Seasonal monitoring of soil erosion at regional scale: An application of the G2 model in Crete focusing on agricultural land uses

• G2 model highlights spatio-temporal variability of soil erosion.
• October to January is the most critical season in Crete for soil erosion.
• Natural grasslands and shrublands are the most vulnerable land uses.
• The mean annual erosion is estimated to 3.065 t ha−1 for agricultural land use.
• The mean annual erosion is estimated to 14.023 t ha−1 for potentially grazed lands.

A new soil erosion model, namely G2, was applied in the island of Crete with a focus on agricultural land uses, including potential grazing lands. The G2 model was developed within the Geoland2 project as an agro-environmental service in the framework of the Global Monitoring for Environment and Security (GMES, now Copernicus) initiative. The G2 model takes advantage of the empirical background of the Universal Soil Loss Equation (USLE) and the Gavrilovic model, together with readily available time series of vegetation layers and 10-min rainfall intensity data to produce monthly time-step erosion risk maps at 300 m cell size. The innovations of the G2 model include the implementation of land-use influence parameters based on empirical data and the introduction of a corrective term in the estimation of the topographic influence factor. The mean annual erosion rate in Crete was found to be 8.123 t ha−1. The season from October to January (the rainy season in Crete) was found to be the most critical, accounting for 80% of the annual erosion in the island. Seasonal erosion figures proved to be crucial for the identification of erosion hotspots and of risky land uses. In Crete, high annual erosion figures were detected in natural grasslands and shrublands (14.023 t ha−1), mainly due to the intensification of livestock grazing during the past decades. The G2 model allows for the integrated spatio-temporal monitoring of soil erosion per land-use type based on moderate data input requirements and existing datasets.