Study On Spring Snowmelt Runoff In Huanren Reservoir Basin

Snow water is a valuable water resource for Huanren Reservoir Basin in spring when water resource is scarce,and for Huanren Reservoir Basin it is of great importance in aspects of fully taking advantages of snow water resource,harzard prevention and mitigation that improving the accuracy of spring runoff forecasting in snowmelt season.For spring snowmelt runoff forecasting,the key steps are estimating parameters that have influence on snowmelt and choosing appropriate analytical methods.The traditional linear analytical methods based only on runoff ignore that snowmelt have influence on spring runoff,therefore the forecasting error is relatively large.In order to analyze which parameters have influence on snowmelt in snowmelt season extensively and improve the runoff forecasting precision in snowmelt season,this paper studies different factors' influences on spring runoff by interpreting the characteristics of varies hydrological factors,and analyses the start date of spring snowmelt runoff,and establishes the period,which is defined as the main snowmelt period,during which the snowmelt have impacts,and at last constructs high-precision snowmelt runoff forecasting models using linear and non-linear analytical methods.The primary achievements in the study are presented as follows:(1)By adopting basic analytical methods such as the Mann-Kendall test,analysis of hydrological elements' properties,like temperature,snowfall,runoff,precipitation in spring,etc.,by the time step of month and their impacts on spring runoff are made,and conclusions are achieved that time from December to February is defined as Stable Snow Accumulating Period;factors that have significant impacts on spring runoff are snowfall in winter,precipitation in spring,temperature in spring,etc.;snowmelt shows a weaker influence on runoff in May comparing to March and April.(2)Start date of spring snowmelt runoff are calculated by using Mann-Kendall test,and forecasting model of the start date of spring snowmelt runoff is constructed by using multiple linear regression,and analysis of main snowmelt period in spring are made by using partial correlation analytical method.Several results are reached that the start dates of snowmelt runoff mainly fall on the second ten-day in March,and the accuracy of the forecasting model is satisfying and the main snowmelt period is March and April.(3)For the forecast of March and April runoff in snowmelt period Linear Regression and Multiplayer Perceptron Neural Network are employed with and without consideration of snowmelt's impact,and 6 groups of factors for every ten-day in March and April are chose,based on which runoff forecast models for each ten-day are constructed.Conclusions are reached that both the goodness of fit and qualification ratio of Multiplayer Perceptron Neural Network model under the consideration of snowmelt's impact are the best,and its high accuracy indicates its excellent referential value for forecast and regulation.