Estimating Wind Direction and Wind Speed Over Lakes With Surface Water Ocean Topography and Sentinel-1 Satellite Observations

Abstract

Wind at the water-air interface is an important driver of hydrologic and biogeochemical processes in lakes. Satellite synthetic aperture radar (SAR) is commonly used over the ocean to retrieve wind fields using backscatter coefficients which are sensitive to wind-driven surface water roughness; however, its application to lakes has been largely unexplored. Here we assess the utility of SAR to retrieve wind fields specifically for lakes. We estimated wind direction from SAR backscatter using the Modified Local Gradient method for Surface Water Ocean Topography (SWOT) and Sentinel-1 data. The estimated wind direction was then used as an input into a C-band geophysical modeling function (GMF) to invert wind speed from Sentinel-1 data. Comparisons between SWOT backscatter and in situ wind speeds were used to provide a foundation for understanding how SWOT could be used to study wind speeds. Using buoy data for validation, we found wind direction (1 km) mean absolute error (MAE) ranged from 31° to 40° for Sentinel-1 and 28° to 38° for SWOT. Sentinel-1 wind speed (100 m) MAE ranged from 1.05 to 2.09 m/s. These retrievals were more accurate and at higher resolution compared to global reanalysis dataset ERA5 (0.25°), with wind direction MAE from 23° to 50° and wind speed MAE from 1.49 to 2.35 m/s. SWOT backscatter sensitivity to wind speed depended on incidence angle, and demonstrated utility for developing a GMF for lakes. These methods could be used to better understand wind dynamics globally, especially over small lakes and in data poor regions.