Simulation of Urban Land Surface Temperature Based on Sub-Pixel Land Cover in A Coastal City

Xiaofeng Zhao,Lei Deng,Huihui Feng,Yanchuang Zhao

DOI: https://doi.org/10.1117/12.2069115

2014-01-01

Abstract:The sub-pixel urban land cover has been proved to have obvious correlations with land surface temperature (LST). Yet these relationships have seldom been used to simulate LST. In this study we provided a new approach of urban LST simulation based on sub-pixel land cover modeling. Landsat TM/ETM+ images of Xiamen city, China on both the January of 2002 and 2007 were used to acquire land cover and then extract the transformation rule using logistic regression. The transformation possibility was taken as its percent in the same pixel after normalization. And cellular automata were used to acquire simulated sub-pixel land cover on 2007 and 2017. On the other hand, the correlations between retrieved LST and sub-pixel land cover achieved by spectral mixture analysis in 2002 were examined and a regression model was built. Then the regression model was used on simulated 2007 land cover to model the LST of 2007. Finally the LST of 2017 was simulated for urban planning and management. The results showed that our method is useful in LST simulation. Although the simulation accuracy is not quite satisfactory, it provides an important idea and a good start in the modeling of urban LST.