Sentinel-2 MSI image time series reveal hydrological and geomorphological control of the sedimentation processes in an Amazonian hydropower dam

Abstract

Monitoring sediment transport in a run-of-river hydropower dam was investigated to detect and quantify the siltation processes using Sentinel-2/MSI satellite images. We developed a method allowing us to assess the fate of sediment discharge in reservoirs and to map the locations of eroding and silting stream cross sections. This monitoring was achieved by retrieving the seasonal and interannual variation in suspended sediment concentration (SPM) longitudinally and by developing indices to detect local deviations from the overall SPM trend. A comparison between Sentinel-2 reflectance, processed for atmospheric and glint correction, and field measurements showed fine accuracy for the b8a/b4 band ratio (RMSE = 0.23 sr-1, N = 50) as well as a very good SPM retrieval performance for a wide range of concentrations [10–926 mg/L] with an RMSLE of 0.23 mg/L. SPM trend analysis demonstrated that significant (p < 0.001) sedimentation processes (i.e., decreases in SPM concentration upstream to downstream) occur for most parts of the hydrological cycle except during peak flooding when there was no significant SPM siltation. The overall rate of SPM decrease is −0.52 mg/L/km, varying from +0.11 mg/L/km down to −1.10 mg/L/km. We developed a deviation index to detect local anomalies from the overall upstream–downstream satellite-derived SPM trend. A satellite-derived map of eroding and silting reaches is produced evidencing the control of the sediment transport pattern by geomorphological features, such as rapids. Using independent bathymetric surveys, we showed that satellite data allow the locations of eroding and silting river reaches to be correctly mapped.