Spatial And Temporal Distribution Of Snow Cover And Snowmelt Runoff Simulation In Xilin River Basin

The Xilin River Basin is a typical seasonal snow-covered area in China with abundant snowfall resources.Spring snowmelt water is an important source of water in this basin.However,the occurrence of disasters such as snowstorms caused by snow cover and snowmelt floods will bring serious consequences to the life and farming in this area.Therefore,it is of great practical significance to know the temporal and spatial distribution characteristics of snow and its variation and to simulate the changes of snowmelt runoff to reasonably exploit the regional water resources and disaster prevention.Based on the remote sensing data and meteorological data,this paper studied the trend changes,abrupt changes and periodic changes of the main snowpacking elements in the basin,and clarified the spatial and temporal distribution of snow cover in the Xilin River Basin.Using the path analysis And GAM were used to study the response characteristics of snow elements to meteorological factors.The linear multiple regression method was used to analyze the factors influencing runoff during the snowmelt period.The PSO-LSSVM model was used to simulate the snowmelt runoff during 2001-2016.PSO-LSSVM and multivariate linear regression model for snowmelt runoff simulation accuracy.The main results are as follows:(1)During the year,the area of snow cover and snow water equivalent in Xilin River Basin showed a single peak,and the overall change tended to increase during the recent 16 years.From the perspective of spatial distribution,with the snowfall,the snow gradually increases from south to north.The average snow depth in recent 35 years and snow days showed an upward trend in time series.Among them,the rate of increase of snow depth was 0.35 cm / 10 a,the depth of snow suddenly changed in 1984 and the first main period appeared in 21a;the rising rate of snow days was 7.35 d / 10 a,In 2009,there was a sudden change,but there was no obvious change in annual snow days.The first main period appeared in about 20 a.(2)Stepwise regression shows that the main influencing factors of snow depth during snow cover are snow equivalent,sunshine duration,effective accumulated temperature>0 and average temperature.The dominant factors affecting the number of ?snow days and snow area are the average temperature,>0°C effective accumulated temperature and relative humidity.Path analysis shows that the main controlling factors for the snow element are>0 ?effective accumulated temperature and average temperature.The results of GAM fitting show that the average temperature and snow water equivalent are the main factors affecting the snow depth.The average temperature,relative humidity,effective accumulated temperature and snow equivalent are the main factors affecting snow days.Relative humidity,effective accumulated temperature and snow water equivalent are the main factors affecting snow area.(3)During the snowmelt period,there was a significant correlation between runoff and snow depth,snow cover area,snow water equivalent,temperature and relative humidity(p <0.01),and positively correlated with precipitation(p <0.05).As can be seen from the regression equation,the most influential factors on runoff are the average temperature,relative humidity,precipitation and snow cover area.The PSO-LSSVM model was used to predict the snowmelt runoff from 2001 to 2016.The runoff decision coefficients in March and April were 0.80 and 0.86,respectively.The simulation accuracy was improved compared with the multiple linear regression model.It shows that the PSO-LSSVM snowmelt runoff estimation model can better express the complex relationship between meteorological factors and snow cover and corresponding runoff during snowmelt,and is more suitable for predicting snowmelt runoff in the Xilin River Basin.