Mapping Post-Flood Water Persistence with SWOT in Low-Gradient Drylands, Australia

Western Queensland’s dryland floodplains are exceptionally low-gradient, with broad anabranching channels and extensive backswamps where the key driver of pastoral impact is not only flood magnitude but the residence time of standing water. Prolonged ponding induces waterlogging and sediment smothering of pastures, creating multi year feed deficits. Leveraging the exceptional rainfall flood sequence of early 2025 as a natural experiment, we investigate the space–time structure of post-flood water persistence across the Cooper Creek floodplain near Windorah and adjacent Channel Country. 

We exploit SWOT KaRIn Level-2 “pixel-cloud” hydrology to retrieve day-by-day water surface elevation and inundation and pair these with Bureau of Meteorology rainfall analyses to decouple local rainfall forcing from upstream flood-wave contributions. A conservative, class-aware filtering is applied to stabilize water detection; we then compute persistence diagnostics (e.g., pixel-wise residence time and a backswamp/active-channel persistence contrast index), quantify observational uncertainty via threshold perturbations, and cross-check timing and extent against independent situational reporting. The workflow is designed to be reproducible and extensible, providing assimilation-ready surface water states suitable for integration with hydrodynamic models or multi-sensor frameworks. 

By centering persistence rather than peak alone, this study targets a hydrologically meaningful and impact-relevant variable. The contribution is a rigorously specified method for mapping where water lingers after major floods in very low-slope rangelands, with clear pathways to generalization beyond the Cooper system and to fusion with national precipitation reanalyzes and river intelligence products for improved flood-recession understanding, forecasting, and rangeland decision-support.