The bathymetry of a water body can be measured in the field using bathymetric probes; however this process is costly and time-consuming. Thus, the main objective of this study was to fit bathymetry estimation models of a lake from satellite data. The study was conducted in the Saco reservoir, located in the Pajeu river basin near the city of Serra Talhada - PE, Brazil. Regression models were adjusted between the bathymetry values collected with a probe, the green and near infrared spectral bands of the Landsat 5 TM satellite and the normalized difference water index. Evaluation of the adjusted models was performed by analysis of the coefficient of determination, mean squared error and model efficiency. The bathymetry of the reservoir varied between 0.3 and 10 meters deep. The best fits were found when using the power and exponential models, with coefficients of determination greater than 0.46 and model efficiency close to one. It can be concluded that the bathymetry models adjusted to the remote sensing data provided a satisfactory accuracy for estimating bathymetry values of the Saco reservoir, especially in the shallow regions of the lake.
Landsat; NDWI; Pajeú river; Remote sensing
