Evaluating satellite rainfall products against ground-based observations over Zambia

Abstract

Evaluating Satellite Rainfall Products (SRPs) is essential for reliable precipitation estimates, especially in the face of climate change and inadequate networks of ground-based observations. This study evaluated the accuracy of SRPs as a proxy for precipitation data recorded by ground-based observations. Four SRPs—ARC2, CHIRPS, RFEv2, and TAMSAT— were evaluated against ground-based observations using false alarm ratio, statistical metrics accuracy, probability of detection and percentage. Results demonstrated significant variations in the performance of SRPs at seasonal, monthly, and daily time steps. RFEv2 exhibited superior rainfall detection capabilities compared to CHIRPS, TAMSAT, and ARC2. The daily rainfall estimates from RFEv2 were more accurate than those of TAMSAT, ARC2, and CHIRPS. Among the four SRPs, the datasets with comparable long-term data periods are ARC2, CHIRPS, and TAMSAT. ARC2 performed better at daily and monthly (January, February), while CHIRPS and ARC2 showed the highest efficacy at the seasonal (October, November, December) scale. In the December-January-February seasonal scale, TAMSAT and ARC2 surpassed CHIRPS, and during the January-February-March period, CHIRPS outperformed the rest. The daily Pearson correlation values ranged from 0.29 to 0.74, depicting a moderate to strong correlation between the SRPs and ground-based observations. This study provides novel perspectives for selecting SRPs that could be applied across water, agriculture, health, and energy sectors because of their high spatio-temporal resolution over every place globally. SRPs are crucial for planning and disaster risk reduction strategies, especially where ground-based observations are scarce.