Remote Sensing of Active Water Storage in Reservoirs and Lakes

The method follows a six-step procedure:

1. Generate Surface Water Occurrence Probability (SWOP) maps from optical satellite imagery (GEE JavaScript).
2. Download ICESat-2 laser altimetry tracks (R and Python).
3. Match SWOP with terrain heights to generate the bathymetry DEM of the reservoir (R).
4. Upload the DEM to Google Earth Engine (Python).
5. Map deposition and erosion patterns (GEE JavaScript).
6. Derive time series of water levels using Sentinel-2 and Landsat 7/8/9 imagery (GEE JavaScript).
7. Export monthly median values of water level, water volume and water area per reservoir (GEE JavaScript).

The sediment balance analysis in Step 5 ensures the accuracy of water level time series (Steps 6 and 7) by identifying and excluding shoreline pixels with unstable terrain, preventing erroneous water level measurements due to sediment deposition or erosion. Step 5 of the workflow builds on a previously published Google Earth Engine script for mapping deposition and erosion patterns in lakes and reservoirs (Ragettli et al., 2022b). However, the earlier approach relied on an existing DEM, while with the present scripts we generate reservoir bathymetry in Steps 1–4, making the method independent of in-situ data and adaptable to different reservoirs. 

 

The methodology is suitable for (operational) monitoring of the water level in reservoirs and lakes, as showcased by the following Earth Engine apps:

- Sahel-Water: Monitoring Water Bodies and Wetlands in the Sahel using Continental Scale Remote Sensing

- India-Reservoirs: Reservoir Bathymetry and Water Accounting in Irrigated Systems of India using Earth Observation Technology