Multi-Resolution Optical Satellite–Based Water Body Detection Using CNN

Water body spatial distribution and dynamic characteristics provide essential information for water resource management as well as hydrological and environmental analyses. In regions where rivers and reservoirs of various sizes are intermingled, monitoring approaches capable of continuously capturing changes in water body morphology and boundaries are required. Satellite remote sensing data have been widely used for water body detection due to their ability to provide large-scale observations, and optical satellite imagery enables relatively accurate water body extraction through water index–based analyses using near-infrared reflectance characteristics. However, differences in spatial resolution among satellite sensors limit the representation of small water bodies and complex boundaries. To address these limitations, medium-resolution Sentinel-2 imagery and high-resolution CAS500 optical imagery were employed, and the applicability and performance of a convolutional neural network (CNN)–based water body detection approach were investigated. The Normalized Difference Water Index (NDWI) was used as the primary input feature, and a CNN model based on the HRNet architecture was applied for water body detection. The analysis was conducted under three scenarios: water body detection using Sentinel-2 imagery, water body detection using CAS500 imagery, and transfer-based water body detection in which a model trained on Sentinel-2 imagery was applied to CAS500 imagery. Results indicate that spatial resolution differences and training data characteristics influence water body detection performance, demonstrating the potential of integrating multi-resolution optical satellite data with CNN-based methods. This approach provides a useful foundation for efficient water body monitoring, particularly in environments where the availability of high-resolution satellite data is limited.Keywords: Water body detection, CAS500, HRNet, Transfer learningAcknowledgmentThis research was supported by the BK21 FOUR (Fostering Outstanding Universities for Research) funded by the Ministry of Education (MOE, Korea) and National Research Foundation of Korea (NRF). This work is financially supported by Korea Ministry of Land, Infrastructure and Transport (MOLIT) as 「Innovative Talent Education Program for Smart City」. This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Water Management Program for Drought Project, funded by Korea Ministry of Climate, Energy and Environment(MCEE)(RS-2023-00230286). This work was supported by Korea Environment Industry & Technology Institute (KEITI) through Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change Project, funded by Korea Ministry of Climate, Energy and Environment(MCEE)(RS-2024-00332300). This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Technology development project to optimize planning, operation, and maintenance of urban flood control facilities, funded by Korea Ministry of Climate, Energy and Environment(MCEE)(RS-2024-00398012).