Evaluation On Snow Cover And Snow Depth Simulated By VIC-CAS Model Based On Multisource Remote Sensing Observations In The Upper Reach Of Shule River Basin

Cryosphere is an important part of global climate system.In the context of climate change,cryosphere change has a great impact on hydrological processes and water resources,especially in water shortage areas.Snow is the most widely distributed element in the cryosphere,which is an indicator of climate change.Meanwhile,snowmelt runoff is also an important component of surface water resources.The influence of snow cover change on hydrological processes is one of the hot issues in climate change research.At the watershed scale,glaciers and permafrost are inlaid each other in space,and snow is superimposed on them.Therefore,in the typical cryosphere basin which has a good research foundation for precipitation,glaciers and frozen soil,it is a useful means to analyze the hydrological effect of snow elements and to strengthen the simulation of snow elements in the slope and watershed scale.Therefore,this study selected the upper reaches of Shule River as the study area,which has a good observation basis,more than 10 years of observation data and a good simulation of glacier runoff.Based on MODIS,500 meters snow cover area data,and snow depth data retrieved from long-range remote sensing in China,the snow cover and depth simulated by VIC-CAS model,which has been used to simulate the glacier melt water well in the study area,are compared with the snow cover and depth observed by remote sensing at the annual scale,monthly scale and different elevation zones,so as to provide reference for improving the snow cover process in VIC-CAS model.The main conclusions are as follows:(1)On the whole,the accuracy of VIC-CAS model was lower.Especially for thin snow and low altitude areas,the simulation effect was relatively poor,which may be related to the insufficient consideration of the influence of terrain on snowfall and snow redistribution caused by wind-blown snow in the model.It is necessary to strengthen the observation and simulation of snowfall and wind-blown snow.(2)By comparing the annual snow cover distribution simulated by VIC-CAS and the annual snow cover distribution calculated by 500 m product,it was found that VIC-CAS model can better simulate the snow cover in a more snow year,which indicated that there is great uncertainty in the simulation of thin snow by VIC-CAS model.Specifically,VIC-CAS model was underestimated when compared with daily snow cover data,and partially overestimated when compared with 500 m product.Cloud removal algorithm may be the main reason for this difference.From the perspective of spatial distribution,the spatial distribution of snow cover in multi snow years simulated by VIC-CAS was more consistent with the topographic characteristics of the study area,which was closer to the spatial distribution of daily snow cover products in a more snow year.From the perspective of monthly scale,VIC-CAS model can better simulate the snow cover in May,but worse simulate the snow cover in December.The main reason was that the snowdrift in December had a great impact on snow redistribution.From the perspective of spatial distribution,the relationship between snow cover degree simulated by the model and elevation distribution in the study area was strong,and the snow cover and elevation change of daily snow cover products was more obvious than that of 500 m products.From different altitudes,the accuracy of snow cover was the highest in 4000?5000m,followed by 3000?4000m,and the lowest in 2000?3000m,which indicated that the model had great uncertainty in 2000?3000m simulation algorithm at low altitudes.(3)By comparing VIC-CAS and snow depth data,it was found that the consistency of snow depth simulated by VIC-CAS was high between snow years and snow depth data.From the perspective of spatial distribution,the correlation between snow depth and terrain was stronger,while the relationship between snow depth and elevation was weaker due to its coarse spatial resolution.From the monthly scale,the consistency of snow depth was good in November,and the simulated value was underestimated,while the consistency was low in September,and the simulated value was obviously overestimated.From the perspective of spatial distribution,the snow depth simulated by the model was more consistent with the characteristics of elevation decreasing from southeast to northwest in the study area.From different altitudes,the consistency of snow depth simulated by VIC-CAS model and observed by products at different altitudes had no obvious difference,which were characterized by high consistency in more snow years and low consistency in less snow years and normal years.