Remote Sensing Estimation Of Land Surface Temperature In Mountainous Areas Of Southwest China

Land Surface Temperature(LST)is a crucial parameter in the study of Land Surface processes.It’s effective to estimate the radiation balance and energy budget at local and global scales using remotely sensed data.Currently,the fast development of LST retrieval algorithm based on thermal infrared remote sensing has made a series of progress.Its accuracy can be obtained in the uniform area of flat surface coverage.However,it is still a great challenge for their application over complex surface area.Because of cloud obstruction,there are many gaps in thermal infrared remote sensing LST products in southwest China.It is necessary to obtain spatial-temporal continuous LST products in complex topographic mountainous areas,which is an important guarantee for evaluating hydrologic,climate and ecosystem dynamic changes.This study analyzed LST over complex mountainous terrain in Southwest China.Firstly,the driving factors of LST dynamic change was analyzed in regional scale area.Secondly,considering various parameters,an improved LST reconstruction method for cloud-covered pixels was proposed.Based on this method,a spatial-temporal continuous LST dataset was produced combining the assimilation data in southwest China.This study is expected to provide a scientific basis for assessing the impact of global climate change and the formulation of adaptation strategies.The main conclusions of this study are as follows:(1)LST was simultaneously affected by topography and vegetation in complex mountainous areas.LST was consistent with the terrain in different seasons.LST decreased with increasing elevation and slope,and it was higher on the north slope than on the south.The mountain region had higher elevation,more homogeneous surface cover types,thicker vegetation coverage and lower LST.The trough valley area was flat,with relatively lower elevation,wider artificial surface distribution and higher LST.LST values decreased with the increasing elevation and slope,and it was higher on the northern slope than on the south.LST in mountain surface was lower than trough valley surface throughout the year,where elevation was higher,slope was steeper and it was mainly covered with woodland.Therefore,topographic and vegetation factors should be considered in the study of LST in complex mountainous areas.(2)The reconstruction method based on random forest regression can effectively filled the gaps in cloudy pixels over the complex terrain areas.Reconstruction of LST covered cloud was studied in Chongqing City as an example,which is a mountainous area.The accumulated solar radiation from sunrise to satellite overpass collected from the surface solar irradiance product of the Feng Yun-4A geostationary satellite was used to represent the impact of cloud cover on LST.The reconstructed method was developed by building a linking model for the moderate resolution imaging spectroradiometer(MODIS)LST with other surface variables using a fandom forest(RF)-based approach.LST assimilation data of China Meteorological Administration Land Data Assimilation System(CLDAS)was used as a supplement for images with severe missing values.The visual assessment indicated that the reconstructed gap-free LST images can sufficiently capture the LST spatial pattern associated with surface topography and land cover conditions.Additionally,the validation with in situ observations revealed that the reconstructed cloud-covered LSTs have similar performance as the LSTs on clear-sky days,with the correlation coefficients of 0.92 and 0.89,respectively.The unbiased root mean squared error was 2.63 K.In general,the validation work confirmed the good performance of this approach and its good potential for regional application.(3)Assimilation data and remote sensing data were fused to produce a complete spatial-temporal seamless LST product of Southwest China in 2019.The performance of the RF model depends on a sufficiently large data sample.However,there are few effective LST pixels for some days.Therefore,CLDAS LST assimilation data was used to fuse with MODIS LST dataset.Firstly,the RF method downscaled CLDAS LST with the reference of auxiliary data,such as vegetation index data,topographic data and cumulative radiation factor.Next,the system biases correction between MODIS LST and CLDAS LST are removed and product the special-temporal continuous LST datasets.The result showed that this method can effectively fill the MODIS LST gaps,and the LST products had a strong temporal-spatial continuity with ground data and other LST products.It could provide effective data support for various researches related with LST in southwest China.