Comparability of radar and optical methods in identifying surface water in a semi-arid protected area

Abstract

Surface water assumes a pivotal role in sustaining a wide range of wildlife species in semi-arid protected areas. However, differences in surface water body typology, underlying soil type, wildlife activity, the presence of phytoplankton amongst other factors, result in high variability of surface water spectral reflectance and detection accuracy. In this study, the performance of radar and optical methods was evaluated in detecting surface water of variable spectral reflectance in Hwange National Park, Zimbabwe using Sentinel-1 radar and Sentinel-2 optical images for the period 2016–2023. Results demonstrated that radar methods had low surface water detection accuracy which was highly variable as shown by overall accuracy and kappa statistic measures which continuously changed over time compared with optical methods. The overall best-performing method was the optical AWEI_nsh (sharpened) which showed high surface water detection accuracy and consistency (OA: 94%–100%) and (κ: 0.88–1.00) from 2016 to 2023. Therefore, optical methods present a stable and robust way for surface water monitoring in heterogeneous semi-arid protected areas. However, radar-based methods should be continually explored where optical-based technologies are impeded as a result of vegetation cover and cloud conditions.