Analysis of fine-grained urban temperature collected with a sensor network

Temperature in a metropolitan area exhibits a complicated tendency. Rather than geographical closeness, structures of a group of buildings and streets can affect changes in temperature. To identify the tendency of fine-grained distribution of temperature, we installed a densely-distributed sensor network called UScan. In this paper we describe a system of UScan and effective placement of sensors based on our experiment in downtown Tokyo. We also propose a clustering method to analyze the correlation between the tendency of temperature and the environmental factors.