Refining a Distributed Linear Reservoir Routing Method to Improve Performance of the CREST Model

As one of the most important components of hydrologic models, routing module determines model performance to a large degree. In this study, the authors proposed a fully distributed linear reservoir routing scheme that is more compatible to fully distributed hydrologic models than the previous version of the cell-to-cell routing scheme proposed in the coupled routing and excess storage (CREST) model. The new scheme effectively solves issues existing in the previous version of the cell-to-cell routing scheme in CREST, including the tendency to underestimate channel flow and discontinuous flow values along the river after storms. Tested over three basins with two in the United States and one in China at spatial resolutions from 250 m to 1 km and temporal resolutions from hourly to daily, the proposed scheme is proved spatiotemporal scalable. At daily scale, the Nash-Sutcliffe coefficient (NSCE), correlation coefficient (CC), and absolute relative bias range from 0.65, 0.82, and 10.3% to 0.90, 0.95, and 1.57% in the three tested basins respectively. At hourly scale, CREST v2.1 successfully replicated 11 out of 12 flood events during 2002–2013 in the Tar and Kankakee Rivers with mean NSCE, CC, and absolute differences of peaks being (0.78, 0.95, and 10.9%).