Improving flood simulation capability of the WRF-Hydro-RAPID model using a multi-source precipitation merging method

• Published in: Journal of hydrology (Amsterdam) 2021-01, Vol.592, p.125814, Article 125814
• Main Authors: Chao, Lijun, Zhang, Ke, Yang, Zong-Liang, Wang, Jingfeng, Lin, Peirong, Liang, Jingjing, Li, Zhijia, Gu, Zhao
• Format: Article
• Language: English
• Subjects: Flood simulation capability; Model evaluation; Multi-source merged precipitation; Precipitation validation; WRF-Hydro-RAPID model
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2020.125814

•Developed an a evaluation method for distributed hydrological models using nested stations.•Investigated the applicability and forecast capacity of the WRF-Hydro-RAPID model.•WRF-Hydro-RAPID shows robust and consistent performance at the basin outlet and interiors.•A multi-source precipitation merging method further enhances the model flood forecasting capability. Flash floods-caused losses are rapidly increasing due to climate change induced extreme weather events and economic development in the world. The WRF-Hydro-RAPID model coupled with land surface model and a vector-based flow routing module is able to simulate discharge at any reach of a watershed, making it a good tool for flood simulation and forecasting. We investigated the flood simulation capability of the WRF-Hydro-RAPID model and evaluated the utility of a multi-source precipitation merging method based on the mixed geographically weighted regression model and Bi-square function (MGWR-BI algorithm), which produces precipitation as forcing with improved quality and resolution, to enhance the simulation accuracy of the WRF-Hydro-RAPID model for the Daheba Watershed, a first-order sub-basin of the Yangtze River Basin. The merged precipitation data have substantial higher quality than the downscaled original CPC MORPHing technique satellite precipitation data (CMORPHd) (r = 0.64–0.74 and RMSE = 1.59–6.64 mm/h for the merged data vs. r = −0.11–0.06 and RMSE = 3.31–8.25 mm/h for the CMORPHd data) by comparing to the ground observations. Floods are better forecasted and simulated by the WRF-Hydro-RAPID model driven by the merged precipitation data than the CMORPHd precipitation data for the four nested medium and small watersheds. Performance of the WRF-Hydro-RAPID model at the watershed outlet station does not differ those at the three inner stations, proving that the WRF-Hydro-RAPID model has a consistent performance in space. The combination of the WRF-Hydro-RAPID with the precipitation merging method makes it a valuable tool for flood simulation of medium and small watersheds.