Downscaling And Spatial-temporal Variation Of Land Surface Temperature In The Greater And Lesser Khingan Mountains During 2010-2021

In the context of global warming,the Greater and Lesser Khingan Mountains,as the main distribution area of permafrost in northeast China,are more vulnerable to the effects of climate change.As a measurement index of climate change,surface temperature affects climate change by directly acting on the underlying surface reflectance,moisture content and heat in soil.Therefore,accurate inversion of surface temperature can provide a reference for the research of permafrost degradation and freeze-thaw change.Currently,remote sensing images cannot obtain land surface temperature data products with both high time resolution and high spatial resolution.Therefore,downscaling land surface temperature data with high time resolution and low spatial resolution to improve data accuracy is a conventional method in current research.Therefore,in this paper,MODIS land surface temperature data is reconstructed by Hants algorithm,and random forest downscaling model is adopted to improve the accuracy of land surface temperature data.On the basis of high-precision data,it is of great significance to analyze the spatio-temporal variation of land surface temperature in the Greater and Lesser Khingan Mountains from 2010 to 2021 and to study the influencing factors of the spatial distribution of land surface temperature,so as to further understand the evolution law and formation mechanism of land surface temperature in the permafrost region,analyze the problem of permafrost degradation,study climate change in the permafrost region,and monitor disasters.The main conclusions are as follows:(1)The surface temperature data was reconstructed by Hants software.MAE and RMSE of reconstructed results were 0.91 and 1.37,respectively,and the degree of fitting R~2was 0.81,indicating a good result of data reconstruction.The random forest model was used to downscale reconstructed land surface temperature data,and the spatial resolution was downscaled from 1000 m to 250 m.MAE and RMSE of the downscaling results and actual land surface temperature data were 0.83 and 1.26,respectively,and the degree of fitting R~2was 0.77.The downscaling images can more accurately describe the spatial differences of land surface temperature and present the details of the spatial distribution of land surface temperature.(2)In terms of time variation,the average annual surface temperature and monthly surface temperature in the Greater and Lesser Khingan Mountains during 2010-2021showed an upward trend,and the average annual surface temperature increased by2.46℃on the whole.In addition,the monthly mean surface temperature in January,February,March,April,July,November and December in the Greater and Lesser Khingan Mountains from 2010 to 2021 showed an obvious upward trend.In terms of spatial changes,the annual average surface temperature in the Greater and Lesser Khingan Mountains from 2010 to 2021 presents obvious regional characteristics.The Greater and Lesser Khingan Mountains are located in high latitude and high altitude areas with large areas of forests,and the surface temperature is relatively low,while the Songnen Plain and Hulun Buir Grassland are relatively high due to the influence of vegetation cover and bare land.Trend analysis method was used to analyze the average annual land surface temperature data of the Greater and Lesser Khingan Mountains from 2010 to 2021,and it was concluded that the average annual land surface temperature of the whole Greater and Lesser Khingan Mountains showed a rising trend,and the area area of the average annual land surface temperature rise reached 88.5%.From the spatial distribution,it can be seen that the average annual land surface temperature in Hulun Buir Grassland and some parts of the Greater Khingan Mountains increased significantly,and the area of the average annual land surface temperature increased significantly reached 9.3%.(3)The spatial distribution of land surface temperature is affected by latitude,DEM,elevation,NDVI and surface reflectance,but the influence of these five variables on land surface temperature varies.Five parameters were used as independent variables to establish a multiple linear regression model.The analysis results show that DEM,latitude and NDVI are the factors that have a great influence on the spatial distribution of land surface temperature in spring and summer.Influenced by monsoon,Xingan Mountains in summer has a hot and rainy climate,which can meet the demand of vegetation growth.High vegetation coverage in summer has the greatest influence on the spatial distribution of land surface temperature.DEM and latitude are the most important factors affecting the spatial distribution of land surface temperature.In autumn and winter,the factors that have the greatest influence on the surface temperature in the greater and lesser Hinggan Mountains are latitude and surface reflectance.Affected by the cold northwest wind from high latitudes,the surface temperature decreases rapidly,and the solar altitude Angle in high latitudes is small,which affects the surface’s absorption of solar radiation.Meanwhile,snow cover also affects the surface reflectance,resulting in the decrease of surface temperature.Therefore,latitude and surface reflectance have great influence on the spatial distribution of land surface temperature in autumn and winter.