Satellites have the capability to continuously collect data over large areas andprovide valuable information for various applications, including land and watermanagement, disaster prediction and response, and more. In the field of water resources management, researches on waterbody detection and their changes using satellite images have recently been conducted in various regions in South Korea, utilizing multiple types of sensors.
This study utilizes optical satellite images from Landsat and Sentinel-2, basedon Google Earth Engine, to analyze long-term changes in surface water area for 10 small- and medium-sized water supply dams and agricultural reservoirsin South Korea. The analysis covers approximately 20 years for the six watersupply dams and 30 years for the four agricultural reservoirs.
By employing image analysis methods such as NDWI, Canny Edge Detection,and Otsu's thresholding for waterbody detection, the study reliably extractedwater surface areas, allowing for clear annual changes in waterbodies to be observed. When comparing the time series data of surface water areas derived from satellite images to actual measured water levels, a high correlation coefficient above 0.8 was found for the water supply dams. However, the agricultural reservoirs showed a lower correlation, between 0.5 and 0.7, which was attributed to the characteristics of agricultural reservoir management and the inadequacy of comparative data rather than the satellite image analysis itself. Theanalysis also revealed several inconsistencies in the results for smaller reservoirs, indicating the need for further studies on these reservoirs.
The changes in surface water area, calculated using GEE, provide valuable spatial information on waterbody changes across the entire watershed, which cannot be identified solely by measuring water levels. This highlights the usefulness of efficiently processing extensive long-term satellite imagery data. Basedon these findings, it is expected that ...