Document Type : Complete scientific research article
Authors
1
MSc Student, Water Engineering, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran
2
Associated Prof, Water Engineering Department, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran
Abstract
Background and Objective
Water resources have always been a critical issue in human life. Reservoirs, as one of the key water sources, require accurate assessment of their stored water volume for optimal utilization and planned management. Traditional methods for determining water volume, relying on water surface elevation and the volume-depth curve, often necessitate costly and time-consuming corrections due to factors such as floods. This research proposes a new approach to estimate the volume of water stored in a dam reservoir using the relationship between satellite images and water depth, aiming to enhance water resource management efficiency and cost-effectiveness.
Materials and Methods
To estimate the stored water volume in the Zujar dam reservoir (with a maximum storage capacity of 3.2 billion cubic meters) using remote sensing based on the water depth estimation, Landsat8 OLI satellite images were downloaded. After applying radiometric corrections, bands and spectral indices related to different pixels of the image were extracted. Due to the huge input data matrix and the time-consuming nature of multivariate linear regression modeling, a code was developed in MATLAB. The new dataset was then introduced to the Minitab software for linear regression equation fitting. Measured water depths from the DAHITI database were considered as dependent variables, while bands and spectral indices selected as independent variables for the multivariate linear regression.
Results
The results obtained from the water depth equation at three different time intervals (2013, 2019, and 2020) showed that the minimum and maximum root mean square error (RMSE) values in depth calculation were 1.00 and 1.35 meters, respectively, with an average of 1.21 meters. Moreover, the minimum and maximum errors in estimating water stored volume were 3.88% and 14.85%, respectively, with an average of 9.25% for three dates that said. Considering that observed water depths during this period ranged from 16.5 to 39.5 meters, the results indicate acceptable accuracy. Analyses revealed that the normalized difference water index (NDWI) and near-infrared (NIR) band from the spectral indices of satellite imagery had the highest significant correlation with water depth, with coefficients of determination of 0.94 and 0.85, respectively.
Conclusion
The obtained results suggest that a linear regression relationship can be established between measured water depth and extracted spectral bands from satellite images for depths up to approximately 40 meters. This not only ensures accuracy in water depth estimation but also provides acceptable precision in estimating the volume of water stored in the reservoir. Improving and enhancing this approach could enable long-term volume estimation of reservoirs, contributing to better water resource management.
Keywords: Water depth estimation, Volume of stored water, Satellite imagery, Zujar Dam, Multivariate linear regression, DAHITI database.
Keywords
Main Subjects
)