From 2D surface extent to 3D water levels: A SWOT-calibrated area–height framework for multidecadal reconstruction in hydraulically complex floodplain lakes

Abstract

Monitoring water levels in hydraulically complex floodplain lakes has long been constrained by a fundamental mismatch in satellite observations: optical sensors provide multidecadal records of surface water extent but lack vertical information, whereas conventional nadir altimetry retrieves precise elevations along sparse tracks that may not represent basin-wide water levels under strong surface gradients. This has limited the development of robust area–height (A–H) relationships and the reconstruction of historical water levels from archival imagery. The Surface Water and Ocean Topography (SWOT) mission helps address this limitation by providing spatially continuous two-dimensional water surface elevation fields from Ka-band radar interferometry. Here, we show how SWOT can serve as a geodetic anchor to calibrate a monotonicity-constrained cubic A–H model that converts long-term optical and synthetic aperture radar (SAR) water extent records into quasi-daily water level series. The framework is evaluated in Poyang Lake, China, a large river-connected floodplain lake with clear spatial gradients in water level. By fusing SWOT with Landsat, Sentinel, and MODIS archives, we reconstruct a quasi-daily water level time series for 2003–2025 that agrees closely with in situ gauges (RMSE = 0.44 m, MAE = 0.33 m, R² = 0.97). The reconstruction reveals a recent shift in hydrological regime, including an approximately 60-day advance in dry-season onset and an increase in extreme low-water events. This study presents a scalable virtual-station framework showing how short-term SWOT observations can extend long-term horizontal satellite archives into water level records for monitoring ungauged floodplain lakes worldwide.