The performance of satellite-based precipitation data to reproduce observed precipitation in Tanzania's Little Ruaha Catchment
Keywords:
Satellite rainfall, CHIRPS, Tanzania, Remote Sensing, Precipitation validationAbstract
Satellite-based precipitation datasets have emerged as promising tools for addressing rainfall data scarcity in regions with limited ground observations, such as Sub-Saharan Africa. This study evaluates the performance of five satellite-based precipitation products CHIRPS, CPC, GPCC, MERRA-2, and ERA5 in capturing rainfall characteristics over the Little Ruaha Catchment in Tanzania. The assessment was carried out at daily, monthly, and seasonal timescales and focused on key rainfall indices including onset and cessation dates, length of the rainy season, total rainfall, and frequency of rainy days. Validation was conducted using statistical indicators such as the correlation coefficient (r), Nash–Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), and Percent Bias (PBIAS). The results indicate that CHIRPS and CPC had strong agreement with observed rainfall atmonthly and seasonal scales. CHIRPS, in particular, demonstrated consistent performance with r > 0.85 and NSE > 0.7 across most stations. At the daily scale, although all datasets performed poorly, CHIRPS still outperformed the others, showing moderate correlations around r ≈ 0.65. MERRA-2 and ERA5 showed higher errors and inconsistencies across different locations. The rainfall indices revealed that satellite datasets generally estimated earlier rainfall onset and delayed cessation compared to gauge data, resulting in longer estimated rainy seasons. For instance, CHIRPS overestimated the rainy season length by 31 days at Mafinga Bomani. In addition, satellite data often reported a higher frequency of rainy days, with discrepancies up to 16 days in some locations. These variations are attributed to the satellites' ability to detect light and scattered rainfall events missed by ground gauges. Overall, CHIRPS emerged as the most reliable product, offering a credible alternative for climate monitoring, hydrological modelling, and agricultural planning in data-scarce regions