A Physically Informed Transformation of Sentinel-1 SAR Observations for Streamflow Inference in Small and Complex Rivers

Abstract

Accurate measurement of river discharge is central to flood-risk management and water-resources planning, yet consistent monitoring remains difficult in data-scarce and information-restricted settings, particularly for small and narrow rivers. Here, we show that surface-water dynamics observed by Sentinel-1 synthetic aperture radar (SAR) can be transformed into a hydrologically meaningful surrogate discharge signal without ground-based ancillary observations. We develop a scalable SAR-based framework that extracts temporally consistent surface-water extent by combining two-stage adaptive thresholding with connectivity-, geometry-, and temporal-stability constraints, yielding a surrogate river discharge (SR) time series. Across 19 rivers in 13 countries, median level Pearson rwas 0.71; elapsed-time-normalized change rwas 0.28, and 18 of 19 combined series remained significant under autocorrelation-preserving circular shifts. Temporal persistence therefore contributed to, but did not fully explain, most contemporaneous associations. Performance was weaker in levee-confined, storage-influenced, persistently low-flow, and steep reaches. As a separate hydrological demonstration, SAR-derived SR was integrated into a parsimonious rainfall–runoff model. In the Imjin and Daly basins, SAR-based SR produced higher median model performance than SMOS L-band surrogates, although intermittent sampling increased sensitivity to stochastic calibration. These results define both the potential and the practical limits of SAR-observed water extent as a streamflow surrogate in data-restricted basins.