Sentinel-2 Bathymetry Indicates River-Mouth Migration and Asymmetric Sediment Redistribution at Cua Dai, Vietnam

Abstract

Satellite-derived bathymetry (SDB) can support monitoring of dynamic river mouths, but its interpretation depends on survey–image timing, validation design and spatially explicit uncertainty. This study applies an established Sentinel-2 blue–green log-ratio workflow to Cua Dai, Vietnam, using four archived image–echo-sounding comparison data sets from 2016–2018. Recalculation from all valid paired records (n = 1,974–3,483 per date) yielded bias from −0.184 to +0.259 m, MAE (mean absolute error) of 0.394–0.651 m, RMSE (root mean square error) of 0.495–0.849 m and predictive R2 of 0.751–0.924. The same-day Sept 2018 comparison performed best; the other image–survey pairs were separated by 36–61 days. Inter-date maps indicate alternating deepening and shoaling, river-mouth/bar reorganisation and an asymmetric pattern of northern vulnerability and southern deposition. Provisional LoD95 values are 2.19 m for Apr 2016–May 2017 and 1.73 m for May 2017–Sep 2018, so smaller differences are treated as unresolved. The latter interval spans the Nov 2017 flood and about ten subsequent months and therefore cannot isolate flood effects. A wave-shadow/sediment-connectivity explanation is presented as a testable hypothesis rather than a demonstrated mechanism. The revised analysis shows how SDB can guide targeted field monitoring while preventing unsupported volume estimates and causal attribution.