A Study Of Land Surface Temperature Retrieval And Angular Effect Of Surface Brightness Temperature Based On Sentinel-3A/SLSTR Data

Land surface temperature(LST)is a significant parameter indicating the physical process of land surface.It is one of the indispensable driving factors in energy exchange between land surface and atmosphere,as well as a key parameter in various fundamental scientific research.LST is widely used in climate,hydrological,ecological and meteorological research,and can also be used for monitoring and forecasting of global climate change,forest fires,agricultural drought.Satellite remote sensing is an important method to quickly obtain temporal and spatial variation of LST at global and regional scales,with currently available satellite LST products such as MODIS,Landsat,ASTER,VIIRS.European Satellite Agency(ESA)launched Sentinel-3A satellite in 2016.The SLSTR onboard Sentinel-3A contains forward and nadir observing directions in two thermal infrared channels,providing the capability to analyze the angular effect of LST.Therefore,it is necessary to implement the researches on LST retrieval algorithms using SLSTR measurements.This research took SLSTR measurements as main data sources,adopted high-accuracy LST and emissivity estimating methods to estimate LST.This research also estimated land surface brightness temperature in two observing directions using split-window algorithm,compared the discrepancies of land surface brightness temperature in two observing directions.Main contents and conclusions of this research are:(1)To obtain land surface emissivity with higher accuracy,this research took ASTER GED emissivity product to estimate emissivity of bare soil,and used ASTER spectral library to estimate emissivity of vegetation.Then land surface emissivity can be estimated using NDVI threshold method.(2)An LST retrieval method based on split-window algorithm.The results of this method and SLSTR LST product were compared using simulated and in-situ measurements.The results indicated that the accuracy of this method is better than 1 K in terms of RMSE,though the accuracy may decline in situations with high atmospheric water vapor.The difference of the retrieval LST and SLSTR LST product is minimal over grasslands,and higher over barren surfaces.The accuracy of this method is 1.29 K and 0.95 K comparing with in-situ measurements in Dalad banner and Wuhai sites.(3)The factors that influences the differences of land surface brightness temperatures in two observing directions were analyzed over five land covers selected using IGBP classification system.The results show that the difference of land surface brightness temperature is large over evergreen needleleaf forests,croplands,grasslands and barren surfaces in daytime.The difference of land surface brightness temperature is stable in nighttime,with no significant season variation.The difference is minimal over evergreen broadleaf forest during both daytime and nighttime.In addition,when NDVI is greater than 0.7,the difference in surface brightness temperature between 0 ° and 55 ° is small.