Reconstruction And Spatiotemporal Analysis Of Land Surface Temperature Data In Huaihe River Basin

Land Surface Temperature(LST)is an important index to evaluate climate change.The study of temporal and spatial variation characteristics of LST and its relationship with geographical location,meteorology and underlying surface can further understand the interaction mechanism between earth and atmosphere,clarify the response law of climate change,and predict the future development trend of LST.Traditional observation method based on the meteorological site is very difficult to obtain high spatial resolution land surface temperature data,using remote sensing technology can obtain large area and spatial continuous distribution of land surface temperature data,because of the remote sensing image affected by aerosol,observation angle of the sun and clouds,make remote sensing inversion of surface temperature in space and time exist on the lack of different level,this hinders the accurate analysis of spatiotemporal characteristics of surface temperature.To solve the above problems,this paper takes MODIS remote sensing data as the main data source,adopting HANTS(Harmonric Analysis of Time Series),S-G(Savitzky-Golay),IMA(Interpolation of the Mean Anomalies)three methods to reconstruct the surface temperature data of MODIS MYD11A2 in the Huaihe River Basin(2003~2018)in 8days and compare the accuracy of the results.Based on the high-precision reconstruction method,the spatial-temporal variation of land surface temperature in the Huaihe River Basin in the past 16 years was analyzed.The influence of different geographical factors,climate factors and land cover types on land surface temperature was analyzed.The research contents and main conclusions of this paper are as follows:(1)Comparison of results of different reconstruction methodsSpatial distribution difference results show that IMA has the strongest denoising ability compared with HANTS and S-G,and the reconstructed surface temperature value range is the closest to the true value range.S-G has the worst denoising ability,and HANTS changes the original true value more,and the reconstructed surface temperature is higher.The correlation coefficient of pixels before and after IMA reconstruction was 0.90~0.98,which was higher than HANTS and S-G.The reconstructed surface temperature curve by IMA method is close to the measured surface temperature curve,and it is smooth,with RMSE(Root Mean Squard Error)of1.61? and MAE(Mean Absolute Error)of 1.17? are smaller than Hants and S-G;The correlation coefficient with the measured surface temperature is 0.956,which is higher than that of Hants and S-G.The reconstructed curves of HANTS and S-G have obvious fluctuations,and the reconstructed surface temperature of HANTS at high values is higher than the measured value,and the reconstructed error of S-G in winter is larger.(2)Spatiotemporal variation characteristics of land surface temperature were analyzed based on IMA reconstruction methodThe interannual and monthly mean temperature showed an overall warming trend,and the mean warming rate in August was the largest,reaching 0.21?/a.The temperature increase was most obvious in summer,and the temperature increase rate reached 0.25?/a in daytime and 0.09?/a at night.The temperature difference between day and night is the largest in autumn,with a maximum of 16.89?.The regional high temperature in January was concentrated in the south of Huaihe River and Taizhou and Yangzhou to the north of Yangtze River in Jiangsu Province,while the low temperature was mainly in the vicinity of Yimeng Mountain Area.The regional high temperature was distributed in the upper and middle reaches of the Huaihe River in spring,mainly concentrated in the middle reaches of the Huaihe River in summer,had no obvious characteristics in autumn,and mainly concentrated in the south of the upper reaches of the Huaihe River in winter.Interannual regional low temperatures were mainly distributed in Yi Mu Si River Basin,Yimeng Mountain,Funiu Mountain and Dabie Mountain with higher altitudes,while regional high temperatures were mainly distributed in the upper reaches of Huaihe River,Yangzhou and some areas of Taizhou far from the coastline.(3)Correlation analysis of influencing factors of surface temperatureThe correlation between LST and latitude is strongest in winter.The results show that the LST decreases by 0.84? for every 1° increase in latitude.The seasonal mean temperature was negatively correlated with the altitude,and the negative effect was most obvious in spring.The results showed that the land surface temperature decreased 0.71? for every 100 m elevation increase.The correlation between surface temperature and air temperature is obvious,and the correlation coefficient in spring and winter is above 0.9.In summer,the negative effect between LST and precipitation is the most obvious.The results show that the LST decreases by 1.91? for every1 mm increase in precipitation.The difference of land surface temperature is obvious under different land use types.The results show that the construction land > cultivated land > unused land > water area > grassland > woodland,in which the construction land is mainly distributed in the high temperature area.When the construction land increases,the surface temperature increases significantly.