Research Of Conceptual Hydrological Models For Unguaged Catchments

A lack of observation data of watershed hydrological model research is a hot topic and difficult problem in today’s world of hydrological research.Based on the statistical regression method to find relation between the satellite remote sensing rainfall data and the the measured rainfall data, and analyzing the characteristics of the respective data, and then by a lack of observation data of remote sensing data show a "real" data.Using inversion data to construct a lack of observation data of watershed hydrological model, and the choice of hydrologic model parameters are used to it.Combined with bayesian theory and is suitable for the highly nonlinear complex model of MCMC algorithm, improve the efficiency of parameter calibration.Through a series of research to extend the PUB(Prediction in Ungauged Basins) problem field of study, and research direction is given.Prediction process of watershed hydrological models exist all kinds of uncertainty,among them, the model parameter uncertainty is a very important source of uncertainty of hydrological model, this paper through the GLUE method to study the CMD yi chun city wan load river valley model parameter uncertainty, find out the sensitive parameters, and give 90% coverage of the scatter distribution parameters in pairs, 90% confidence interval of the runoff value and the most advantage of parameter space in pairs.This paper’s main work includes:(1)Compared to single station precipitation data and stand average rainfall data and the correlation between remote sensing rainfall data in this area, and found many station precipitation data and the correlation of the area remote sensing remote sensing data better.(2)And measured rainfall data and remote sensing rainfall data together with the return of the remote sensing found respectively for hydrological forecasting rainfall data, remote sensing rainfall data contains information of torrential rain is real, the result of the flood forecast is more accurate.( 3) Compared with small and big basin found directly forecast results of remote sensing data, we found that remote sensing data is more suitable for large watershed hydrological forecast.(4)Compared to general global optimal method and is suitable for highly nonlinear DREAM algorithm, we found the rate of DREAM algorithm is more efficient, better result of simulation.(5)Through a variety of runoff confluence module combination, we found that CMD +armax model module is more suitable for jiangxi yichun area.