Reference
Chung, U., Choi, J. and Yun, J.I. 2004. Urbanization effect on the observed change in mean monthly temperatures between 1951-1980 and 1971-2000. Climatic Change 66: 127-136.
Background
Chung et al. say that "due to the overlapping of the rapid urbanization-industrialization period with the global warming era in Korea" (when the population of Seoul rose from 3.8 million in 1966 to 9.6 million in 1985), the background climatic trends from these urbanized areas "might be contaminated [by the] urban heat island effect." Hence, they proceeded to investigate this subject.
What was done
In the words of the authors, "monthly averages of daily minimum, maximum, and mean temperature at 14 synoptic stations were prepared for 1951-1980 (past normal) and 1971-2000 (current normal) periods," after which "regression equations were used to determine potential effects of urbanization and to extract the net contribution of regional climate change to the apparent temperature change." Twelve of these stations were growing urban sites of various size, while two (where populations actually decreased) were rural, one being located inland and one on a remote island.
What was learned
In terms of change over the 20 years that separated the two normal periods, Chung et al. report that in Seoul, where population increase was greatest, annual mean daily minimum temperature increased by 0.7°C, while a mere 0.1°C increase was detected at one of the two rural sites and a 0.1°C decrease was detected at the other, for no net change in their aggregate mean value. In the case of annual mean daily maximum temperature, on the other hand, a 0.4°C increase was observed at Seoul and a 0.3°C increase was observed at the two rural sites. Hence, the change in the annual mean daily mean temperature was an increase of 0.15°C at the two rural sites (indicative of regional background warming of 0.075°C per decade), while the change of annual mean daily mean temperature at Seoul was an increase of 0.55°C, or 0.275°C per decade (indicative of an urban-induced warming of 0.2°C per decade in addition to the regional background warming of 0.075°C per decade). Corresponding results for urban areas of intermediate size defined a linear relationship that connected these two extreme results when plotted against the logarithm of population increase over the two-decade period.
What it means
In light of the significantly intensifying urban heat island effect detected in their study, Chung et al. say it is "necessary to subtract the computed urbanization effect from the observed data at urban stations in order to prepare an intended nationwide climatic atlas," further noting that "rural climatological normals should be used instead of the conventional normals to simulate ecosystem responses to climatic change, because the urban area is still much smaller than natural and agricultural ecosystems in Korea." Their advice should be heeded by everyone, including the IPCC.