Monthly Time-Step Erosion Risk Monitoring of Ishmi-Erzeni Watershed, Albania, Using the G2 Model

Abstract

In this study, soil erosion was mapped in Ishmi-Erzeni watershed, Albania, using the G2 model. The G2 model has been proposed as an agri-environmental service by the Global Monitoring for Environment and Security (GMES) initiative (now Copernicus programme). Based on the principles of the Universal Soil Loss Equation (USLE), G2 provides maps of actual soil loss at a monthly time-step. The main innovations of the model with regard to previous USLE family models are as follows: the introduction of a ‘storm factor’, which differentiates rainfall erosivity (R factor) per month when detailed rain intensity records are not available; the use of standardised biophysical parameters derived from satellite image time series in combination with land use information for calculating the vegetation retention factor (denoted here as V factor, corresponding to C factor of USLE); and the use of satellite imagery for calculating a new factor, namely the slope intercept factor (denoted as I factor), which expresses landscape feature alterations, thus functioning as corrective to the topographic influence factor (denoted here as T factor, the slope length and steepness (LS) factor of USLE). The model was originally implemented in the cross-border Strymonas river basin and on the island of Crete after revision; in both cases with encouraging results. The G2 model follows a data-driven methodology, while providing alternatives for all factor estimations with moderate data requirements. For the model application in the Ishmi-Erzeni watershed (covering 2200 km2), rainfall data were collected from ten weather stations, and soil properties were measured from sampling at 47 locations. Vegetation layers were downloaded from the GMES portal, while land use information was extracted from a Landsat-7 TM image. Finally, terrain properties were calculated from a 250-m digital elevation model (DEM) of the area. The G2 model showed Ishmi-Erzeni to have moderate soil erosion, with a mean annual soil loss estimated to be 6.5 t/ha; however, 18 % of the area is facing an annual risk of soil removal more than 10 t/ha, which is considered to be a sustainable threshold for Albania. Winter months appear to be the most risky, with all months contributing substantially to the annual erosion rate (i.e. between 4 and 12 % each). Areas of coniferous and mixed forests, together with mountainous agricultural land, appear to be the most risky land uses. In conclusion, the G2 model proved to be useful and efficient for predicting erosion at monthly time-steps for all land uses in the Ishmi-Erzeni watershed. Future research will focus on an Albania-wide erosion mapping task, using the G2 model.