In situ, modeled, and earth observation monitoring of surface water availability in West African rangelands

Introduction Rangeland ponds are vital to the livelihoods of pastoral and agropastoral communities in Africa, providing an important source of water for livestock. However, sparse instrumentation across much of Africa makes it extremely challenging to monitor surface water availability in these areas. Model estimates of surface water, for example, as used by the Famine Early Warning Systems Network (FEWS NET) Water Point Viewer, are one of the few operational tools available to monitor surface water stress across pastoral areas of the Sahel and East Africa. Methods Water availability data from these models are difficult to validate. New methods using satellite data to classify surface water provide an opportunity to assess the performance of these tools. This study compares water availability estimates derived from Landsat and Sentinel 1 satellite imagery to in situ observations and model simulations of water availability in 22 ephemeral ponds located in the Ferlo region of Senegal. Results and discussion The Active-Passive Water Classification (APWC) algorithm detected surface water at each location. Over 2022 and 2023, water was detected in pond locations annually at a frequency of 68.2% for all ponds and at a frequency of 43.8% for ponds with a surface area <10,000 square meters (m 2 ). The APWC results outperform global and continental surface water datasets in the Ferlo region. Seasonal water availability was captured in 12 ponds over the 2022 and 2023 seasons. The 12 locations can function as sentinel ponds to monitor local water availability. Study results demonstrate the viability of satellite methods to assess water availability in the region, as well as the challenges to using satellite-based methods to estimate water availability in small ponds.