Correlation Analysis Between Land Surface Temperature And Remote Sensing Index Based On RS In Daxinganling Burned Area

In this paper,the Jinhe-Genhe forest fire in 2003 in Daxing'anling,Inner Mongolia was used as the research area.Using multi-source remote sensing data,combined with the measured surface temperature and DEM data,Analyzed the extraction method of the burned area and the optimal surface temperatur.Established an exploratory model to explore the optimal interpretation factor affecting the surface temperature,and compares the two influence models of the surface temperatures.The main conclusions are as follows:(1)The random forest classifier has a great advantage in accuracy and operability.Using the mainstream 6 algorithms applied to the LANDSAT 8 data band 10,band 11 inversion of the surface temperature.By comparison,the accuracy of the two-band inversion under different algorithms is ranked from high to low as:ISC10>ISC11>MW10>SWA>IMW10>IMW11>RTE10>JSW>RTE11.(2)The surface temperature before the forest fire occurred relatively low.The surface temperature of the fire occurred year reached the highest value of 320.26k,and then decreased year by year in the next three years.In 2006,it reached the lowest value of 312.56K,but in 2007 The change in 2010 was a shock up,and in 2010 it reached the highest value of 318.94K after the fire.Combined with the long-term sequence NDVI,it can be found that the vegetation distribution and the surface temperature has a significant correlation annually.(3)The change rate of surface temperature with altitude is slower than that of air with altitude.The rate of surface change with altitude changes in different seasons is different.The surface temperature changes in winter and summer have special patterns.The surface temperature changes of different slopes are relatively regular.On different slopes,the surface temperature from the east slope to the northeast slope showed a trend of increasing first,then decreasing,and increasing.(4)The DNBR,NDISI,VNDWI,NDVI,SGI,and MNDWI indices have better interpretation effects on surface temperature.DNBR,NDVI,NDISI and surface temperature showed a positive correlation,and VNDWI and MNDWI showed a negative correlation with surface temperature.In the interpretation of the surface temperature model,the local regression GWR model has more ideal regression analysis results than the global regression OLS model.The GWR model established by the five remote sensing indices can be found that MNDWI has better interpretability.(5)By comparing the image values and measured values of the UAV remote sensing index in the study area,it is found that the NGRDI inversion vegetation coverage fits well with the actual value.The downscaling model established by NGRDI vegetation coverage and NDVI vegetation coverage is y.=1.27x3-1.63x2+0.95x+0.6,R2 reaches 0.923.