INTEGRATING OPTICAL AND RADAR IMAGERY TO ENHANCE RIVER DROUGHT MONITORING

Abstract

Drought events are growingly affecting European and Italian territories, hampering local environments and biodiversity, such as the ones relying on rivers for their subsistence. Monitoring of rivers is becoming an important issue to face drought crisis and may be exploited with different tools. Among the most commons, satellite imagery is exploited to map water coverage, basing on optical or radar sources. This work proposes a combination of the two sensors to overcome possible limitations of the single dataset exploitation, reaching a reliable result. The methodology is applied to a stretch of Po River in Lombardy region (Italy). Through Google Earth Engine platform, optical satellite Sentinel-2 and radar satellite Sentinel-1 data are processed. The combination of the radar data and of the optical spectral indices is carried out through a pixel-based supervised classification, with a Random Forest classifier. Maps of water coverage are obtained, numerical outcomes of water surface evaluation are recorded and validated by the mean of reference hydrometric data. A multitemporal analysis is then reported, aiming to prove the efficiency of the procedure. All iterations show reliable accuracies and correlation among water surface estimation and water table measurements in two sections of interest. In perspective, the proposed methodology will be implemented in tools for supporting drought monitoring to be integrated in environmental public administration policies.