Time Series Modeling And Analysis Of Remotely Sensed Land Surface Temperature Over The Tibetan Plateau

Land surface temperature(LST)is an important parameter of surface physical processes at global and regional scales.The variation of LST is a response to the changes of climate factors and surface characteristics.In recent years,the warming of high-altitude areas has aroused much concern.With an average altitude of over 4,000 m,the Tibet Plateau(TP)is known as the world's "third pole",plays an important role in global climate and atmospheric circulation.Deeper understanding of the temporal and spatial variations of LST is important for research in a range of fields including climate,vegetation and hydrology.Although there have been numerous studies on analysis of spatiotemporal variation of LST over TP based on remote sensing data,a prominent drawback of these studies is that only clear-sky LST based on optical/thermal remote sensing data,which is blocked under cloudy conditions,is utilized.As a consequence of the bias between the time series of clear-sky LST and all-weather LST,the inference and conclusions of these studies can be unreasonable or have considerable deviations with the actual conditions.To the best of our knowledge,study on analysis of LST variations over TP based on all-weather LST is very rare.Therefore,based on the all-weather LST data set,this paper analyzes the spatial and temporal variation of LST on the TP,and discusses the relationship between LST and related factors over TP.The main work of this paper is as follows:(1)Near surface air temperature(SAT)has a strong correlation with LST and is an important factor in the study of LST changes.However,the meteorological stations on TP are sparse,and it is difficult for the station data to represent the regional SAT.The resolution of the reanalysis data is relatively coarse,and it is difficult to meet the analysis of this study.Therefore,before the spatio-temporal analysis of the LST,the SAT data with the same spatio-temporal resolution as the LST is first estimated using machine learning methods to provide support for subsequent analysis.(2)The latest international seasonal adjustment method X-13-ARIMA-SEATS is applied to the time series modeling of LST for the first time,and the differences in the analysis results are discussed from the perspective of the missing data.Combined with the linear tendency estimation and empirical orthogonal decomposition method,the temporal variation and spatial distribution characteristics of the all-weather LST over the Tibetan Plateau were analyzed from three different time scales: year,season and month.(3)Analyzed the all-weather LST and related factors of the TP.The study found that the trend of LST decreased with the increase of altitude,which was different in different altitude segments.The spatial distribution of LST is similar to that of SAT,but the variation trend is different.LST is correlated with precipitation,wind speed and sunshine hours.In this study,the all-weather SAT data set is produced of TP.The variation trend of the LST of the TP is analyzed at different scales,and the relationship between the LST of the TP and different factors is discussed.This is of great significance for understanding the climate change and environmental planning of the TP.