Research Of Uncertainty Of SRM Model Parameter In The Upper Reaches Of Heihe River

Watershed hydrological modeling and forecasting is an important technology in understanding water resources. However, hydrological model almost all contains some parameters which cannot be directly observed, how to determine these parameters become is a very important factor in the hydrological modeling process. In the cold and arid inland river basin in Northwest Chinese, due to the lack of information, many model parameters need to be set artificially, the problems of model uncertainty is particularly prominent. Research on the uncertainty of hydrological model parameters can effectively identify the key parameters, optimization parameter calibration efficiency, reduce the uncertainty of model parameters, and provide a basis for model widely.SRM(snowmelt-runoff model) model is necessary for the watershed hydrological model of remote sensing data support for the few, data requirements are not high, but with the simulation and the prediction accuracy is very good. In this paper, we take the upper reaches of Heihe River as the study area, the establishment of the main 2008 snowmelt period(3-7 months) for the SRM model, to determine the parameters required by the model. On the basis of this, a sensitivity analysis of model parameters, determine the uncertainty interval model simulation, and the factors causing uncertainty of parameters is discussed, and obtained the following conclusions:(1) Surface precipitation China 0.5 * 0.5 degree grid data issued by the National Meteorological Information Center, was set reflecting the temporal and spatial distribution features of precipitation. To make up for the alpine valley spatial rainfall data are scarce, can provide spatial precipitation data necessary for snowmelt runoff model simulation.(2) The SRM model in the upper reaches of Heihe basin in 2008 mainly snow melt period(3-7months) model efficiency coefficient was 0.84, the volume difference of 3.8%, are in good agreement with the simulated hydrograph and the measured flow process line in the trend, can basically reflect the variation of runoff. But the simulated daily runoff is not accurate, especially the simulated error of the main peak flow is larger, there is a big uncertainty.(3) There are nonlinear effects and interaction between the parameters of SRM model, the simulation results of the model is not decided by one or several factors, but by a set of parameters of common effect.(4) In the beginning of the snow melt period, more sensitive factors associated with snowmelt, a greater impact on SRM model. With the reduction in snow cover and increased rainfall, more sensitive to rainfall runoff coefficient, the contribution rate of the larger models.(5) The SRM model uncertainty size related from of runoff supply, supply more complexity and uncertainty is greater. In the given parameter distribution and interval conditions, due to many factors, the measured flow cannot be completely contained within the 95% confidence lower flow, boundary, but the 81% was in the range, in the upper reaches of the Heihe river basin model still has a good application prospect.