An analysis of similarity of digital river networks with different resolutions in various river basins

In order to apply hydrological models and establish computer aided hydrological discharge nowcast eventually forecast system for large regions like Asian continent or whole world, accuracy of digital river network is one of the key issue to produce useful information. In this study, we propose and apply a method which analyses the similarity of digital river networks with different spatial resolutions. We use the coordinates and upstream catchment areas at major confluent points in order to represent the features of river networks. Turning counter- clockwise around the river network from the estuary, we arrange confluent points into one string. The sets of the one-dimensional data obtained from river networks are compared by using DP matching. A set of river networks with different scales ranging from 90 m to 20 km grids were developed based on this scaling algorithm (Masutani et al., 2006) using 90m HydroSHEDS. We evaluated the developed dataset using the proposed method to small to large rivers in Asian Regions (including Mekong River, Yellow River. etc.). A possibility of utilizing the proposed method was successfully shown. The proposed method and the river network dataset developed can provide better hydrological simulation with different grid scales with little effect.