Bridging the Data Gap: Global Mapping of Small Water Reservoirs and Estimating their Storage Capacity

Small agricultural reservoirs play a critical role in meeting irrigation and livestock water needs during dry spells. However, systematic mapping of these reservoirs and estimating their storage capacity remain challenging due to their small size and high temporal variability caused by evaporative losses and frequent water withdrawals [1, 2]. This study integrates Sentinel-1 and Sentinel-2 satellite remote sensing data to develop a global detection methodology for agricultural reservoirs smaller than 0.1 km². Water pixels are identified using region-specific Otsu’s dynamic thresholding applied to Normalized Difference Water Index (NDWI) derived from Sentinel-2 and VV-polarized backscatter from Sentinel-1 data. A comprehensive ground-truth database was developed through manual reservoir delineation, enabling machine learning algorithm development and validation across diverse geographic and climatic regions. This research addresses a critical data gap in the extent and storage estimation of small reservoirs, which are often overlooked in existing datasets of surface water bodies. It supports key applications including irrigation management, water budgeting, drought resilience planning, and hydrological and climate modeling, with particular relevance to water-stressed regions where small reservoirs serve as essential agricultural water resources [3].