The ongoing land degradation processes visible across Europe contribute to raising the question whether policies are effective enough to face current trends [1
] and their possible exacerbation due to climate change [2
]. The soil governance in the EU has gained increasing importance in recent years, and it is indirectly addressed in different policy areas (agriculture, water, and climate) [3
]. In 2006, the European Commission developed a common EU strategy (COM(2006)23) for soil protection including the proposal of a EU Soil Framework Directive. The proposed directive, which would have guided the EU Member States to take actions to prevent soil degradation [4
], was withdrawn in 2014 following the lack of agreement between the EU Member States. Today, the Soil Thematic Strategy is the only EU legislation in force that merely concerns soil. With regard to soil erosion, it (a) recognizes soil degradation due to erosion as a major threat for food security and the environment; (b) provides a set of overarching principles and calls for stronger integrated policy and; (c) promotes ‘raising awareness’ and the need for additional research related to soil degradation and protection [5
Before trying to put in place the Soil Thematic Strategy, the EU promoted a more environmentally friendly agriculture by introducing the so-called cross-compliance mechanism in the Common Agricultural Policy (CAP) reform of 2003 [7
]. With the introduction of the cross-compliance mechanism, farmer support payments were conditioned on the adoption of environmental, animal welfare and food safety standards. This led to the definition of Good Agricultural and Environmental Conditions (GAEC), first established by the Council Regulation No.1782/2003 and subsequently Council Regulation (EC) No 73/2009. Since 2009, EU Member States have the compulsory requirement to keep their land in good agricultural and environmental conditions under the CAP [8
]. The introduction of the GAEC was a further legislative effort to promote prevention and mitigation of soil erosion and maintenance of soil organic matter.
In 2013 (Figure 1
), the CAP 2014–2020 reform proposed the sustainable management of natural resources as one of its main objectives [9
]. Under the cross-compliance mechanism, Member States were required to define minimum standards for soil protection, such as the minimum soil cover (compulsory) and other management practices such as terraces and grass buffer strips (optional) to limit soil erosion [10
]. Under Pillar 1 of the 2014–2020 CAP, the greening practices required the application of crop diversification, maintenance of permanent grasslands (less susceptible to erosion compared to croplands) and ecological focus areas favoring soil conservation. Under Pillar 2 of the 2014–2020 CAP, the prevention of soil erosion (through practices of conservation agriculture, and green covers) was one of the priorities of rural development. The 2014–2020 CAP also put in place a monitoring and evaluation framework to assess the environmental impact of farming through indicators of soil protection, namely: (a) soil erosion by water and; (b) soil organic matter.
At the global scale, the United Nations agreed, in 2015, on the adoption of the Sustainable Development Goals (SDGs) as a major policy driver towards sustainable development and well-being [11
]. The EU has been a front-runner in monitoring the SDGs by developing a set of indicators at the EU level [12
], taking into account recent policy developments and improved data availability. This comprises 100 indicators relevant for monitoring progress towards the SDGs, including soil erosion by water which refers to both SDG2 (zero hunger) and SDG15 (life on land) [13
The post 2020 CAP reform (2021–2027), currently under discussion in the European Parliament, includes nine objectives and three of them aim to implement environmental and climate change actions: (a) Contribute to climate change mitigation/adaptation; (b) foster sustainable development and efficient management of water soil and air, and; (c) contribute to biodiversity protection and ecosystem services (European Commission, 2018) [14
]. The proposed 2021–2027 CAP regulation sets soil erosion among the 28 indicators to monitor the impact and performance of this policy [15
]. Among others, the Member States will establish management plans to reduce the percentage of agricultural land under severe erosion (> 11 t ha−1
). In the post 2020 CAP, the proposed GAEC include, among others, a ban on burning arable stubble, protection of bare soils in winter with cover crops, and crop rotation and tillage management in areas of soil degradation (sloping arable lands).
Briefly, the current policy framework (Figure 1
) for soil protection is fragmented and cannot entirely implement measures against soil threats and towards improving soil functions [16
]. However, monitoring indicators can measure and evaluate the soil governance instruments [18
]. The development of effective indicators, related to land processes occurring at continental scale, requires a state-of-the-art methodology, able to detect changes under environmental pressures and policy drivers [13
In the past, different modelling approaches have been developed at the continental scale, such as the PESERA [19
] and the erosion rates based on runoff plot data [20
]. Even though these studies are advancing the knowledge on soil erosion processes and modelling, they do not offer the possibility to assess the impact of conservation practices on mitigating soil erosion and the role of agri-environmental policies. For these reasons, the application of the RUSLE2015 model [21
], previously developed and tailored to the European context, was applied to estimate the erosion rates in 2016.
The main objective of this paper is to present the soil erosion indicator development as a response to the EU agri-environmental policy requests. Here, we present an update of the assessment of soil loss by water erosion for the year 2016 and the changes from 2010.
4.1. Progress in the Soil Erosion Indicators
The progress of the soil erosion indicators is more obvious at regional level. Even if the spatial patterns of the soil erosion datasets (2010, 2016) are similar, the differences in mean soil erosion rates are evident when we aggregate the differences at regional level (Figure 5
Among the 281 regions studied, the soil erosion indicator (mean soil loss rates as t ha−1
) shows a mean decrease of -0.8%. For the majority of the regions (73 out of 281), the mean soil loss rates for the period 2010–2016 show an insignificant change of -1 to 1% (yellow color in Figure 5
). In practice, there is no significant change of applied soil conservation practices in those regions, mainly located in Central Europe. For 42 regions the decrease of soil loss rates was between 1–3%, while 37 regions have a significant decrease in the range 3–5%. Finally, 38 regions have increased management practices to reduce soil loss rates by more than 5% in the period 2010–2016. This decrease is mainly observed in Western Europe (Denmark, France, and Germany) as well as in Portugal.
In contrast, a slight increase of 1–3% of soil loss rates was modelled in 56 regions; these are located in Spain, Poland and part of Italy (Figure 5
). Finally, an increase of soil loss rates by more than 3% is noted in 25 regions, mainly located in South Europe (Greece, Italy and Bulgaria). This mostly has to do with the decrease of conservation tillage in those regions.
4.2. Model Comparison with Modelled Regional Assessments
Since 2015, the soil loss estimates by water erosion and the input layers are available in the European Soil Data Centre (ESDAC). This allows public users and researchers to download the data, replicate the methodology and compare their results with the European dataset which is often used as a baseline.
In order to evaluate the performance of the RUSLE2015 at continental scale (>4.5 million Km2
), we decided to do an inter-comparison with independent RUSLE-based assessments at national, regional and local scales. In general, local/regional modelling studies use more accurate (and detailed) input datasets (higher spatial resolution, geo-referenced crops, etc.), and we expect that local/regional experts know their study area better compared to the authors. We have collected nine national datasets (Belgium, Bulgaria, Germany, Denmark, Netherlands, Poland, and Slovakia, 10 regions in Italy and agricultural soils in Austria) from the European Environment Information and Observation Network (EIONET) [48
]. In addition, we collected 20 studies which compare their modelling outputs with our RUSLE2015. In total, the 29 local/regional studies are presented in the Appendix A
; Figure A1
). The RUSLE2015 outputs fell in the estimates range of local and regional modelled studies (Figure 6
4.3. Uncertainties and Model Development
The RUSLE model does not mechanistically represent the temporal climate variability and, more specific, the changes in rainfall intensity, which is calculated as a mean of measured data (R factor) for multiple years. Indeed, a more dynamic (annual) assessment of rain erosivity would be necessary. The northern parts of Europe are experiencing frequent rain events, while the Mediterranean part rarer but more extreme intense rainfall due to climate change (mean increase of rain erosivity by 18% in the EU from 2010 to 2050) [49
]. For a more dynamic approach, phenological changes and the inter-annual vegetation protection are necessary to capture the seasonal and inter-annual erosion variability.
Another important uncertainty is the lack of georeferenced data for crops and management practices. The ideal model for estimating soil loss from arable lands should also incorporate the annual crops and the applied soil conservation practices in the field. However, those data are not yet available at the continental scale (ca. 170 million ha); therefore we used statistical data on crop types provided by countries at regional level. Finally, the extremely erosion-sensitive areas (e.g., sparsely vegetated areas according to CORINE) have a very high variable cover spanning from Mediterranean badlands to steep rock slopes in highlands.
We also performed an uncertainty analysis due to possible misclassification of the land cover type as the C-factor shows the highest variability and is the most sensitive input factor [50
]. The analysis is based on the assumption that land cover misclassification is more likely to occur in highly fragmented pixels compared to more homogenous areas [51
]. The uncertainty can be estimated with the variability of neighboring pixels. After calculating the variance of a pixel neighborhood (Figure A2
, Appendix A
), it is possible to simulate the probability that a given pixel belongs to a different land cover and the possible impact on the model estimates. For the C-factor, the effect of uncertainty results in an estimate of ±488 million tons (MT) (at 0.95 confidence interval) at EU level in 2016. However, two European validation studies have shown that the achieved accuracy both in CORINE Land Cover (CLC) and the CORINE land cover changes layers are above the 85% [52
]. Taking into account the 15% uncertainty in CORINE layers, we estimated the uncertainty due to land cover misclassification at ±73.2MT (±7.6 %) of the 2016 estimates.
To overcome those uncertainties, a more dynamic modelling framework should be developed at continental scale. The dynamic soil erosion approach will incorporate improved estimates of vegetation changes at a high temporal resolution using remote sensing products in combination with detailed databases on crop types, soil characteristics and soil loss information collected at the field parcel and catchment scale in Europe. In addition, the availability of global atmospheric data re-analysis and development of numerical models to estimate rain erosivity regularly are promising for annual soil erosion estimations at continental scale.
4.4. Scenario Analysis Based on Policy Developments
RUSLE2015 model is adapted to run policies scenarios. In the impact assessment for the future Common Agricultural Policy (CAP 2021–2027), scenarios with different application rates of cover crops have been used (European Commission, 2018) [14
]. The flat rates scenarios showed that cover crops can reduce soil loss in arable lands by 3.9%–14%, depending on the application rate of cover crops ranging between 25%–75% of the total arable land. A more targeted scenario envisages the cover crops application in the 75% of arable lands with soil erosion exceeding 5 t ha−1
, 50% cover crops application where soil erosion exceeds 2 t ha−1
and 25% cover crops application where soil erosion exceeds 1 t ha−1
. This targeted scenario (variation of cover crops application based on erosion rates) will result in soil erosion rates reduction by 9.5% in the EU arable lands.
4.5. Remote Sensing to Model Soil Erosion by Water at Pan-European Scale
Anthropogenic influences through land use and agriculture substantially accelerate the magnitude of soil erosion processes [53
]. Recent advancements in remote sensing, the greater availability of earth observations data and the development of geospatial technologies for data processing have been remarkably improving the capacity of models to represent land cover changes and management practices and their potential effects in terms of soil erosion [54
]. Remote sensing is playing an increasingly crucial role in the development of a new generation of soil erosion models. However, the level of integration of remote sensing data and analytical techniques in soil erosion modelling is still heavily influenced by the spatial scale of the modelling application. Small- to local-scale soil erosion modelling applications are moving towards dynamic approaches, which use vegetation indices and phenology obtained from remote sensing based canopies to describe the intra-annual variability of crops and natural vegetation in field parcels [55
] or representative hillslopes [56
]. Recent large-scale modelling approaches like RUSLE2015 show substantial progress in the spatial description of land conditions using remote sensing [57
]. However, the increase in the study area generally leads to limitations related to the lower availability of input data and computation capacity. Today’s challenge in large-scale soil erosion modelling is to integrate more remote sensing to define the spatiotemporal occurrence of tillage, tillage intensity [58
] and other scarcely vegetated surfaces (e.g., wildfires and forest harvesting) [59
], which are more exposed to the effect of rainfall and overland flow [60
]. Meanwhile, RUSLE2015 uses pan-European remote sensing derived products such as the CORINE land cover (CLC) and the Copernicus fraction of vegetation cover (FCover) to provide first harmonized estimates at pan-European scale which can help decision-makers in both ex-ante and ex-post policy evaluation.
4.6. Data Availability
The data of the 2016 soil erosion indicator will be available at the European Soil Data Centre (ESDAC) both as raster files and as aggregated indicators at different regional levels (National; NUTS2: Regions; NUTS3: Provinces). In addition, the updated C-factor and P-factor data will also be provided. Besides the soil erosion data layer and the input factors, we made a number of derived pan-European datasets available such as the stoniness layer, the Rainfall Erosivity Database at European Scale (REDES), the vegetation density, the stone walls density and the grass margins distribution.
5. Conclusions and Outlook
The results from this study indicate a slight decrease, both in the mean continental soil loss rate (and, as a consequence in the total soil loss), and in the area affected by severe erosion. This decrease is due to the small increase of applied soil conservation practices in most countries and some small land cover changes (increase of urban areas, a minor decrease of shrublands and conversion of arable lands to pastures). The small mean increase in soil conservation practices at European scale (implying a decrease in the C- and P-factors) has been offset by a decrease of such conservation practices (and, as a consequence, increase of C- and P-factors) in more erosion-sensitive areas such as the Mediterranean basin. The small increase of grass margins (+8.2%) that was observed between the two LUCAS surveys (2012 and 2015) is expected to further increase due to the greening of the CAP 2014–2020. However, this assumption should be verified with the survey after 2020.
Soil formation rates found in the literature vary quite significantly as early studies report rates of 0.05–0.5 mm per year (≈ 1 t ha−1
] or 1.4–2 t ha−1
], while recent ones suggest that soil production can be 3.2–4.5 t ha−1
]. Around 25% (948 × 103
) of the potentially erosion-sensitive study area has soil loss rates above 2 t ha−1
). Comparing the soil loss rates, which only partially represent all soil erosion processes (gullying, wind erosion, harvest erosion and piping are not included in this study) [53
], we conclude that the total soil loss rates are much larger than soil formation rates in large parts of the EU and mitigation measures are therefore very urgent.
The marginal decrease of soil erosion in the EU in the period 2010–2016 compared to the 9% decrease between 2000–2010 needs further investigation to disentangle the drivers behind this trend. In addition to the almost double studied period (10 years vs 6 years) and the land-use change dynamics, we identified additional reasons. The first one is the gradual shift of the agricultural production model from a highly productive one (in the 1990s) with zero conservation practices, to an environmental friendly and sustainable agricultural model at the beginning of the century. Then, as we described in the introduction, there is a reinforcement of sustainability policies relevant to soil protection both at the national scale and in the EU. The third reason is the local adjustments driven by farming attitude, awareness raising, national/regional policies, and markets. Finally, we also consider that farmers’ incentives should be further linked to their environmental performance (conditionality) guided by a shift of the post-2020 CAP from compliance towards performance and results. An adequate EU legislation for soil protection is currently missing as the proposed Soil Framework Directive was not voted due to a blocking minority of 5 countries causing the consequent withdrawal in 2014. However, the future Common Agricultural Policy (CAP) 2021–2027 can be a framework for better monitoring soil erosion in the EU and for applying soil conservation practices to reduce soil erosion.
The strong bond between remote sensing and inventory statistics formed the basis for pan-European consistent characterizations of soil erosion with local importance and utility. The modelling approach uses harmonized input datasets and highlights the relative differences among regions and countries beyond national borders and local adapted models. RUSLE2015 includes scenario analysis and uncertainties, has performed in different time-steps (2000, 2010, 2016) and allows for hotspot identification. We do recognize that RUSLE2015, like any other soil erosion prediction approach, is not able to 100% reproduce reality, but we argue that the presented RUSLE-based pan-European approach provides a suitable basis for policy support as it offers process understanding, relative spatial and temporal variations, hotspots identification, scenario development and controlling factors to mitigate soil erosion impact. As such, this modelling approach has been used for both ex-post policy assessment (CAP 2014–2020) and ex-ante policy design (post 2020 CAP). In addition, the results of this pan-European assessment can also provide the basis for a well-planned approach to evaluate regional/national strategic plans against soil erosion and to direct new local monitoring/modelling efforts.
The proposed dataset and the derived indicators (in combination with local/regional studies) can be used in the context of the future CAP as baselines to identify areas vulnerable to soil erosion and to develop better targeted management plans to reduce soil loss rates. The current agri-environmental policies in place need to focus on hotspots and reduce soil erosion rates in agricultural lands where current rates exceed sustainable ones. In addition, more than 6% of agricultural areas suffer from severe erosion (i.e., soil loss rates >11 t ha−1 yr−1). An important step for environmental protection would be to set a sustainable soil loss target to reduce severe and extreme erosion on agricultural land by 2030. Taking into account the expected future increase of rainfall intensity (and thus rain erosivity) due to climate change, business as usual is not an option and additional policy actions are necessary to ensure environmental sustainability.