Spatial propagation of streamflow data in ungauged watersheds using a lumped conceptual model

Abstract

Streamflow data are required for the effective management of flood damage; however, streamflow is rarely measured in small watersheds. In this study, a lumped conceptual model was adopted to generate streamflow values at small, ungauged watersheds using the spatial propagation concept. During the process of spatial propagation, parameters related to physical properties were fixed, while event-based initial conditions were optimized to ensure that error rates on the simulated streamflow data were comparable to those for measured data. Then the concept was validated using data from 21 flood events in the ChungJu Dam (CJD) watershed, Korea, which was divided into 22 small, ungauged watersheds. The propagated peak streamflow data at ungauged cross-validation points have average Nash-Sutcliffe efficiency (NSE) values of 0.91–0.99. Averages of NSE values for volume and time to peak streamflow are over 0.85 which satisfies the model criterion (NSE > 0.5). It is concluded that streamflow data for small, ungauged watersheds located upstream can be generated by one gauged downstream streamflow without an extensive amount of gauged streamflow data from other locations using the proposed concept. For accurate simulation, availability of rainfall data is essential for accurately modeling the spatial propagation of streamflow using a lumped conceptual model.