Enhanced Modeling Of Annual Temperature Cycles And Urban Heat Island Monitoring With Temporally Discrete Remotely Sensed Thermal Observations

Satellite thermal remote sensing provides land surface temperatures(LST)over extensive areas that are vital in various applications,but this technique suffers from its sampling style and impenetrability to clouds,which frequently generates data gaps.Annual temperature cycle(ATC)models can estimate continuous daily LST dynamics from a number of thermal observations and fill these gaps.These models can not only be used to generate spatio-temporally seamless LST products,but also can be the cornerstone of LST temporal upscaling methods which are conducive to further improve the accuracy of urban heat island(UHI)remote sensing monitoring.However,the standard ATC model(termed ATCS)remains incapable of quantifying the short-term LST variations caused by synoptic conditions,leading to uncertainty in the results of UHI monitoring.Focusing on these problems,this paper designs a new ATC model that can characterize short-time LST changes,and further explores the impacts of different temporal upscaling methods on UHI monitoring.First,this paper proposed an enhanced ATC model(ATCE).By incorporating in-situ surface air temperatures and satellite-derived normalized difference vegetation index based on the ATCS,the ATCE described the daily LST fluctuations.With Aqua/Moderate-resolution Imaging Spectrometer(MODIS)daily LST products as validation data,we implemented and tested the ATCE over the Yangtze River Delta region of China.The results demonstrate that,when compared with the ATCS,the overall root mean square errors of the ATCE decrease for the day and night.The accuracy improvements vary with land cover types with greater improvements over the forest,grassland,and built-up areas than over cropland and wetland.The assessments at different time scales further confirm that LST fluctuations can be better described by the ATCE.Second,based on the ATC models,this paper explored the impacts of different temporal upscaling methods on the calculations of surface UHI intensity(SUHII).A systematic comparison of three temporal upscaling methods,such as the simple moving window(SMW)average method,the ATCS-and the ATCE-based methods,shows that great differences exist between the temporal upscaling data produced by the three methods.The SUHII is significantly affected by the proportion of clear-sky pixels.The intensities calculated by the three methods reach the similar level as the proportions increase.In addition,the SUHII calculated by the SMW average method show larger variation along with different time scales compared with the ATC-based ones.Through quantitative analysis of the effect from temporal upscaling strategies on the SUHII calculations,it can be conclude that,when the number of satellite observations is large,the SMW average method should be preferred.In contrast,the ATCE model method is a priority.In conclusion,the enhanced ATC model proposed in this paper can simulate the daily LST variations more accurately over the year,and provide better LST products for applications such as urban thermal environment.It can potentially play a greater role in more practical applications.