Study On Probabilistic Flood Forecasting Of Huanren Reservoir Basin Using TIGGE Data

The flood forecast can provide the information of the flood,which plays an important role in flood control and disaster relief,the exploitation of water resources.Due to the uncertainty of the hydrologic model structure and the input information,the flood forecast usually is uncertain.The historical flood forecast can only provide the single and certain results,which cannot reflect these uncertainty.The development of ensemble precipitation forecast makes the ensemble flood forecast become a new research,and makes it possible for the probabilistic flood forecast.Take the Huanren reservoir basin as study area,In this paper,firstly,the availability and precision of the precipitation form the ECMWF?UKMO?CMA?NCEP and CMC in TIGGE are evaluated.Then,the BMA model is used to obtain the probabilistic precipitation forecasting.Finally,the Probabilistic flood forecasting is obtained by using the probabilistic precipitation forecasting and the Xin'anjiang model.The result are list below.(1)The ensemble precipitation forecast from ECMWF,UKMO,CMA,NCEP and CMC are evaluated using CRPS,BS,Talagrand,TS,missing alarm rate and false alarm rate,the results show that with the increase of the lead time,CRPS and BS of the five centers increase,and TS of the precipitation at the same grade decrease,which indicates that the five centers perform worse as the lead time increase.At the same lead time,CRPS,BS,TS of ECMWF perform better than the other four centers,which indicate that ECMWF has a better forecast skill.(2)Xin'anjiang model is used as the flood forecast model in the paper.The period1970-2006 is used as calibrate period for calibrate the parameters and the period 2007-2011 is used as validation period for validate the model.The results show that,the CC and NSE in the calibration period is 0.90 and 0.88,respectively,and the CC and NSE in the validation period is 0.81 and 0.77,respectively.The performance in the two periods are both good,which indicates that Xin'anjiang model is suitable in the basin.(3)With the July 21 to September 30 in 2007 to 2011 as the validation period,the BMA model is evaluated in the case the calibration period is defined as 15 days,20 days,25 days,35 days,40 days,45 days and 50 days.The results show that,the BMA model of the five centers perform better that the original ensemble forecast,and the best calibration period is 35 days,the BMA model of ECMWF performs best.(4)In order to forecast the floods,the ensemble precipitation forecast and the BMA model results of ECMWF are both used as the input data of the Xin'anjiang model.The results show that the observed flood can be concluded in the range formed by the 25% and80% probabilistic precipitation of the BMA model,and the BMA model can provide more information,but the results dependents on the original forecasts.