원격탐사 기반 미계측 저수지 수체탐지 및 변화 분석
Remote Sensing-Based Assessment of Water Body Dynamics in Ungauged Agricultural Reservoirs
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Understanding reservoir storage volume is critical for ensuring a stable water supply and effective management of agricultural reservoirs, which serveas essential infrastructure for irrigation. This study aims to detect and analyze temporal variations in water bodies within ungauged reservoirs, wheredirect water level measurements are not available, through the application of remote sensing techniques. Optical satellite imagery from the Landsat andSentinel-2 missions was processed using the Google Earth Engine (GEE) platform. The Normalized Difference Water Index (NDWI) was calculatedto enhance water body identification, and the canny edge detection algorithm combined with Otsu thresholding was employed to delineate reservoirboundaries and estimate water surface area. In the case of a gauged reservoir (CT), an R2 value of 0.7 was observed between satellite-derived andobserved water surface areas during the period of reservoir storage depletion, indicating high accuracy. For the ungauged reservoirs TB and BD,correlation coefficients of 0.35 and 0.61, respectively, were obtained, with improved accuracy associated with larger reservoir size. High accuracy wasalso achieved in estimating full water surface areas; however, correlation tended to decline during periods of dynamic water level fluctuation. Thesefindings suggest that enhanced spatial resolution in future satellite missions will further improve the detection and monitoring capabilities for waterbodies in ungauged agricultural reservoirs.
Abstract
Understanding reservoir storage volume is critical for ensuring a stable water supply and effective management of agricultural reservoirs, which serve as essential infrastructure for irrigation. This study aims to detect and analyze temporal variations in water bodies within ungauged reservoirs, where direct water level measurements are not available, through the application of remote sensing techniques. Optical satellite imagery from the Landsat and Sentinel-2 missions was processed using the Google Earth Engine (GEE) platform. The Normalized Difference Water Index (NDWI) was calculated to enhance water body identification, and the canny edge detection algorithm combined with Otsu thresholding was employed to delineate reservoir boundaries and estimate water surface area. In the case of a gauged reservoir (CT), an R2 value of 0.7 was observed between satellite-derived and observed water surface areas during the period of reservoir storage depletion, indicating high accuracy. For the ungauged reservoirs TB and BD, correlation coefficients of 0.35 and 0.61, respectively, were obtained, with improved accuracy associated with larger reservoir size. High accuracy was also achieved in estimating full water surface areas; however, correlation tended to decline during periods of dynamic water level fluctuation. These findings suggest that enhanced spatial resolution in future satellite missions will further improve the detection and monitoring capabilities for water bodies in ungauged agricultural reservoirs.