토지이용․토지피복 특성이 서울시 도시기온에 미치는 영향

This study is to develop statistical models to explain relationships between air temperature at Automatic Weather Stations(AWS) and urban land use characteristics inside 1 km-radius area around each AWS. Four regression models are initially presented, which explain the variations in average air temperature during January, April, August, and October of 2005. Then, these model specifications are adopted to estimate seasonal regression models of 2005. Lastly, the same specifications are applied to obtain seasonal regression models of 2004 as well as 2006, which are tested with null hypotheses that the model for each season of the two years has the same model specification as each of 2005 seasonal models by using Chow test. The monthly regression models explain about 78~90% of the variations in average air temperatures. ‘Green and open space(G)’ is identified to be negatively related to temperature, while ‘Transportation facility area(TA)’ and ‘Residential area(RA)’ are identified to be positively related to it. Among those explanatory variables, variable TA is analyzed to be the most influential one. The monthly model specifications are adopted to estimate regressions models for seasonal average air temperature of 2005, which shows that the seasonal models have more explanatory power than monthly ones do. The same model specifications are also extensively applied to explain seasonal air temperatures of 2004 and 2006, showing that the newly estimated models are not better than models of 2005. Accordingly, in order to verify that there is no structural change in seasonal models during 2004~2006, Chow test is conducted. As for summer seasons during that period, no evidence for its structural change is found, which confirms that a model specification for one specific time period is applicable to models for other adjacent time periods.