Evaluation of RFE Satellite Precipitation and its Use in Streamflow Simulation in Poorly Gauged Basins

Abstract

Abstract: The performance of satellite Rainfall Estimates (RFE, version 2.0) at daily resolution was evaluated in comparison with ground-based meteorological datasets (GBD) by applying statistical and hydrological modeling approaches. In-situ daily rainfall observations from 5 stations in and around the periphery of the Nasia river basin in Ghana, covering a period of 15 years (2001–2015), were used in this research. Comparison of the observed rainfall data with satellite-based estimates revealed a strong positive correlation, which yielded the highest correlation coefficient (R2) of 0.74 at the monthly timescale as against the weak positive linear relationship with the highest R2 of 0.41 at the daily timescale. Mean annual precipitation computed from both datasets also showed close correspondence yielding 978.83 mm/annum and 977.12 mm/annum for RFE and GBD, respectively. Calibration at the daily timescale showed that the ground-based data (GBD) performed better in simulating the observed streamflows compared to the satellite-based (RFE) simulations yielding a Nash Sutcliffe Efficiency (NSE) of 0.81 and 0.67 for the GBD and RFE, respectively. At the monthly timescale, the performance of both datasets improved, resulting in an NSE of 0.89 and 0.80 for the GBD and RFE, respectively. Although the RFE-based simulations could not perfectly reproduce the observed discharge, it can be used to supplement traditional in-situ gauge data to address the problem of non-availability of observed rainfall data for hydrological applications such as water resources planning and assessment. Future research into the usability of the RFE in other medium scale river basins could be carried out to compare with these results. Highlights: • There is a strong correlation between RFE version 2.0 and ground-based monthly rainfall data • RFE-based simulations appear to underestimate the observed hydrographs compared to ground-based data simulations • RFE could be used to supplement in-situ observations of rainfall for hydrological analysis