Abstract
The growing demand for better understanding of urban microclimate change has led to the need for new models to explain relationships between urban air temperature and various urban environmental factors. In this paper, statistical models are presented that assess the impacts of various urban characteristics on air temperature observed at Automatic Weather Stations (AWS) in Seoul during 1998 and 1999. Model specification and estimations arc based on an explicit and new spatial framework under the assumption that the temperature of an AWS is affected by its surrounding urban factors, namely, open space area, transportation facilities area, building floor area, total road length, impermeable pavement area, and number of car registration. After spatially connecting those factors with AWS data using GIS techniques simple linear regression models for each explanatory variable are estimated for August, 1999, showing that all variables are statistically very significant. As a pilot study, multiple linear regression models are estimated with impermeable pavement area, transportation facilities area and open space area variable, explaining more than 79 percent of the variability in air temperatures. Then, the models are extensively used to check if the model specification of the pilot study could be adopted for the temperatures in other time periods (summer 1999; year 1999; August, 1998; summer, 1998; year 1998). The results show that the estimated models explain more than 74 percent of the variations in the temperatures, confirming its appropriateness for urban air temperature models. Various implication and several areas of further research are outlined.