Remote sensing of a transboundary Chinese reservoir cascade and its influence on Lai Chau reservoir in the upper Da river basin

Abstract

Satellite-based monitoring of reservoir operations in transboundary, data-scarce basins remains a major challenge due to limited in situ observations and restricted data sharing. This study develops an end-to-end workflow on Google Earth Engine to monitor a cascade of nine Chinese reservoirs in the upper Da River and assess their influence on Viet Nam’s downstream Lai Chau reservoir. DEM-derived area–elevation–volume curves are generated from NASADEM and reservoir polygons, while a fast Sentinel-1 OtsuSmart classifier produces 7-day water-surface area series that are converted to water level and storage. Validation against survey curves for Lai Chau and Huoi Quang yields r = 1.00, with RMSE values of 1.86 and 0.23 km², respectively, while reconstructed Lai Chau water levels achieve r ≈ 0.80 and RMSE ≈ 5.6 m. Cascade and regulation indices along the Jupudu–Gelantan–Tukahe–Lai Chau chain show that Chinese operations exert first-order control on Lai Chau’s seasonal water-level regime, with positive correlations and 0–2-month lags at the monthly scale, while most net dry-season release originates from upstream branches. Daily lag-correlation analysis further reveals tight synchrony between Gelantan and Tukahe but a compensating, one-week out-of-phase response at Lai Chau, indicating active buffering. The framework uses only global open data and is transferable to other transboundary basins. Keywords: Transboundary river, sentinel-1, storage dynamics, Da river basin