Potential of Gridded Precipitation Estimate Products for Hydrological Modelling: The Case of Small Watersheds in Côte d'Ivoire

Rainfall information measured on the ground is not sufficiently representative the watershed scale in Côte d'Ivoire due to a low density of the rain gauge network. Many watersheds are ungauged or present a high rate of missing data on flow record since 1990s. The use of gridded precipitation estimate products is an alternative to remedy this lack of information. Therefore, the aim of this study was to evaluate the performance of CHIRPS, GPCC, MSWEP, PERSIANN-CDR, PERSIANN-CCS-CDR, TAMSAT, WFDE5-CRU, and WFDE5-CRU-GPCC precipitation estimates for hydrological modelling in Côte d'Ivoire over the period from 1984 to 2004. A continuous-based flow simulation was made for 14 small watersheds using GR4J lumped model. The methodology was based on a global analysis to identify the products that gave the best scores in both calibration and validation over the selected watersheds. KGE2 was the major metric considered along with NSE, VE, PBIAS, and ME. The performance of these gridded precipitation estimate products was then evaluated according to the different climatic zones of Côte d'Ivoire. The results showed that WFDE5-CRU-GPCC performed well in all climatic zones with KGE2 > 0.6 for at least the half watersheds for each climatic zone. CHIRPS and PERSIANN-CDR exhibited the best performance in the transitional tropical and mountainous zones (KGE2mean > 0,7), whereas WFDE5-CRU was good, mainly in the mountainous zone with KGE2 > 0,7 for 2 watersheds out of 3.