Retrieval Of Land Surface Temperature With High Temporal Resolution In The Source Regions Of Rivers In Southwest China And System Construction

Land surface temperature(LST)plays an important role in the process of surface-atmosphere energy flux exchange and it is widely applied in many fields,such as it plays a key input parameter in the estimation of surface evapotranspiration and serves as the crux of weather forecasting.Satellite thermal infrared(TIR)remote sensing provides a direct way to derive the LST through observing the thermal radiation emitted from the ground surface in large regions.Although the applicability of the TIR remote sensing is limited in cloudy conditions and topography conditions,it can provide LST with better resolution and accuracy than the passive microwave remote sensing.The environmental characteristics of our study area(the source regions of rivers in southwest china-SRRSC)are mainly reflected in cloudy weather and great topographic relief,in addition,heterogeneous land surfaces and scarcity of meteorological stations provide clear challenges to overcome in actual LST retrieval process.The climate and topographic conditions of the study area need to be considered carefully,since we can take these as bases for adopting appropriate methods to eliminate the effect of environmental conditions and choosing relatively suitable remote sensing data and input parameters to retrieve accuracy LST relatively.In this paper,high temporal resolution remote sensing data(the disc image scanning is set at 30 minutes per cycle)from FY-2C/D/E is used for overcoming the cloudy weather of study area.Otherwise,The climate and topographic conditions of the study area give a wealth of environmental features and information to analyze LST retrieval result.The main purpose of this study is generating a long term and high temporal resolution LST product for the study area.Nowadays,a variety of satellites can provide a long term remote sensing data for using.But for the target of high temporal resolution,the geostationary satellite has incomparable advantages.In this paper,the second group satisfied our project time requirements including FY-2C,FY-2D,FY-2E provides TIR image data for LST retrieval.In recent decades,many algorithms and methods have been developed to retrieve LST from satellite TIR remote sensing data.Among these algorithms,17split-window(SW)algorithms have been adopted to retrieving LST in this paper,adapted to the advantage of S-VISSR sensor which has two adjacent TIR bands.Different SW algorithms have different scopes of application,using simulate data to evaluate LST retrieval algorithms' qualities is wildly used under the scarcity of in situ measurement.Based on the evaluation results,a part of unsatisfied algorithms are removed from the next LST retrieval section.Finally,the validation of high temporal resolution LST product is processed by single point method using in situ measurement data,and in situ measurement LSTs are calculated from four component data.When long term and high temporal resolution remote sensing data needs to be processed,time-consuming and storage issues are becoming significant challenges to overcome in practice.Under this circumstance,the parallel cluster computing and distributed file system demonstrated great potential to process big data.This paper generates a long term and high temporal resolution LST product in the source regions of rivers in southwest china,and uses the advantages of parallel cluster computing and distributed file system to solve big data problem in remote sensing.In order to simplify the data processing flow and achieve visualization target,a LST retrieval system is designed and realized.