Hide/Show Apps

Detection of reservoir water levels using landsat remote sensing data over Ermenek and Altınkaya dams

Şenocak, Ali Ulvi Galip
Detection of water border using remote sensing observations at the visible bands and incorporating them with the digital elevation map is a useful approach for detecting water volume of dams and the water bodies with existing DEM images. In this study, NDWI, NDPI, WI2015 and AWEI indices retrieved using Landsat 8 images and ASTER/SRTM DEM maps are utilized to infer about the water levels of Ermenek and Altınkaya dams’ reservoir water levels. To reduce the water level retrieval errors during the cloudy and the snow-covered areas, F-Mask cloud masking algorithm and a TCW-based custom index with optimized parameters have been introduced. Moreover, in order to prevent the affection of pixels that are located far away from the area of interest, a water-area-based shape file and proximity buffer have been introduced. Lastly, after the completion of the analysis, a statistical model has been applied to combine the results with DEM to get the elevation value as a result. Results show RMSE of the water level estimation over Ermenek and Altınkaya are 3.63 m and 3.34 m, respectively for the best index/DEM scenario when the models are trained and calibrated over the same dam. On the other hand, the errors increase to 5.13 m and 5.09 m respectively for Ermenek and Altınkaya dams when the validation and the calibration are done over different dams