Reconstruction of MODIS Land Surface Temperature in Northeast Qinghai-Xizang Plateau and Its Comparison with Air Temperature

Land surface temperature(LST) derived from remote sensing data is a good alternative because the traditional LST from meteorological station has a local problem.Yet remote sensing LST may suffer from noises from various sources,which greatly degrade the LST quality and hamper its efficient application.This paper presents a novel algorithm which can reconstruct the complete LST based on relationship of DEM and LST.TERRA MODIS-LSTs in northeast Qinghai-Xizang Plateau in 2008 are reconstructed using this method.Comparison of origin LST and reconstructed LST data with maximum air temperature(T) observations at 62 meteorological stations in the study region indicates that LST is significantly correlated with T,with correlation coefficient of 0.88 and 0.91,respectively,and average correlation coefficient is 0.898 at each station.By the average,LST has a 6.98 ℃ bias over T.The land-atmosphere temperature difference is related with the sea level elevation,vegetation cover,season and noise.The heighter sea level elevation is,the larger land-atmosphere temperature difference is;the denser vegetation coverage is,the smaller earth-atmosphere temperature difference is;the higher temperature is,the larger temperature difference is.It is also found that there are still some residual noises in well-vegetated areas after reconstruction.Therefore,the further study is required to identify the sources of noises,for instance,using the quality control information and 1 day LST.