138 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 4, NO. 1, MARCH 2011 Maximum Nighttime Urban Heat Island (UHI) Intensity Simulation by Integrating Remotely Sensed Data and Meteorological Observations Ji Zhou, Yunhao Chen, Jinfei Wang, and Wenfeng Zhan Abstract—Remote sensing of the urban heat island (UHI) effect has been conducted largely through simple correlation and regression between the UHI’s spatial variations and surface characteristics. Few studies have examined the surface UHI from a temporal perspective and related it with climatic and meteorological factors. By selecting the city of Beijing, China, as the study area, the purpose of this research was to evaluate the applicability and feasibility of the support vector machine (SVM) technique to model the daily maximum nighttime UHI intensity (MNUHII) based on integration of MODIS land products and meteorological observations. First, a Gaussian surface model was used to calculate the city’s MNUHIIs. Then, SVM regression models were developed to predict the MNUHII from the following variables: the normalized difference vegetation index (NDVI), surface albedo, atmospheric aerosol optical depth (AOD), relative humidity (RH), sunshine hour (SH), and precipitation (PREP). Results demonstrate that the accuracy of the SVM regression in predicting the MNUHII was around 0.8 C to 1.3 C; in addition, the SVM regression outperformed the multiple linear regression and the artiﬁcial neural network with backpropagation. A scenario analysis indicates that the relationships between the MNUHII and its inﬂuencing factors varied with time and season and were impacted by previous precipitation. The RH and AOD were the most important factors that inﬂuenced the MNUHII. In addition, previous precipitation could signiﬁcantly mitigate the MNUHII. The results suggest that future investigations on the surface UHI effect should consider the climatic and meteorological conditions in addition to the surface characteristics. Index Terms—Climatic and meteorological conditions, MODIS, support vector machine, surface characteristic, urban heat island. Manuscript received November 30, 2009; revised June 18, 2010 and August 14, 2010; accepted August 17, 2010. Date of publication September 20, 2010; date of current version March 23, 2011. This work was supported by the Na- tional Natural Science Foundation of China (Grant 40771136), the Fundamental Research Funds for the Central Universities of China, the Opening Funding of State Key Laboratory for Remote Sensing Science (Grant OFSLRSS201003), and an NSERC Discovery Grant. J. Zhou is with the Institute of Geo-Spatial Information Science and Tech- nology, University of Electronic Science and Technology of China, Chengdu 610054, China, and also with the State Key Laboratory of Earth Surface Pro- cesses and Resource Ecology, College of Resources Science and Technology, Beijing Normal University, Beijing 100875, China (e-mail: jzhou233@uestc. edu.cn). Y. Chen and W. Zhan are with the State Key Laboratory of Earth Sur- face Processes and Resource Ecology, College of Resources Science and Technology, Beijing Normal University, Beijing 100875, China (e-mail: cyh@ires.cn; zhanwenfeng@ires.cn). J.Wang is with the Department of Geography, University of Western Ontario, London, ON, N6A 5C2 Canada, and also with the State Key Laboratory of Earth Surface Processes and Resource Ecology, College of Resources Science and Technology, Beijing Normal University, Beijing 100875, China (e-mail: jfwang@uwo.ca). Digital Object Identiﬁer 10.1109/JSTARS.2010.2070871 I. INTRODUCTION R APID urbanization in cities across the world causes sig- niﬁcant land use and land cover changes on the Earth’s surface and has been recognized as one of the most important anthropogenic inﬂuences on the climate [1]. One of the most well-known adverse effects that is induced by urbanization is the so-called “urban heat island” (UHI) effect, which describes the phenomenon of temperatures in urban areas being higher than those in nearby rural areas [2]. Heat-waves that are caused by the UHI effect inﬂuence human health, the quality of life of urban residents, energy consumption, and other adverse climate effects, such as air pollution. Therefore, the UHI effect has been a concern for several decades. Satellite remote sensing is an excellent tool for examining the UHI effect. Remotely sensed images with medium spatial resolutions, e.g., the Landsat Thematic Mapper (TM) and the Terra Advanced Spaceborne Thermal Emission and Reﬂection Radiometer (ASTER) data, are extensively used to analyze the UHI’s spatial patterns because their thermal channels obtain higher spatial resolutions than other sensors. Many studies have focused on examining urban thermal patterns and their rela- tion to urban surface characteristics, such as vegetation abun- dance, impervious surface fraction, and land use and land cover [3]–[8]. On the other hand, sensors with high temporal reso- lutions, such as the Moderate-Resolution Imaging Spectrora- diometer (MODIS) on board both the Terra and Aqua satel- lites as well as the Advanced Very High Resolution Radiometer (AVHRR) on the NOAA satellites, provide the possibility to monitor UHI variations at different temporal scales. Researchers have found that the UHI obtains evident features at the diurnal, monthly, and seasonal scales [9]–[12]. Although the spatial variation of the UHI has been investi- gated in the literature, there are still some research questions that need to be addressed urgently. First, the inﬂuences of cli- matic and meteorological conditions as well as urban surface characteristics on the UHI effect need to be investigated. Al- though meteorologists have pointed out that the atmospheric UHIs are impacted by climatic and meteorological elements, [13], [14], to our knowledge, studies on the surface UHI with re- mote sensing are still rare. In addition, a greater understanding of the best models to describe the relationships between the UHI and its inﬂuencing factors needs to be investigated. Cur- rently, simple correlations and regressions, such as the linear regression, are commonly used to describe the relationships be- tween the UHI and its inﬂuencing factors [15]. In fact, the fac- 1939-1404/$26.00 © 2010 IEEE