Reservoir metrics estimated by remote sensors based on the Google Earth Engine platform

Abstract

Artificial reservoirs produce most of the electric power generated in Brazil, and they are essential for water supply. Hence, sustainable management of water reservoirs, along with constant monitoring of water quality and quantity through effective and low-cost technologies, such as remote sensing, is vital to avoid hydrological stress and prevent water crises. However, monitoring long-term time series of reservoir parameters can be limited by data availability and the computing power of the available machinery. Therefore, the present study proposes a method that combines the use of a digital elevation model (Shuttle Radar Topography Mission [SRTM]) and Landsat images with the Google Earth Engine cloud processing platform to monitor multiple reservoir metrics (surface area, water level, and useful storage). The method was applied to study these metrics between 2009 and 2020 for the Corumbá IV water dam, Goiás, Brazil, where 24 mosaic images, one for each 6-month period, were obtained by filtering cloudy pixels and selecting pixels of better quality. Composite RGB images of short-wave infrared, near-infrared, and red (SWIR1/NIR/R) bands were transformed to HSV to classify the water pixels and estimate the surface area of the reservoir. Area estimates were then 1) used to assess water level and useful storage through linear fitting (regression method [RM]); and 2) combined with SRTM elevation to obtain water level and useful storage (elevation method [EM]). Metrics of rainy and dry seasons that were calculated using linear RM of surface area estimates produced good results, with a coefficient of determination of 0.74 and 0.78 for determining water level and useful storage, respectively. However, the use of SRTM elevation information (EM) provided an even more efficient estimation method, with coefficients of determination of 0.84 and 0.95 for water level and useful storage estimates, respectively.