| Land Surface Temperature(LST)and its annual or inter-annual changes play an important role in global climate change,urban heat islands,and land-atmosphere energy exchange.Since the rise of remote sensing technology,satellite sensors have accumulated a large amount of thermal infrared data and have been widely researched and applied.Moderate Resolution Imaging Spectroradiometer(MODIS)sensors can provide four instantaneous LSTs of the day across the world.However,many important studies such as climate change and hydrology require the input of daily mean LST rather than instantaneous LST.Moreover,the satellite's sampling cycle and cloud contamination have caused a serious lack of LST data and hinder the application of LST products.In response to these problems,this paper proposed a method to estimate the daily mean LST based on the instantaneous measurement at the MODIS observation time,explores the interannual variations law of LST,and revealed the global trend of interannual LST variations from 2000 to 2019.There are three primary research contents as follows:(1)Estimation of daily mean LST from remote sensing.Based on the ground measurements from 208 stations distributed around the world,a daily mean LST estimation model was constructed.An algorithm for estimating the daily mean LST using MODIS daily instantaneous LSTs was proposed.The ground validation results show that the accuracy of daily mean LST estimated by MODIS LST products is 2.22 K.The algorithm has the ability to accurately estimate the daily mean LST.(2)Comparison and development of the annual temperature cycle(ATC)model.The comparison results of the three-parameter and five-parameter based ATC models(ACP3 and ACP5models,respectively)show that the ACP5 model performs better than the ACP3 model.For most samples,the fitting accuracy of the two models is similar;for the samples near the equator,both models perform poorly with R~2 less than 0.5;ACP5 model performs much better than ACP3 model at the pixels in Antarctic and some mid-latitude pixels.Considering the continuity of the multi-year variation of daily mean LST,a multi-year variation model of daily mean LST(YYCD?ACP3 and YYCD?ACP5)was constructed based on the ACP3 and ACP5 models to simulate the inter-annual variation of the daily mean temperature.The YYCD?ACP3 model was evaluated using 8representative samples with MODIS daily mean LST from 2014 to 2018.The results show that the YYCD?ACP3 model can accurately simulate the inter-annual variation of daily mean LST.For samples in tropical and polar regions,the fitting accuracy of the YYCD?ACP5 model is significantly better than that of the YYCD?ACP3 model.(3)Analysis of global trends of interannual variation of daily mean LST from 2000 to 2019.Based on the MOD11C1 and MYD11C1 LST products from 2000 to 2019,the daily mean LST product and ACPs(i.e.annual mean LST,annual amplitude and annual phase)were calculated using the daily mean LST estimation method and the YYCD?ACP5 model.By calculating the Theil-sen slope of the global ACPs from 2000 to 2019,and using the Mann-Kendall test method for significance test,the global trends of annual mean LST,amplitude and phase in the past 20 years were obtained.The area where annual mean LST,amplitude and phase rise and fall significantly were located.The correlation analysis results between the annual mean LST and the annual mean NDVI,annual mean soil moisture,annual cumulative rainfall and annual mean aerosol optical thickness show that the annual mean LST has the greatest correlation with the annual mean NDVI,and has a smaller correlation with several other surface parameters.Overall,the estimation method of daily mean LST proposed in this paper can accurately estimate the daily mean LST.The continuous and derivable model of annual temperature cycle for several years can effectively simulate the interannual variation of daily mean LST,which is expected to play a huge potential in the inter-annual change of surface temperature and other practical applications. |
PDF Full Text Request