Monitoring time series of reservoir water surface area changes using remote sensing approaches

Background and objectives
Dam reservoirs are considered as one of the most important sources of water supplies which estimating of their water surface area is necessary for many hydraulic and hydrodynamic subjects. Estimating of the water surface area is crucial for flood routing, diffusion and transport of pollutants, and modeling of reservoir Thermal Stratification. Due to the development of remote sensing and improvement
of the quality of satellite images, it is possible to derive valuable information about the trend of water area’s changes and evaluate them in long-term time series. In this research, a precise method based on remote sensing has been presented with the purpose of calculating the water surface area at any water level of reservoirs and lakes

Materials and methods
Monthly images of the Landsat-8 satellite were downloaded from April 2013 to September 2023 for the Zujar dam reservoir located in Spain with longitude -5/2318 °W and latitude 38/9295 °N from the archives of the United States Geological Survey (USGS) site. In order to validate the results of this study, Database of Hydrological Time Series of Inland Waters (DAHITI) has been used including the hydrological data of the lake and reservoirs of different regions from 1992. After radiometric correction, by combining five water indices MNDWI، NWI ،AWEIsh ،AWEInsh and TCwet, a threshold between water mask and land mask was detected for each monthly image which included the gaps in data caused by shades, clouds, cavities, ice and etc. To determine the status of the pixels, the status of water, land, or gap is determined. Using a identification threshold, monthly images were classified as land-water masks that included the gap in the data. To fill these gaps, a long-term water probability graph was computed that each gap pixel in monthly land-water mask was compared with its value in longterm probability. With the iteration method in different water probabilities and using an equation that minimized area between the monthly water area and the long-term water probability, the area of the data gaps was filled for each monthly image and finally a ten-years’ time series of water surface area was formed
Results
The water surface area values calculated in a ten-years period showed that the largest water area of the Zujar dam reservoir was at 351.2 m water level corresponding to April 2013 and the lowest water area at 316.1 m water level corresponding to September 2023 that are equal to 140.3 and 17.8 Km2, respectively. Investigations showed that the water surface area of the reservoir had a continuous downward trend in the last decade that climate changes can be one of the most important resoans. The comparison between the obtained results with the measurements of the database showed that the largest difference in the calculation of the water surface area was in July 2016, at the water level of 344.4 m, equal to 4.3 Km2 (3.6%), and the average error in the ten-year period was about 2.5%

Conclusion
Although the water surface area of the reservoir at each level can be calculated with the help of the initial of the area-elevation curve of the reservoir, but this curve changes over time due to various reasons such as the deposition of sediment by floods. Therefore, water surface area has a dynamic behavior and its time series variations should be considered. The proposed method in this study based on remote sensing was able to estimate the water surface area of the dam reservoirs in a long-term time series of ten years with high accuracy by separating water and land areas