Performance Evaluation of Near-Real-Time Satellite Rainfall Estimates over Three Distinct Climatic Zones in Tropical West-Africa

The performance of four Near-Real-Time Satellite-Based Rainfall Estimates (NRT_SREs) was evaluated across the Volta basin from January 2019 to December 2020: Global Satellite Mapping of Precipitation (GSMaP_NRT), Integrated Multi-satellitE Retrievals for Global Precipitation Measurement-Early run (IMERG-E), Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN)-Cloud Classification System (PERSIANN-CCS), and PERSIANN-Dynamic Infrared Rain Rate (PDIR_NOW). They were also compared to their post-real-time counterparts: PERSIANN, IMERG-Final run (IMERG-F), IMERG-Late run (IMERG-L) and GSMaP_MVK. Quantitative and categorical metrics were used in conducting hourly and daily evaluations at individual stations across the basin, as well as at zonal and seasonal scales. The results revealed that all the NRT_SREs had weak correlations (third quartile: 0.3625) at the hourly timescale. IMERG-F had the best correlation (R) of all the SREs, but it also had the worst Root Mean Square Error (RMSE) and False Alarm Ratio (FAR), being outperformed by IMERG-E and IMERG-L. IMERG-E also outperformed the NRT_SREs in most cases. However, in the arid Sudano-Sahelian zone, PDIR_NOW had the highest probability of detecting rainfall of all SREs (at the daily timescale) and all NRT_SREs (at both timescales). This was most likely because of PDIR_NOW's increased maximum temperature threshold. Seasonal analysis revealed that the RMSE of the NRT_SREs was significantly lower during the dry season than during the wet season, and vice versa for FAR. The findings of this study are expected to provide not only valuable feedback to algorithm developers in order to improve NRT_SREs, but also guidance to data users worldwide. Highlights center dot All NRT_SREs performed poorly at hourly timescale, but improved at daily timescale IMERG-E outperformed all NRT_SREs in most cases, irrespective of the season and zone IMERG-E had better RMSE and pBIAS than PRT IMERG-F at hourly and daily timescale IMERG-E could supplement rainfall measurements within the basin at daily timescale