The climatic response of cities to the heat island effect is thought to vary by city size and location. In this paper, we analyzed weather data for representative Japanese cities from 1961 onwards and examined characteristics of time series meteorological data statistics such as annual number of tropical nights. The number of winter days decreased, and the number of summer nights, tropical nights and extremely hot days increased in all cities. As a result of principal component analysis, the rise of temperature was identified as the first principal component, and the decline of the daily range of temperatures was identified as the second principal component. A cluster analysis using the first four principal components categorized the 16 Japanese cities into four sets. Factors related to the results of the principal component are discussed with the aid of multiple regression analysis. Numerical results show that the magnitude of the first principal component is described by latitude and distance from the sea, the magnitude of the second principal component is described by the logarithm of the DID (Densely Inhabited District) population and distance from the sea.
