Analysis of “furnace cities" in China using MODIS/LST product (MOD11A2)

In China, hot big cities are called “furnace” cities. With the development of urbanization and global warming, more and more big cities have become hot like a “furnace”. To compare the hot degree among cities, long time series of meteorological data can be used. However, due to a high patial heterogeneity of urban air temperature, point-based meteorological observation is unable to fully represent the whole city. In this paper, a new method based on satellite LST (Land Surface Temperature) product is proposed. First, MODIS/LST product (MOD11A2) and meteorological data during 2000-2010 for the 31 big cities are collected, and tree ensemble algorithm is used to build air temperature estimation models for each city. Then the model is applied to get spatially continuous air temperature and the urban regional mean temperature is calculated. Through analyzing the frequency of extreme hot days (regional mean temperature is greater than 35°C), we find out that the top 4 hottest cities are Changsha, Chongqing, Fuzhou, and Hangzhou. The list is slightly different from the one based on meteorological data. LST product quality, model accuracy, the length of data series and other factors have an impact on the result. The satellite-based method can represent the whole big city well and the spatial detail of hot degree within a city can also be characterized. More comprehensive criteria are needed to improve the new method.