Application Of Uncertainty Method In Hydrological Forecast Of Small Watershed In Mountainous Area

Due to the many uncertainties in the hydrological process,there are many blind spots in the forecast operation.Uncertainty exists objectively in the hydrological process.Uncertainty issues are currently hot topics in hydrological forecasting.Studying the practical application of uncertainty methods in small basin hydrological forecasting can provide a wide range of application possibilities for strengthening basic hydrological theory and flood control.In this paper,the Yongcui River Basin is selected as the research object,and the uncertainty theory is used to forecast the monthly runoff sequence.The results obtained are as follows:(1)The seasonal first-order autoregressive model cannot give the ideal forecast accuracy at the sudden runoff point,individual monthly runoff during the special peak period,and monthly runoff during the transition period.However,the model has a simple structure and is convenient for modeling,and the process line given in the forecast can describe the level and periodic variation characteristics of the monthly runoff sequence,which can be used as a reference.(2)In the seasonal least squares multiple linear regression model,without considering the precipitation,the model fitting and forecasting effects are general,and the forecasting effects are stable in June during the wet season and December during the dry season.After considering the precipitation,the prediction accuracy of the model is greatly improved,the error qualification rate and the peak relative error are in line with the hydrological prediction error standard,and have certain adaptability to the study area.(3)The adaptive fuzzy neural inference system(ANFIS)has a high degree of fitting in the training process.During the inspection process,the relative error value of most months is within 20%,and some high-precision month errors reach 0.8%,and the forecast results have a good trend The correlation between the forecast value and the actual value is high.Except for individual months,the determination coefficient is relatively close to 1,which is suitable for forecasting the monthly runoff sequence in the Yongcui River Basin.(4)Establish a joint forecasting model.On the basis of ANFIS forecasting precipitation,a least squares multivariate linear regression model considering the current month's rainfall is established.The prediction accuracy of the model is closely related to the forecasting accuracy of precipitation.The accuracy of rainfall forecast meets the standard,and the runoff process obtained by the model is also satisfactory,which is suitable for the study area.Provides a new way for the wider application of forecasting models...