Remote-sensing-based characterizing and storage monitoring of dam reservoirs: bathymetry, capacity curves, sedimentation, and loss of storage capacity of the Ouled Mellouk reservoir (Algeria)

This paper presents a remote sensing-based approach for dam reservoirs characterizing and monitoring using the Ouled Mellouk reservoir (Algeria) as a demonstration. This reservoir experienced a continuously decreasing water volume (during 2019–2021), reaching zero storage by Jul. 2021. This aspect was exploited to produce a new bathymetric map of the reservoir by stacking water surfaces contours (at selected dates) delineated from remote sensing products. This new bathymetry (representing the reservoir’s state in 2021) is then used to construct new capacity curves (linking water levels to water volumes and surfaces) for the reservoir. The comparison of the new bathymetric map to a previous in-situ map (from 2006) allowed the assessment and mapping of the sedimentation of the reservoir and the associated loss of storage capacity. It was found that the reservoir has accumulated 8.57Hm3 of sediments (equivalent to a 2.40 m thick layer of sediments) during the period 2006–2021 and lost 7Hm3 of its storage capacity. In the second part, water surface delineation using automated binary threshold-based classification techniques (using the Otsu algorithm) is employed for both multispectral (Landsat-8 and Sentinel-2) and SAR (Sentinel-1) imagery. The study evaluates the use of several optical spectral indices and the single or combined use of SAR polarizations to detect water surfaces. The performance assessment was done against field observation data. The MNDWI and S2WI water indices were found to be the best-performing for Landsat-8 and Sentinel-2 data, respectively, and the VV×VH polarization combination for Sentinel-1 data. The best-performing optical and SAR classifications are then fused to increase the temporal resolution (up to 1-day) of the water surface, level, and volume monitoring. The fused data fitted well field observations (R2 = 0.972). The proposed approach is a reliable and cost-effective solution for assessing sedimentation and monitoring surface waters and can support decision-making.