Research On Coupled Forecast Of Atmosphere And Hydrology In Qingjiang River Basin Based On Data Assimilation

The majority of hydrological models used for flood forecasting are driven by gauge rainfall data.In this situation,the leading time is generally short.It is an effective way to couple the quantitative precipitation forecast(QPF)to a distributed hydrological model with the goal of extending leading time for flood forecasting.However,the QPF products contain a certain degree of uncertainty,and would affect the accuracy of flood forecasting,especially in the mountainous regions with complex terrain.While data assimilation has been proved to be an important method to improve the quality of QPF and further improving the flood forecasting.In this paper,the data assimilation was applied to construct a high resolution atmospheric-hydrological coupling model based on the WRF and WRF-Hydro models in the Qingjiang River basin.Firstly,the quality of the QPF from the Numerical Weather Prediction(NWP)model based on conventional observational and radar data assimilation were evaluated in the basin during eight typical flood events between 2017 and 2018.Then the parameters of WRF-Hydro model were calibrated in the Qingjiang River basin,and the flood simulation results of the model of thirteen flood events between 2015 and 2018 were analyzed.Finally,the atmospheric-hydrological coupling forecast studies were conducted based on the eight flood events between 2017 and 2018 in order to analyse the impacts of data assimilation on flood forecasting capability of the coupling model,and the main conclusions are as follows:(1)The QPF results from the NWP model with data assimilation show that the error of the forecast areal rainfall peak time is reduced overall with conventional observations assimilated and the shape of the forecast areal rainfall is more accurate.In addition,the uncertainty of the shape of the forecast areal rainfall among the eight flood events is decreased.However,the uncertainty of the forecast rainfall volume still exists.While assimilating radar data,the peak time of the forecast areal rainfall are more accurate,and the error of the areal rainfall peak volume is smaller compared with conventional observations assimilation.In addition,the uncertainty of the QPF among the eight events is decreased significantly.In general,the QPF of the Qingjiang River basin is improved with radar data assimilation.(2)The flood simulation results from the distributed hydrological model with the model parameters calibrated show that the flood simulation results are more sensitive to the infiltration coefficient and the Manning's roughness for river channel than the maximum retention depth and the Manning's roughness for overland.The calibrated parameters could express the characteristics of the Qingjiang River basin,including the steep topography,the narrow and steep river channel,and the high forest coverage.The thirteen flood events are simulated well overall after the model parameters calibrated,and the WRF-Hydro model can be applied to the Qingjiang River basin with complex terrain for flood simulations.In addition,it is found that the streamflow simulations of single-peak flood events are generally better than that of multi-peak flood events.The simulation deviation of the multi-peak events is mainly manifested in the over-estimation of the flood peak intensity.(3)The flood forecast results of the atmospheric-hydrological coupling model show that flood forecasting is highly dependent on the QPF.While assimilating conventional observational data,the shape of the forecast hydrographs is improved.However,the uncertainty of the forecast flood volume among different flood events still exists.While assimilating radar data,the shape of the forecast hydrographs is more accurate in general and the error of the forecast flood peak volume is also smaller,thus the Nash-Sutcliffe efficiency coefficient of the forecast streamflow is significantly improved.In addition,the uncertainty of the flood forecast results among different flood events is also distinctly decreased with the radar data assimilation.In conclusion,radar data assimilation can improve the flood forecasting effectively based on the atmospheric-hydrological coupling model.