms it can constitutes a threat to cropping
and contributes to the degradation of a sustainable cropping agriculture
(Lyles, 1975). The wind induced movement of soil occurs when three
environmental conditions coincide: i) the wind is strong enough to mo-
bilize soil particles, ii) the characteristics of the soil make it susceptible
to wind erosion (soil texture, organic matter and moisture content)
and iii) the surface is mostly devoid of vegetation, stones or snow
(Bagnold, 1941; Nordstrom and Hotta, 2004; Shao, 2008).
Wind erosion has always occurred as a naturalland-forming process
(Livingstone and Warren, 1996) but, today, the geomorphic effects of
wind are locally accelerated by anthropogenic pressures (e.g. leaving
cultivated lands fallow for extended periods of time, overgrazing range-
land pastures and, to a lesser extent, over-harvesting vegetation (Leys,
1999)).
Land degradation due to wind erosion is also an European phenom-
enon (Warren, 2003) which locally affects the semi-arid areas of the
Mediterranean region (Gomes et al., 2003; Lopez et al., 1998; Moreno
Brotons et al., 2009) as well as the temperate climate areas of thenorth-
ern European countries (Bärring et al., 2003; De Ploey, 1986; Eppink and
Spaan, 1989; Goossens et al., 2001). According to the EU Thematic
Strategy for Soil Protection (European Commission, 2006), an estimated
42 million hectares are affected by wind erosion in Europe. However,
the latest investigations within the framework of EU projects (Wind
Erosion on European Light Soils (WEELS)) and Wind Erosion and Loss
of Soil Nutrients in Semi-Arid Spain (WELSONS; Warren, 2003) suggest
that the areas potentially affected by wind erosion may be more wide-
spread than previously reported by the European Environment Agency
(EEA, 1998). Field observations and measurements found that the areas
that the European Environment Agency reported as being only slightly
affected by wind erosion (EEA, 1998) have actually undergone severe
erosion (Böhner et al., 2003; Riksen and De Graaff, 2001). These field
research findings reveal that the European Environment Agency (EEA,
1998) currently has an incomplete picture about the occurrence and
scope of wind erosion in Europe. This could lead to incorrect decision
making by national and European institutions in seeking to mitigate
wind erosion. To fulfil the goal of the EU Thematic Strategy for Soil
Protection (European Commission, 2006), research must aim to better
understand where and under which conditions land degradation by
wind erosion is most likely to occur. The methodologies that are applied
must be harmonised in order to effectively locate the wind erodible
areas in Europe.
Geoderma 232–234 (2014) 471–478
⁎Corresponding author.
E-mail address: pasquale.borrelli@jrc.ec.europa.eu (P. Borrelli).
http://dx.doi.org/10.1016/j.geoderma.2014.06.008
0016-7061/© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
Contents lists available at ScienceDirect
Geoderma
journal homepage: www.elsevier.com/locate/geoderma

This study provided an assessment of the susceptibility of European
soils to wind erosion. It is a key parameter of integrated modelling for
the spatial assessment of the wind erosion risk (Hagen, 2004). The
erodibility of European soil was estimated as the wind-erodible fraction,
a simplification of Chepil's (1941) work (Woodruff and Siddoway,
1965). Soil characteristics were obtained from the first topsoil survey
of the whole European Union (Tóth et al., 2013). The assessment pre-
sented in this paper is part of a preliminary investigation that aims to
further investigate the patterns of soil susceptibility to wind erosion
across Europe, and to research the occurrence of wind erosion at region-
al and European scales.
2. Material and methods
2.1. Study area
The study area was made up of 25 member states of the European
Union. Bulgaria, Romania and Croatia were excluded from the study be-
cause data from their LUCAS soil samples were not available. The total
land surface is about 4 million km
2
, providing living space for a popula-
tion of about 470 million (Eurostat, 2012). According to Eurostat(2012)
two-fifths (about 1.55 million km
2
) of the total land area was used for
agricultural purposes in 2007.
2.2. Soil database
Soil information for the 25 EU member states was acquired from the
Land Use/Land Cover Area frame statistical Survey (LUCAS) database,
which provided data from 2009 onwards. This was combined with a
topsoil assessment component (‘LUCAS-Topsoil’—Tóth et al., 2013).
LUCAS-Topsoil comprises the first harmonised and comparable dataset
on soil at the Europeanlevel. We used a merged databasethat contained
19,967 geo-referenced samples (each of 0.5 kg of topsoil, collected at a
depth of 0–20 cm), which was selected from a subset of 200,000 poten-
tial LUCAS sampling sites. Budgetary constraints did not allow for a
broader sampling exercise. Geostatistical techniques were employed
to sample representative points (Tóth et al., 2013). All 19,967 samples
were analysed for their coarse fragment percentage, particle size distribu-
tion (% clay, silt and sand content), pH value (in CaCl
2
and H
2
O), organic
carbon content (g kg
−1
), carbonate content (g kg
−1
), phosphorous con-
tent (mg kg
−1
), total nitrogen content (g kg
−1
), extractable potassium
content (mg kg
−1
), cation exchange capacity (cmol + kg
−1
) and multi-
spectral properties.
2.3. Computation of the erodible fraction (EF)
In the early 1950s, the combination of soil sieving and wind tunnel
experiments provided evidence of the relationship between soil loss
by wind and the characteristics of the soil surface (Chepil, 1950;
Chepil and Woodruff, 1954). The field observations revealed that aggre-
gates that were larger than 0.84 mm in diameter were non-erodible
under test conditions. As a result of these findings, the proportion of top-
soil aggregates b0.84 mm in diameter (i.e. the wind-erodible fraction (EF)
of the soil) became a commonly accepted and widely applied measure of
soil erodibility by wind (Colazo and Buschiazzo, 2010; Hevia et al., 2007;
Woodruff and Siddoway, 1965), which has been widely employed ever
since in prediction models (Chepil et al., 1962; Woodruff and Siddoway,
1965). Fryrear et al. (1994) developed a multiple regression equation
for computing the erodible fraction of soils based on the soil's texture
and chemical properties (Fryrear et al., 2000):
EF ¼29:09 þ0:31Saþ0:17Siþ0:33Sc−2:59OM−0:95CaCO3
100 ð1Þ
where all variables are expressed as a percentage. S
a
is the soil sand
content, S
i
is the soil silt content, S
c
is the ratio of sand to clay
Table 1
List of environmental covariates used for the spatial interpolation.
Parameter Data source Spatial resolution
Land use andland cover data International Geosphere–Biosphere Programme 1 km
Monthly temperatures (min & max) WorldClim —Global Climate Data v 1.4 1 km
Monthly precipitations WorldClim —Global Climate Data v 1.4 1 km
Satellite imagery
Red, blue, green, near infrared (NIR) and middle infrared (MIF) bands NASA —Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m
Principal Component Analysis of the satellite imagery NASA —Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m
Vegetation indices: Enhanced Vegetation Index (EVI) & Normalized Differenced Vegetation Index (NDVI) NASA —Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m
Principal Component Analysis of the Enhanced Vegetation Index & Normalized Differenced Vegetation Index NASA —Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m
DigitalElevationModel(DEM) NASA SRTM Digital Elevation Database v4 90 m
Multi-resolution valley bottom flatness index DEM derivative 90 m
Slope gradient DEM derivative 90 m
Drainage network DEM derivative 90 m
Altitude above channel network DEM derivative 90 m
Down–slope distance DEM derivative 90 m
Latitude & latitude Coordinate system (ETRS_1989_LAEA) –
472 P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

contents, OM is the organic matter content and CaCO
3
is the calcium
carbonate content.
The study carried out by Fryrear et al. ( 1994) calculated an R
2
of 0.67,
leaving 33% of the erodible fraction variability unexplained. The recent
applications of the equation in Europe and Argentina, however, have re-
vealed certain limits to its transferability (López et al., 2007). López et al.
(2007) stated that the model proposed by Fryrear et al. (1994) did not
fit with the measured EF values. This was attributed to the high CaCO
3
contents of Spanish soils and the low sand/clay ratios and high organic
matter contents of some Argentinean soils. Despite these limitations,
the equation constitutes one of the most robust and widely tested equa-
tions defined in the literature to assess the intrinsic susceptibility of soil
to wind erosion. Therefore, we calculated the wind-erodible fraction of
the soil by using Eq. (1) and the LUCAS soil data. This enabled us to de-
lineate the susceptibility of European soils to wind erosion into spatial
patterns.
The LUCAS dataset contains 19,967 sample points. A set of 1,237
sample points (mainly organic soils) was excluded from the dataset.
Before applying Eq. (1), some of the remaining 18,730 sample points
were modified as following: For the soil samples with an organic matter
content above 4.79% and a CaCO
3
content above 25.2%, the upper limits
of 4.79% (6209 samples; mean = 10%, σ= 5.9%) and 25.2% (1710
samples; mean = 41%, σ= 13.1%) were applied. Minimum (5%, n =
1359) and maximum (70%, n = 99) threshold values were imposed
upon the calculated erodible fraction values where the physical charac-
teristics of the sampling points were outside the equation's validation
limits. Finally, the erodible fraction values were further adjusted to
consider the rock fraction that is not erodible (Zobeck, 1991). This was
accomplished by subtracting the average surface stonecover (%) report-
ed in the LUCAS dataset. Non-erodible surface (such as lakes, glaciers,
bare rocks and urban areas) were described in the map as ‘No Data’.
2.4. Computation of a soil crust index
The impact of raindrops on the soil surface leads to a redistribution
of soil particles and creates a soil surface crust (Belnap, 2003). Depend-
ing on the soil properties, the surface crust may decrease or increase
wind erosion potential (Zobeck, 1991). The soil crust factor (SCF,
Fryrear et al., 2000) was employed to estimate the influence of the soil
crust occurrence on the wind erosion susceptibility of European soils.
This is an empirical relationship which was developed using laboratory
wind tunnel tests on the resistance of soil aggregates and crusts to
windblown sand of Hagen et al. (1992). According to Fryrear et al.
(2000), the soil crust factor was developed by regressing the soil crust
Fig. 1. Mapof wind erosion susceptibility of European soils (500 m spatial resolution)based on the estimationof the wind-erodiblefraction of soil (EF) (Chepil, 1941; Fryrear et al., 2000).
The map was obtained by interpolating the EF values (Chepil,1941; Fryrear et al., 2000) calculated for 18,730 geo-referenced topsoilsamples (Land Use/Land Cover Area framestatistical
Survey —LUCASdataset). For the interpolation, a Cubist-rule-based model was used for theregression, and a Multilevel B-Splinesfor the spatial interpolation of the Cubist residuals. The
geographical extent of this study includes 25 member states of the European Union. Bulgaria, Croatia and Romania were not included as the LUCAS-Topsoil database currently does not
include them. Non-erodible surfaces (such as lakes, glaciers, bare rocks and urban areas) were described as No Data.
473P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

factor, as determined by the abrasion coefficient, on clay (clay) and
organic matter (OM):
SCF ¼1
1þ0:0066 clay
ðÞ
2þ0:21 OM
ðÞ
2

:
For the soil samples with an organic matter content of above 4.79%,
an upper cut-off point of 4.79% was applied (6209 samples; mean =
10%, σ= 5.9%). With regard to the clay content, an upper limit of
39.3% (1481 samples; mean = 48.2%; σ= 7.44) and a lower limit of
4.9% was used (1428; mean = 2.47%; σ=0.63).
2.5. Spatial prediction of the wind-erodible fraction and surface crust factor
The digital soil mapping approach was employed to calculate the
wind-erodible fraction of soil and soil surface crustingfactor by deriving
its distribution from a series of related but independent covariates
(Goovaerts, 1998). This approach aims to establish a statistical rela-
tionship between the property to be calculated and a set of spatially
exhaustive covariates. Once this relationship is found, the dependent
property is estimated everywhere within the geographic frame of interest
(Goovaerts, 1998).
Table 2
Descriptive statistics of the wind-erodible fraction of soil for European countries.
Country Mean
[%]
Maximum Standard deviation Coefficient of variation
Austria 27.2 46.5 3.6 13.3
Belgium 32.0 61.2 6.9 21.6
Cyprus 18.5 45.2 5.7 30.7
Czech Republic 30.8 54.5 4.6 14.9
Denmark 41.1 61.4 5.7 13.8
Estonia 38.3 61.6 5.8 15.1
Finland 38.6 67.0 8.0 20.7
France 24.4 60.5 7.5 30.8
Germany 35.0 69.0 10.2 29.2
Greece 23.7 61.3 7.3 30.9
Hungary 30.9 64.9 7.6 24.5
Ireland 27.9 41.7 2.8 10.1
Italy 22.0 52.6 6.0 27.4
Latvia 40.1 62.4 5.2 12.9
Lithuania 39.3 62.5 5.5 14.1
Luxembourg 24.6 37.9 3.1 12.6
Malta 21.8 36.7 4.8 22.0
Netherlands 40.6 67.1 9.8 24.1
Poland 45.2 68.8 8.4 18.5
Portugal 25.5 67.8 9.3 36.6
Slovakia 26.2 53.3 4.6 17.8
Slovenia 23.3 44.6 5.1 22.0
Spain 20.4 58.6 7.6 37.3
Sweden 34.5 63.8 6.2 17.9
United Kingdom 27.7 54.9 4.0 14.6
Fig. 2. The average wind-erodible fraction of soil (EF), according to European Union NUTS-3 administrative units.
474 P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

Fig. 3. Comparison of thepredicted wind erosion susceptibility ofsoil (backgroundraster image) withregional observations (represented withyellowish lines).a) The Geest areain Lower
Saxony.This area mainly consists of glacial moraines and sand plains,forming light sandysoils largely endangered by wind erosion (Capelle,1990; Gross and Schäfer,2004, among others).
b) Area affected by wind erosion in Hungary according to Stefanovits and Várallyay (1992).
Fig. 4. Map of the soil crust factor of European soils (500 m spatial resolution).
Based on the work of Fryrear et al. (2000) and Hagen et al. (1992).
475P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

In our study, the value of the wind-erodible fraction of soil and soil
surface crusting factor that were calculated for the LUCAS points were
interpolated using a series of environmental descriptors (covariates),
in order to map its spatial distribution. Regression residuals were then
spatially interpolated according to their covariance function. In general,
this kind of model can be described as:
^
zs0
ðÞ¼
^
ms0
ðÞþ
^
εs0
ðÞ ð2Þ
where ^
ms0
ðÞrepresents the deterministic part fitted by the regression
model and ^
εs0
ðÞrepresents the interpolated residuals.
The two components were then summed to obtain the final estima-
tion of the erodible fraction. This hybrid approach was performed using
the Cubist-rule-based model (Quinlan, 1992) to carry out the regres-
sion, and Multilevel B-Splines (MBS; Lee et al., 1997) to spatially inter-
polate the Cubist residuals. The Cubist model was fitted by k-fold cross
validation in order to evaluate the best combination of committees
and neighbours.
In terms of accuracy and lack of bias, the Multilevel B-Splines algo-
rithm performs as well as the kriging. It is also computationally faster
and allows for an easy estimation of the smoothness of the interpolated
field. In our case, an optimal smoothness was estimated using General-
ized Cross Validation (GCV) (Craven and Wahba, 1979)bytrading
model complexity for prediction error.
Various covariates were considered for the Cubist model (Table 1).
Two main types were considered to be the most appropriate: i) Re-
motely sensed data, derived from the Moderate Resolution Imaging
Spectroradiometer (MODIS), including vegetation indices (Normalized
Differenced Vegetation Index —NDVI, Enhanced Vegetation Index —
EVI) and raw band data which were re-projected using Principal
Component Analysis (PCA). This data comprised the full cycle of
yearly observations of MODIS. ii) Terrain features, derived from the
SRTM Digital Elevation Model (Jarvis et al., 2008), including common
geomorphometric descriptors (including slope, altitude above channel
base level, multi-resolution index of valley bottom flatness).
2.6. Evaluation of the outcomes
A cross validation was carried out to evaluate the performance of the
spatial prediction approach. The extremely limited number of studies
that report soil erodible fraction estimations (Fryrear et al., 1994)
or similar types of soil erodibility by wind assessment (Gross and
Schäfer, 2001) did not allow for the application of further validation
procedures for the calculated values of soil erodibility. We ensured
that our results were consistent with theoretic expectations. Further-
more, we compared our findings with previous studies (Bärring et al.,
2003; Eppink and Spaan, 1989; Gross and Schäfer, 2001; Huber et al.,
2008; Kertész and Centeri, 2006; Stefanovits and Várallyay, 1992) for
the geographical areas where soil susceptibility to wind erosion had
been reported (i.e., Geest area of Lower Saxony, Southern Great Plains
of Hungary and the Dutch provinces of Groningen and Drenthe).
3. Results and discussions
3.1. Soil susceptibility to wind erosion
We estimated the wind-erodible fraction based on the
15. 786 million cells (500 m spatial resolution) into which we subdivided
the surface of the 25 EU countries (Fig. 1). The resulting erodible fraction
values ranged from 3.6% to 69.0%, with a mean value of 30% (σ10.6%). Ac-
cording to the erodibility classification proposed by Shiyatyi (1965),
which has been adopted for European contexts by López et al. (2007),
81.3% (EF b40%) and 13.8% (EF ≥40% and b50%) of the investigated
area are characterised by slight and moderate erodibility, respectively,
whereas 4.9% are characterised by high erodibility (EF ≥50%). As can
be inferred from Fig. 1, the distribution of the spatial wind-erodible frac-
tion patterns suggests a division of the European surface into three re-
gions: i) a north region mostly dominated by the highest EF values, ii) a
central eastern region with average EF values interspersed with some
high/low spots, and iii) the Mediterranean area, which has mainly low
wind-erodible fraction values.
A cross-country comparison of the mean erodible fraction values
(Table 2 —National level; Fig. 2 —NUT-3 level (Nomenclature of Terri-
torial Units for Statistics —Eurostat, 2013)) confirms the regional struc-
ture. The Mediterranean countries (Cyprus, Spain, Malta and Italy) have
the lowest average erodible fraction values (18.5% to 22%). The highest
values appear in the areas surrounding the North Sea and the Baltic Sea,
with Poland, Denmark, the Netherlands and northern Germany show-
ing average values of above 40%. The higher coefficient of variation
values (c
v
) in the southern countries (Table 2) confirms the regional
pattern. These patterns show a spatial distribution of the soils' erodibil-
ity by wind erosion which is clearly distinct from the soil erodibility pat-
tern identified for water erosion (K-factor, Panagos et al., 2014). While
this generally applies across Europe, the situation is particularly true
for the northern part of central Europe, France and Spain.
Fig. 5. The twenty most important covariates and their relative importance in the application of Cubist/MBS model for wind-erodible fraction of soil (EF) prediction. (i) Latitude;
(ii) altitude above channel network; (iii) elevation; (iv) PCAb1 of Red band; (v) PCAb3 of NIR band; (vi) PCAb1 MIR; (vii) PCAb3 MIR; (viii) MIR; (ix) PCAb1 EVI; (x) PCAb5 MIR;
(xi) drainage network; (xii) land use and land cover; (xiii) PCAb2 NIR; (xiv) longitude; (xv) multi-resolution valley bottom flatness index; (xvi) slope gradient; (xvii) PCAb2 Red;
(xviii) NIR; (xix) PCAb4 MIR; and (xx) PCAb2 MIR.
476 P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

The academic literature on soil degradation in Europe identified
wind erosion as a major threat to northern Europe (Warren, 2003).
This is because the phenomenon has especially significant effects on
light sandy soils (Bärring et al., 2003; Eppink and Spaan, 1989;
Goossens and Gross, 2002; Riksen and De Graaff, 2001). The sandy
soils of northern Europe often show a bimodal grain-size distribution
and a secondary maximum in the silt range. Further observations of
their erosion susceptibility patterns confirmed that the sandy soils of
northern Europe are characterised by a higher susceptibility to wind
erosion. Fig. 3, which shows the Geest area of Lower Saxony (Gross
and Schäfer, 2004), provides an example of areas that are highly suscep-
tible to wind erosion. As indicated by the wind tunnel experiments of
Gross and Schäfer (2001), the region, which mainly consists of glacial
moraines and sand plains, is predominantly covered by sandy soils
and is thus highly susceptible to wind erosion. Our results reflect re-
gional soil erodibility dynamics very well. In fact, Fig. 3a shows high
erodibility values for the Geest areas (an average 48% wind-erodible
fractionof soil), intermediate valuesfor the loess at the Geest's southern
boundary(38%), and lower values along the coast and the southern area
of Lower Saxony (30%). Further proof of the good correlation between
our results and the land susceptibility to wind erosion described in the
literature (Kertész and Centeri, 2006; Stefanovits and Várallyay, 1992)
is reported in Fig. 3b. For Hungary, the figure illustrates the degree to
which the areas are affected by wind erosion processes. The findings
of Stefanovits and Várallyay (1992) coincide extremely well with our
predictions. Further encouraging results were obtained from compari-
sons with studies carried out in south-east England (Huber et al.,
2008) and the Dutch provinces of Groningen and Drenthe (Eppink
and Spaan, 1989).
3.2. Soil crust factor
Surface map of soil crust factor (SCF) is shown in Fig. 4. The soil
crust factor values range between 0.02 and 1, with an average value of
0.39 (σ= 0.18). Similar to the wind-erodible fraction (EF), the spatial
pattern of soil crust factor shows higher values in the northeast region.
The soil crust factor values tend to decrease moving towards the south-
west direction. The sandy soils characterizing the glacial deposits of the
northern countries (Denmark, Germany, Netherlands, Scandinavia and
Baltic area) and the soils with a significant percentage of sand (the
north-western Iberian Peninsula, the French region of Limousin and
the coastline of Aquitaine along the Atlantic Ocean) are less affected
by the formation of a soil surface crust. Here, the soils are more easily
eroded by wind as the raindrop-impacted soil surface is aerodynamical-
ly smoother than the cloddy surface before the rain (Belnap, 2003;
Fryrear et al., 2000). By contrast, the soils with high clay content (e.g.,
Sicily, Andalicia) led to the formation of a resistant soil surface crust
that effectively limits the erosive power of the wind.
3.3. Evaluation of the outcomes
Following the approach described in Section 2.5, the wind-erodible
fractionof soil was modelled using a seriesof environmental descriptors
(covariates) and theCubist model, and by interpolating the Cubist resid-
uals using the MultilevelB-Splines. The Cubist method automaticallyse-
lects the mostinformative covariates (Fig. 5). The proposed methodwas
therefore able to predict the distribution of the wind-erodible fraction of
soil with a good performance (R
2
= 0.5) and an RMSE = 10.1 in a k-fold
cross validation. The interpolation by Multilevel B-Splines further in-
creased the prediction performance up to an R
2
of 0.89 (in fitting).
3.4. Data availability
The European maps of the wind-erodible fraction of soil and soil crust
factor are available on the European Soil Data Centre (ESDAC) web plat-
form (Panagos et al., 2012). They can be downloaded free of charge in
GeoTIFF raster format (http://esdac-catalog.jrc.ec.europa.eu/)inorderto
encourage further regional and pan-European investigations into spatial
variability and analysis of soil susceptibility to wind erosion.
4. Conclusions
The elaboration of theLUCAS-Topsoil data combined with digital soil
mapping techniques allowed for the assessment of soil susceptibility to
wind erosion at a European scale. This constitutes, to the best of our
knowledge, the first comprehensive study of its kind. The accuracy as-
sessment confirmed thegood performance of the Cubist method. By in-
terpolating approximately 20,000 wind-erodible fraction values of soil,
we reproduced spatial patterns of soil susceptibility to wind erosion in
line with the published literature. The distribution of light sandy soils
which frequently suffer from degradation due to wind erosion was spa-
tially described and illustrated. In addition, regional observations gave
encouraging results with regard to the reliability of the study outcomes
and its suitability for local-scale applications. The use of the LUCAS
dataset allowed us to take a significant step towards: i) the develop-
ment of spatial analysis of soil susceptibility to wind erosion, ii) the
development of an integrated GIS-based risk assessment of wind erosion
at regional and national scales, and iii) the better identification of areas
that are susceptible to wind erosion. These insights will help to identify
areas that are at risk of wind erosion, and for which conservation mea-
surementssuchasshelterbeltsorwindbreaks(Jönsson, 1994)shouldbe
considered by decision makers.
5. Ongoing and planned future research
This communication describes research carried out on the suscepti-
bility of European soils to wind erosion. Drawing on these insights,fur-
ther studies on land susceptibility to wind erosion processes need to be
pursued. As the physics of wind erosion is complex, soil as well as atmo-
spheric and land-surface processes must be taken into account in order
to assess the wind erosion susceptibility of European soils. Therefore,
our ongoing and planned research activities focus on the development
of integrated modelling approaches that aim to spatially define i) the
EU land surface that is susceptible to wind erosion, and ii) the arable
lands thatare subject to soil degradation processes. To achieve these ob-
jectives, two different modelling approaches are undertaken (a pixel-
and an object-oriented model).
Acknowledgements
This project was funded by the Joint ResearchCentre of the European
Commission in the context of the action nr. 22004 “Soil Data and Infor-
mation Systems”. We also wish to thank Grainne Mulhern for English-
language editing.
References
Bagnold,R.A., 1941. The Physics of Blown Sand and DesertDunes. Methuen and Company,
London, p. 265.
Bärring, L., Jönsson, P., Mattsson, J.O., Åhman, R., 2003. Wind erosion on arable land in
Scania, Sweden and the relation to the wind climate: a review. Catena 52, 173–190.
Belnap, J., 2003. Biological soil crusts and wind erosion. In: Belnap, J., Lange, O.L. (Eds.),
Biological Soil Crusts: Structure, Function, and Management. Springer, Berlin Heidelberg,
pp. 339–347.
Böhner, J., Schäfer, W., Conrad, O., Gross, J., Ringeler, A., 2003. The WEELS model:
methods, results and limitations. Catena 52, 289–308.
Capelle, A., 1990. Die erosionsgefährdete Landesfläche in Niedersachsen und Bremen. Z.
Kult. Landentwickl. 31, 11–17.
Chepil, W.S., 1941. Relation of rind erosion to the dry aggregate structure of a soil. J. Sci.
Food Agric. 21, 488–507.
Chepil, W.S.,1950. Properties of soil whichinfluence wind erosion: I. Thegoverning prin-
ciple of surface roughness. Soil Sci. 69, 149–162.
Chepil, W.S., Woodruff, N.P., 1954. Estimations of wind erodibility of field surfaces. J. Soil
Water Conserv. 9, 257–265.
Chepil, W.S., Siddoway, F.H., Armbrust, D.V., 1962. Climatic factor for estimating wind
erodibility of farm fields. J. Soil Water Conserv. 17, 162–165.
477P. Borrelli et al. / Geoderma 232–234 (2014) 471–478

Colazo, J.C., Buschiazzo, D.E., 2010. Soil dry aggregate stability and wind erodible fraction
in a semiarid environment of Argentina. Geoderma 159, 228–236.
Craven, P., Wahba, G., 1979. Smoothing noisy data with spline functions. Numer. Math.
31, 377–403.
De Ploey, J., 1986. Bodemerosie in de lage landen, een Europees milieuprobleem. Acco,
Leuven.
Eppink, L.A.A.J., Spaan, W.P., 1989. Agricultural wind erosion control measures in the
Netherlands. Soil Technol. Ser. 1, 1–13.
European Commission, 2006. Theamtic Strategy for Soil Protection. COM 2006, p. 231.
European Environment Agency, 1998. Europe's Environment: the Second Assessment.
Elsevier, United Kingdom, (293 pp.).
Eurostat, 2012. Population and population change statistics. [on line] URL: http://epp.
eurostat.ec.europa.eu/statistics_explained/index.php/Population_and_population_
change_statistics (accessed October 2013).
Eurostat, 2013. Nomenclature of territorial units for statistics. [on line] URL: http://epp.
eurostat.ec.europa.eu/portal/page/portal/nuts_nomenclature/introduction (accessed
October 2013).
Food and Agriculture Organization of the United Nations (FAO), 1960. Soil erosion by
wind and measures for its control on agricultural lands. FAO Agricultural Develop-
ment Paper No. 71.
Fryrear, D.W., Krammes, C.A., Williamson, D.L., Zobeck, T.M., 1994. Computing the wind
erodible fraction of soils. J. Soil Water Conserv. 49, 183–188.
Fryrear,D.W., Bilbro, J.D.,Saleh, A., Schomberg, H.M., Stout,J.E., Zobeck, T.M.,2000. RWEQ:
improved wind erosion technology. J. Soil Water Conserv. 55, 183–189.
Gomes, L., Arrue, J.L., Lopez, M.V., Sterk, G., Richard, D., Gracia, R., Sabrea, M., Gaudicheta,
A., Frangid, J.P., 2003. Wind erosion in a semiarid agricultural area of Spain: the
WELSONS project. Catena 52, 235–256.
Goossens, D., Gross, J., 2002. Similarities anddissimilaritiesbetween the dynamicsof sand
and dust during wind erosion of loamy sandy soil. Catena 47, 269–289.
Goossens, D., Gross, J.,Spaan, W., 2001. Aeolian dust dynamicsin agricultural land areas in
lower Saxony, Germany. Earth Surf. Process. Landf. 26, 701–720.
Goovaerts, P., 1998. Geostatistical tools for characterizing the spatial variability of micro-
biological and physico-chemical soil properties. Biology and Fertility of soils 27,
315–334.
Gross, J., Schäfer, W., 2001. Erodibility (wind-tunnel investigations). In: Warren, A. (Ed.),
Wind Erosion on European Light Soils (WEELS). Final Report to the European Union,
pp. 12–28.
Gross, J., Schäfer, W., 2004. Quantification of erosion-induced dust emissions: develop-
ment of an application-oriented method. Wind Erosion and Dust Dynamics: Observa-
tions, Simulations, Modelling, p. 41.
Hagen, L.J., 2004. Evaluation of the Wind Erosion Prediction System (WEPS) erosion
submodel on cropland fields. Environ. Model Softw. 19, 171–176.
Hagen, L.J., Skidmore, E.L., Saleh, A., 1992. Wind erosion: prediction of aggregate abrasion
coefficients. Transactions of the ASAE 35, 1847–1850.
Hevia, G.G., Mendez, M.,Buschiazzo, D.E.,2007. Tillage affectssoil aggregation parameters
linked with wind erosion. Geoderma 140, 90–96.
Huber, S., Prokop, G., Arrouays, D., Banko, G., Bispo, A., Jones, R.J.A., Kibblewhite, M.G.,
Lexer, W., Möller, A., Rickson, R.J., Shishkov, T., Stephens, M., Toth, G., Van den
Akker, J.J.H., Varallyay, G., Verheijen, F.G.A., Jones, A.R., 2008. Environmental Assess-
ment of Soilfor Monitoring: Volume I, Indicators & Criteria. Office for theOfficial Pub-
lications of the European Communities, Luxembourg.
Jarvis, A., Reuter, H.I., Nelson, A., Guevara, E., 2008. Hole-filled SRTM for theglobe version
4. [on line] URL: http://srtm.csi.cgiar.org (accessed October 2013).
Jönsson, P., 1994. Influence of shelter on soil sorting by wind erosion —a case study.
Catena 22, 35–47.
Kertész, A., Centeri, C., 2006. Hungary. In: Boardman, J., Poesen, J. (Eds.), Soil Erosion in
Europe. Wiley, Chichester, pp. 139–153.
Lee, S., Wolberg, G., Shin, S.Y., 1997. Scattered data interpolation with Multilevel
B-splines. IEEE Trans. Vis. Comput. Graph. 3, 229–244.
Leys, J.F., 1999. Wind erosion on agricultural land. In: Goudie, A.S., Livingstone, I., Stokes,
S. (Eds.), Aeolian Environments, Sediments and Landforms. John Wiley & Sons Ltd.,
Chichester, pp. 143–166.
Livingstone, I., Warren, A., 1996. Aeolian geomorphology: anintroduction. Addison Wes-
ley Longman Ltd, Boston,.
Lopez, M.V., Sabre, M., Gracia, R., Arrue, J.L., Gomes, L., 1998. Tillage effects on soil surface
conditions and dust emission by wind erosion in semiarid Aragon (NE Spain). Soil
Tillage Res. 45, 91–105.
López, M.V., de Dios Herrero, J.M., Hevia, G.G., Gracia, R., Buschiazzo, D.E., 2007. Determi-
nation ofthe wind-erodiblefraction of soils using differentmethodologies.Geoderma
139, 407–411.
Lyles, L., 1975. Possible effects of wind erosion on soil productivity. J. Soil Water Conserv.
30, 279–283.
Moreno Brotons, J., Romero Díaz, A., Alonso Sarría, F., Belmonte Serrato, F., 2009. Wind
erosion on mining waste in southeast Spain. Land Degrad. Dev. 21, 196–209.
Nordstrom, K.F., Hotta, S., 2004.Wind erosion from cropland in the USA: a review of prob-
lems, solutions and prospects. Geoderma 121, 157–167.
Panagos, P., Van Liedekerke, M., Jones, A., Montanarella, L., 2012. European Soil Data
Centre: response to European policy support and public data requirements. Land
Use Policy 29, 329–338.
Panagos, P., Meusburger, K., Ballabio, C., Borrelli, P., Alewell, C., 2014. Soil erodibility in
Europe: a high-resolution dataset based on LUCAS. Sci. Total Environ. 479–480,
189–200.
Quinlan, J.R., 1992. Learning with continuous classes. Proceedings of the 5th Australian
Joint Conference on Artificial Intelligence, 92, pp. 343–348.
Riksen, M.J.P.M., De Graaff, J., 2001. On‐site and off‐site effects of Wind Erosion on
European Light Soils. Land Degrad. Dev. 12, 1–11.
Shao, Y., 2008. Physics and Modelling of Wind Erosion. Springer, Cologne.
Shiyatyi, E.I., 1965. Wind structure and velocity over a rugged soil surface. Vestnik
Sel.-khoz. Nauki 10.
Stefanovits, P., Várallyay, G., 1992. State and management of soil erosion in Hungary.Pro-
ceedings of the Soil Erosion and Remediation Workshop, US—Central and Eastern
European Agro-Environmental Program, Budapest, vol. I, pp. 79–95.
Tóth, G., Jones, A., Montanarella, L., 2013. LUCAS topsoil survey methodology, data and re-
sults. Report EUR 26102 EN.
Warren, A., 2003 . Wind Erosion onAgricultural Land inEurope: Research Results for Land
Managers. Office for Official Publications of the European Communities, Luxembourg.
Wolfe, S.A., Nickling, W.G., 1993. The protective role of sparse vegetation in wind erosion.
Prog. Phys. Geogr. 17, 50–68.
Woodruff, N.P., Siddoway, F.H., 1965. A wind erosion equation. Soil Sci. Soc. Am. Proc. 29,
602–608.
Zobeck, T.M., 1991. Soil properties affecting wind erosion. J. Soil Water Conserv. 46,
112–118.
478 P. Borrelli et al. / Geoderma 232–234 (2014) 471–478