Network Connectivity-based Stream Classification for the Conterminous United States
Researchers have developed a novel framework and dataset called NetConUS for classifying streams across the conterminous United States based on network connectivity. Published in Scientific Data, this study addresses the limitation of existing classification schemes that primarily focus on physical habitat and hydrology while neglecting river network connectivity. By utilizing the National Hydrography Dataset Plus version 2 (NHD), the team categorized streams into five types: central, peripheral, mainstem, cluster, and convergent. These classes are differentiated using metrics such as degree centrality, eigen-vector centrality, clustering coefficient, closeness centrality, and betweenness centrality. The classification system was validated using Bayesian Neural Networks to account for uncertainty in class assignments. This dataset captures the structural roles of stream segments within river networks, offering new opportunities to integrate connectivity metrics with geophysical descriptors in freshwater fauna analyses. The research, supported by the National Science Foundation and USDA, aims to advance riverine research and improve the management of freshwater ecosystems at large spatial extents by highlighting how connectivity influences the movement of water, nutrients, sediments, and aquatic species.
Wire timeline
Network Connectivity-based Stream Classification for the Conterminous United States
Researchers have developed a novel framework and dataset called NetConUS for classifying streams across the conterminous United States based on network connectivity. Published in Scientific Data, this study addresses the limitation of existing classification schemes that primarily focus on physical habitat and hydrology while neglecting river network connectivity. By utilizing the National Hydrography Dataset Plus version 2 (NHD), the team categorized streams into five types: central, peripheral, mainstem, cluster, and convergent. These classes are differentiated using metrics such as degree centrality, eigen-vector centrality, clustering coefficient, closeness centrality, and betweenness centrality. The classification system was validated using Bayesian Neural Networks to account for uncertainty in class assignments. This dataset captures the structural roles of stream segments within river networks, offering new opportunities to integrate connectivity metrics with geophysical descriptors in freshwater fauna analyses. The research, supported by the National Science Foundation and USDA, aims to advance riverine research and improve the management of freshwater ecosystems at large spatial extents by highlighting how connectivity influences the movement of water, nutrients, sediments, and aquatic species.
Scientific Data