Sarah Cooley (Duke University, United States); Elizabeth Webb (Duke University, United States); Camryn Kluetmeier (Duke University, United States)

We developed a novel multi-sensor approach that integrates observations from SWOT, Sentinel-2, and MODIS/VIIRS satellite sensors into harmonized time series for individual lakes and river reaches. Our primary research objective is to evaluate whether incorporating Sentinel-2’s frequent 10 m resolution water surface area measurements and MODIS/VIIRS’ daily cloud observations with SWOT altimetry can reduce uncertainty and increase the temporal resolution of water storage measurements. To do this, we first assembled a benchmark dataset of 315 gaged lakes and reservoirs across the US. For each of these water bodies, we produced a harmonized satellite-based time series using the SWOT single pass lake product, Sentinel-2 derived water surface area, and MODIS/VIIRS observed cloud properties. We utilized the 134 water bodies in the benchmark dataset that only have in situ water level data to produce an observation filtering model designed to filter poor SWOT-based water surface elevation measurements that were not captured by the lake product’s quality flags. Using the remaining gaged water bodies, we compared the performance of water storage time series from the filtered and unfiltered SWOT single pass lake product to our storage calculations that leverage Sentinel-2 water surface area. Our results suggest that combining satellite sensors leads to (a) improved SWOT water surface elevation observation filtering; (b) reduced uncertainty in water surface elevation – area relationships; and (c) improved temporal resolution and the ability to reconstruct historical surface water storage trends. We also demonstrate this multi-sensor approach in river reaches, which has the potential to support SWOT-based discharge estimation.