Assessment of remotely sensed precipitation products for climatic and hydrological studies in arid to semi-arid data-scarce region, central-western Morocco

Abstract

The lack of ground-based data remains the main challenge for hydrologists working in semi-arid to arid regions. In this context, remotely sensed precipitation products (SPPs) are crucial for widespread hydrological and climatic applications. The core benefit of SPPs is their high spatial-temporal resolution with evenly distributed precipitation estimates. However, their accuracy can vary considerably with climatic conditions and location and should be concisely evaluated beforehand. Therefore, this study is a first attempt to exhaustively compares and evaluates the performance of three widely used SPPs (CHIRPS, TRMM and PERSIANN) against ground-based observations at daily scale over the Souss data-sparse region (Central-western Morocco). Our proposed approach consisted of three main steps. The reference gauge method was adopted to evaluate daily precipitation data. Next, the hydrological simulation method was used to assess the performance of SPPs in simulating the observed streamflow in four locations of the SRB using multiple models. The non-parametric methods were applied to assess the capacity of SPPs in capturing precipitation extremes and characterizing the hydrological signature of SRB based on streamflow extreme detection. Our findings indicated that the three SPPs were not able to well reproduce the total amount of daily ground-based observations over the Souss region (CCmedian < 0.51; PBIAS > ±35%). The HEC-HMS model coupled with the TRMM product is the most appropriate combination for daily river flow restitution and for hydrologic signatures detection in SRB. For extreme precipitation events, CHIRPS showed satisfactory performance in capturing nearly all extreme events indices while PERSIANN and TRMM demonstrated reasonable ability to capture the wet extreme events. Therefore, it is still rather difficult to replace completely the ground-based data. Consequently, we recommend to further improve the satellite-based algorithms to adapt to the complex topography and special characteristics of semi-arid data-scarce regions.