Abstract
A prominent issue in climate research is whether the global climate is warming related to the greenhouse effect. The warming trend of the global or hemispheric surface air temperature has been noted (IPCC, 1990, 1992) as evidence of the “global warming”. However, it has also been argued that the urban effect might have contributed largely to this warming trend of surface temperature. It is essemtial to estimate quantitatively the urban effect component in the surface air temperature trend.
Based on the assumption that climatic changes due to different mechanisms or forcing may show different seasonal dependencies in the anomaly time series of meteorological elements, Yasunari (1986) proposed an original method of analysis of climatic change related to the seasonal pattern of anomalies for each climatic year.
In this study, based on Yasunari's method, seasonal patterns of surface air temperature anomaly and their changes during the past 90 years (1901-1990) are examined for analysis of the urban climatic component on the surface air temperature trend using principal component analysis for 42 Japanese weather stations (Fig. 1).
More than 50 percent of variance can be explained by the first component. The examination of the seasonal eigen vector patterns shows that the first component is characterized by anomalies with the same sign for all months but with relatively high values for winter at all stations (Fig. 2).
The time series (score) of the first component generally shows increasing trends since 1901 at all stations (Fig. 3). The first component strongly seems to explain the variation due to the warming by the urban climate, although this includes the long-term climatic trend on a hemispheric scale.
The linear increasing rate of temperature anomaly (a) of the first component shows a good correla-tion with the populations of the city as an index of urbanization for large cities with populations over 300, 000 (R2=0.655; level of significance 99%). Although the linear increasing rate of temperature anomaly is greater the larger the urban population is, regional differences are obvious. Table 1 shows that for cities with similar size populations, the linear increase rates of temperature anomalies of basin cities are larger than those of inland plain cities and coastal cities. The relationship seems to be better presented if we divide the samples into two groups because the trends bend at around 300, 000 population. The increment of the linear increase rate for cities with populations over 300, 000 is larger (1.0-3.0°C/100 years: for example, Tokyo, 2.7; Kyoto, 2.4; Fukuoka, 2.2; Sapporo, 2.0) than for cities smaller than 300, 000 (0.5-1.0°C/100 years) where there is no apparent relation between the linear increasing rate of temperature anomaly and population (Fig. 4).
