Variables included were the size of the city population, anthropogenic heat production, elevation above sea level and topographic features (plain, coast or valley). In addition, three different climate classifications were taken into consideration.
Partial and multiple regression analysis showed a significant correlation of UHI as the dependent variable with the logarithm of the city population, with anthropogenic heat production and with the input-number of the climate classification after TERJUNG and LOUIE (1972) as an indicator of the radiation balance at the earth's surface. On the other hand, partial regression calculations did not show a statistically significant correlation between UHI and latitude. If UHI is dependant on latitude, it could be assumed that this is indirectly caused by a strong correlation between anthropogenic heat production and the radiation balance on the one hand and latitude on the other hand. Partial regression analyses with anthropogenic heat production and radiation balance as the dependent variables and latitude as the independent variable explain 43 % and 6 % respectively of the variance observed in these variables.
Factorial analyses show that, despite a significant positive correlation between UHI and a latitudinal energy factor including latitude, anthropogenic heat production and radiation balance, its contribution to the magnitude of UHI is relatively small.
Furthermore, the paper discusses the results of cluster analysis which were employed to show latitudinal variation of UHI. Four different clusters were extracted from the data set . These could be classified as one tropical, one subtropical and two medium latitude clusters.
Supplementary URL: