The Soil Erosion Use Case aims to assess soil susceptibility to water erosion by integrating precipitation datasets, which drive estimates of rainfall erosivity, with non-climate datasets like Earth Observation (EO) imagery, and use integrated data to inform the widely used Revised Universal Soil Loss Equation (RUSLE) model. 

To ensure maximum consistency and homogeneity, the Use Case primarily relies on data from Copernicus services and space components. Rainfall time series coming from climate re-analysis and interpolation of observations will be extracted from the Copernicus Climate Change Service. Similarly, land-based parameters, such as elevation or the intra-annual cycle of land fractional cover, will be retrieved from the Copernicus Land Monitoring Service or by extracting vegetation indices by satellite imagery (e.g. Sentinel-2). 

Building on an existing pilot within the Copernicus Climate Change Service, CMCC plans to develop an updated, high-resolution and cutting-edge service for evaluating soil erosion in Italy, untapping the potential of the most updated EO data, and the tools provided by the EO4EU platform. For instance, new climate re-analysis data at high spatial resolution will be utilized as the main rainfall input, and AI algorithms will allow users to estimate sub-annual dynamics of land cover and their impact on erosion. This Use Case focuses on five administrative regions in southern Italy, an area among the most affected by rainfall-induced erosion hazards in Europe, which also poses transport infrastructures at risk, and involves stakeholders and potential users from the land management and infrastructure sectors. 

Challenge

Rainfall-induced soil erosion occurs when soil particles are detached, transported and deposited away due to rainfall, runoff, snow melting or irrigation. When soil erosion rate is higher than soil formation rate, the soil becomes depleted and the potential of the land to be used productively is reduced. The direct and indirect economic costs of soil erosion are extremely high, as a variety of sectors are negatively impacted. Soil erosion may lead to a decrease in the yield of agricultural areas, physical damage to cultivated fields and deterioration of water and air quality due to suspended soil particles. As a consequence, reducing soil erosion can provide benefits for several Ecosystem Services like habitats, air quality, water quality and the protection, formation and decontamination of soils and sediments, and secondarily for water quantity. For these reasons, soil erosion has raised great attention at EU level, and reducing soil erosion has synergies with Sustainable Development Goals (SDGs), in particular with the goals regarding “Food security” and “Life on land”, and, in cascade, with the SDGs generically related to poverty, water, and climate actions, and finally also for health and wellbeing.

Solution

Through the developed service, a plethora of end-users will have access to information related to water-induced soil erosion for their region of interest. The application will increase awareness of soil erosion patterns and severity. Also, it will assist decision-makers, like land management actors and territorial planners, informing them about the impact of farming practices, forest management or post-disturbance recovery of soil (i.e., after floods and fires) in mitigating the hazards associated with soil erosion and, consequently, the related physical and economic risks. The Use Case can also support climate-informed design and management of infrastructures, ensuring a more robust evaluation of hazards and risks posed by rainfall-induced soil erosion. Soil erosion estimates for future horizons, also provided by the Use Case building on the most updated and fine resolution climate projections, will allow for the design of more resilient infrastructures, as well as an appropriate and climate-informed allocation of investments for maintenance, update and retrofitting. 

Current status

A preliminary, embryonic-stage service for the evaluation of rainfall-induced soil erosion in Italy is already available as Demo Case (dataset and applications) on the Climate Data Store of the Copernicus Climate Change Service. However, it relies on state-of-the-art and less region-specific empirical methods and data, especially for the quantification of rainfall erosivity, and on low-resolution datasets for the evaluation of soil-related features. The proposed Use Case will provide significant enhancements in the robustness of soil erosion estimates, using novel, cutting-edge approaches, tools and data for all the key parameters of the erosion process. 

According to the adopted Revised Universal Soil Loss Equation (RUSLE), the potential soil loss due to erosion depends on rainfall erosivity and soil susceptibility. Regarding the former, a sub-annual empirical model recently calibrated across Italy will be soon implemented on the EO4EU platform. The estimation of land susceptibility to erosion will be based on three different algorithms. The first innovatively starts from consolidated models analyzing the morphological features of the terrain, such as the slope, and adds to them a new module accounting for the presence of transport infrastructures. The second processes specific datasets on soil properties, like the texture and the content of organic matter, to inform a combination of empirical models from the literature and estimate the overall soil erodibility. The third algorithm is a cutting-edge solution considering the influence of land cover and plant phenological dynamics in inhibiting or reducing erosion. It is an artificial neural network trained on Earth Observation multispectral images and a state-of-the-art lookup table to assign spatially distributed erosion correction factors depending on dynamical cover properties. A prototype of this algorithm is already implemented in the EO4EU platform, while a more robust version will be deployed soon.    

Sentinel-2 multispectral images underlie the determination of land cover and management impact on soil erosion. All available 10 and 20 m bands are used to train an artificial neural network algorithm accounting for diverse land cover spectral response in different cover type and phase of the year.