What was done
The authors studied the characteristics of urban heat islands in several cities in Australia with populations ranging from approximately 1,000 to 3,000,000 people.

What was learned
The maximum urban-rural temperature differences of the Australian cities were found to scale linearly with the logarithms of their populations; and Torok et al. noted that the same was true for cities in Europe and North America, but that the heat islands of Australian cities were generally less than those of similar-size European cities (which were less than similar-size North American cities) and that they increased at a slower rate with population growth than did European cities (which increased slower than did cities in North America).  The regression lines of all three continents essentially converged in the vicinity of a population of 1,000 people, however, where the mean maximum urban-rural temperature difference was approximately 2.2 ± 0.2 °C.

What it means
The four researchers note that their experimental results "were sampled during atypical conditions which were suitable for maximum urban heat island genesis," and, therefore, that "they cannot be used to adjust long-term annual mean temperature records for urban influences." Nevertheless, they say their results imply that "climatological stations in large cities should preferably be excluded from studies into long-term climate change," and that "those in small towns should be located away from the town centers."