Flood Forecasting Research Of Xitiaoxi River Watershed Based On WRF Model And HEC-HMS Hydrologic Model

The leading time and accuracy of flood forecasting is directly related to the timeliness and effectiveness of flood control. The higher forecast accuracy and longer forecast period could bring the greater economic value for flood control and disaster reduction. In order to improve the accuracy of flood forecasting and prolong the forecast period of flood, one of the effective ways is to couple a high-resolution weather numerical model in flood forecast. The Weather Research and Forecasting model(WRF)-with its convenient and perfect forecasting system-has its usefulness and advantage in mesoscale simulation, especially for the rainfall simulation, however, the model is barely used in coupling flood forecasting at present. Xitiaoxi River Watershed, located in the Middle and Lower Reaches of the Yangtze River, of which climate and terrain characteristics were typical in the humid area of south-eastern China.Therefore, a flood forecasting experiment of the watershed was carried out based on numerical weather forecast model WRF and the distributed hydrological model HEC-HMS, with observations from rain gages and flow gauging stations over the Gangkou sub-region, which realized preferably the unilateral coupling of meteorological model and hydrological model,and established a real-time flood forecasting scheme for a mesoscale watershed. In this thesis,the following research parts was included:Firstly, a typhoon rainstorm flood case in August,2009and a Meiyu frontal rainstorm flood case in June,2011from five flood cases during2008-2011in the Xitiaoxi River Watershed according to its hydrological and climatic characteristics. Based on the NCEP reanalyzed meteorological data, after the analysis of the weather backgrounds and flood characteristics, in this thesis, the two cases were simulated for10roll days and the output fields was analyzed. The results showed as follows:WRF performed an ideal forecasting capability in the heavy rainfall area. The simulated rainfall bands,rainfall intensity and spatio-temporal pattern of average precipitation in the sub-watershed Xitiaoxi River had a good correspondence to the observations. The model has been working better in the forecasting of heavy rain, but it showed high a false acceptance rate for the light rains that the daily preciptation was.less than1mm (or the daily area preciptation was less than1mm). With the predicted time was elongated, the acceptability of quantitative analysis in the units of sub-watersheds reduced gradually and the average acceptability maintained in the level of50%. Also, the bigger area of a basin and the greater waviness of a undulate terrain could bring a bigger prediction errors of area preciptation. Overall, the reasonable and reliable simulation results from WRF can be used as an input to hydrological model for runoff simulation and flood prediction.Secondly, during the operation of HEC-HMS model, two kinds of methods were used to calculate runoff yield and flow concentration in the simulation of the precipitation-runoff processes of watershed systems on the basis of a large number of geographic information processing and estimation of the model parameters, and the effect of the hydrologic simulations was analyzed according to the acceptable errors in "Forecasting norm for hydrology intelligence"(GB-22482-2008-T). It showed that HEC-HMS model has a satisfying accuracy in the simulation of the rainfall and runoff processes of the Xitiaoxi River Watershed.In the calibration and validation periods, the fitting between the daily simulated flow and the observed daily flow was so good that the efficiency coefficients (N-S coefficient), the relative error of the peak flows and the peak time difference were less than the acceptable errors except the N-S coefficient of NO.20110611flood (Scheme2), and it indicated that the model has a good applicability in Xitiaoxi River Watershed. Especially, the simulating effect of HEC-HMS model in the flood with a single peak was better than the flood of two peaks and its parameter scheme was more applicable to the rainfall events of short duration. The Scheme one performed better than the Scheme two and it illustrates that the relative simple parameter schemes had a better practical value than the complicated schemes when the observed data and survey information of the selected watershed was scarce.Thirdly, considering an unidirectional coupling, the roll forecasting rainfall field (5kmx5km grid) with the leading time of48h output by WRF model was used to drive HEC-HMS model for a real-time roll flood forecast after the transformation of spatial scales in the rainfall processes and the two flood events in the watershed were forecasted successfully in real-time. The forecasting precision was improved over time, while the flood peak dischargeed and its occurring time were adjusted constantly. Forecasting results stablized when the rainfall tended to the end. The results of the coupling experiments displayed that this coupling forecast had an obvious superiority to a traditional forecast with a satisfactory accuracy and the effective extension of48hours in the flood frecasting period and it had a high forecasting accuracy.The researches of the above three parts showed that rainfall forecasting error of WRF model is a key factor deciding on the accuracy of flow forecast and the errors of the peak flows and peak time differences depend on the prediction deviation of rainfall regions and strong rainfall period.