광학 및 SAR 영상 기반 농업용 저수지 수체의 시·공간적 변화 모니터링 = Enhancing agricultural reservoir water body monitoring through the integration of optical and SAR remote sensing

다국어 초록 (Multilingual Abstract)

Changes in water bodies are critical indicators for diagnosing and predicting extreme hydrological hazards, and the ongoing decline in storage across dams and reservoirs has recently posed significant challenges for water resource management. As a result, there is a growing demand for reliable systems capable of monitoring the spatiotemporal dynamics of reservoir water bodies. In this study, we quantitatively compared optical and SAR-based methods for monitoring reservoir dynamics using Sentinel imagery processed on the Google Earth Engine (GEE) platform. The optical indices—the Normalized Difference Water Index (NDWI) and the Modified Normalized Difference Water Index (MNDWI)—generally produced lower RMSE values and closely matched observed water surface areas; however, their reliability diminished during periods of reservoir drawdown due to indistinct boundaries and data gaps. In contrast, Synthetic Aperture Radar (SAR)-based approaches—including the Kittler-Illingworth (KI), Chan-Vese (CV), and K-means (KM) algorithms—offered greater year-round data availability and more stable time-series detection. Among these, the CV method provided the best balance between accuracy and robustness, yielding an R² of 0.7 for the Geumgwang Reservoir, an RMSE of 7.6 ha for the Madun Reservoir, and an overall classification accuracy of 87% relative to observed water surface areas. Overall, both optical- and SAR-based approaches effectively captured reservoir variations during periods of rising water levels, though performance declined under stable or decreasing conditions. Future research should focus on developing segment-specific corrections and integrating optical and SAR imagery to establish a continuous monitoring framework for medium- to large-scale agricultural reservoirs, with potential application as a core component of national hazard monitoring systems.