Development and Application of Stream Network Dataset Based on Actual River Channel Geometry in South Korea

Stream environments exhibit complex network structures characterized by linear connectivity among segments and interactions with terrestrial conditions, such as catchments and watersheds. Therefore, developing digital representations of stream networks that integrate geomorphological, environmental, and climatic information is crucial for understanding key factors shaping aquatic ecosystems. Here, we developed a stream network dataset for river basins of South Korea, incorporating segment-level topographic attributes, catchment characteristics, and climate variables. Stream network was delineated using a 30-m resolution digital elevation model. Each segment was assigned nine basic attributes (e.g., segment ID, basin ID) and 22 environmental variables, including elevation, slope, stream order, upstream watershed area, and land use type. In addition, 19 bioclimatic variables were extracted from historical and four future time periods based on shared socioeconomic pathways. All variables were spatially linked to their corresponding stream segments and catchments to construct an integrated dataset. The developed stream network encompassed 21 river basins, comprising 31,381 stream segments and associated catchments. Validation with previous datasets (e.g., Korea reach file and the Water Resources Management Information System) showed a mean absolute percentage error below 0.2% across the five major basins, confirming the spatial accuracy of delineated stream networks. To demonstrate applicability, we used a MaxEnt model to predict future distributions of the cold-water fish, <i>Rhynchocypris kumgangensis</i>. Results of modeling well explained ecological characteristics and potential distribution shifts under climate change. Overall, the developed stream network can serve as an important dataset in aquatic ecology, hydrology, and environmental sciences across river basins in South Korea.