Research On Flash Flood And Prediction In Medium-or Small-Scale Basin Of South Guizhou

Qiannan has abundant rainfall and numerous mountainous areas.Rivers in small and medium-sized watersheds are widely distributed in the region,which is a high incidence area for flash floods.Mountain torrent disaster threatens the safety of local people’s live and property,so it is urgent to strengthen mountain torrent forecasting in Qiannan region to improve the accuracy of regional early warning and forecasting and the efficiency of flood disaster emergency response.Based on the existing work and data of flash flood early warning and forecasting technology in small and medium-sized watersheds in southern Guizhou,this paper studies flash flood early warning and forecasting technology in small and medium-sized watersheds from two aspects of model forecasting and flood season early warning.The main results are as follows:(1)Mountain torrent forecasting model is the most widely used tool for mountain torrent forecasting,but it is difficult to build mountain torrent forecasting model for areas with lack of data or relatively lack of data.In order to solve this problem,based on the theory of variable velocity geomorphic unit hydrograph(R-V-GIUH),a flash flood prediction model of velocity geomorphic unit hydrograph(R-V-GIUH)for small watersheds in southern Guizhou is established.Nineteen historical floods in the study area were selected for calibration and nine were verified.Runoff depth,flood peak,peak occurrence time and flood process were taken as accuracy indexes for evaluation.The results show that the R-V-GIUH flash flood forecast model can well reflect the influence of rainfall process on runoff process in the study area under the condition of less historical data.Even for multi-peak flood simulation,the simulated curve fits well with the measured flood series.This shows that the model has wide applicability and can be used as a forecasting method to carry out hydrological forecasting for small watersheds lacking hydrological data.(2)Because the relevant parameters of some subprocesses of the hydrological model cannot be directly obtained by monitoring methods,they can only be calibrated by means of measured data.The problem of "different parameters and the same effect" caused by calibration is an important source of uncertainty in hydrological models.In this paper,the deterministic flash flood forecast is realized,and the probability flash flood forecast model is also developed based on Bayesian theory.According to the distribution range,distribution probability and influence on flood forecasting results of parameter groups,the model can list the value intervals of parameter groups through,and obtain the probability and confidence intervals of mountain torrent forecasting,thus increasing the safety margin of mountain torrent forecasting and perfecting the conventional mountain torrent forecasting method.(3)In addition to the model forecast,flash flood early warning according to rainfall can effectively improve the response efficiency of flood control emergency.Based on the information of soil water content,this paper constructs the critical rainfall index and the composite rainfall index integrating rain intensity,early influence rainfall and this accumulated rainfall,which has the characteristics of simple use,low dependence on data and platform,and can realize rapid flash flood early warning.Especially for the composite rainfall index,the "early influence rainfall" is used instead of the "initial soil water content" in the dynamic critical rainfall.The problem that the dynamic critical rainfall index method needs both soil water content information and rainfall information to give early warning is transformed into the problem that only rainfall information is needed to give early warning in the composite rainfall index method,and the composite rainfall index can be dynamically calculated in real time as the basis for judging whether flash flood warning needs to be issued.