Comparing rural and urban sites in London and Vienna
The paper begins by looking at 5 sites in and around London. Figure 1 shows absolute temperatures, clearly indicating a UHI influence on the urban sites at London Weather Centre (brown) and St. James Park (dark blue). The coolest record is the rural based Rothamsted (dark green). However, the excess urban warmth has no effect on the temperature trend - all sites show the same overall trend.
Figure 1: Annual temperature trends for five sites in and around London. Brown and dark blue are urban sites, green are rural.
A similar comparison was made between two sites in Vienna. Again, the absolute temperature is greater for the urban site but both sites show near identical trends.
Figure 2: Annual temperature trends for two sites in Vienna – Hohewarte in the center (brown) and the rural location of Grossenzersdorf (green).
Comparing rural and urban networks in China
So established urban areas show the same trends as surrounding rural areas. What about urban areas that are still developing? China, in contrast to Europe, has experienced rapid economic growth over the last 30 years with a dramatic increase in its city areas. If there were to be significant urban-related warming, it ought to be in this region and over recent decades. Figure 3 compares a range of temperature datasets:
Figure 3: Annual average temperature anomalies. Jones et all (dotted green and brown) is a dataset of 42 rural and 42 urban sites. Li et al (solid green and brown) is a homogenized dataset of 42 rural and 40 urban sites. Li (blue) is a non-homogenized set of 728 stations, urban and rural. CRUTEM3v (red) is a land-only data set (Brohan et al., 2006). This plot uses the 1954–83 base period.
That there are hardly any differences between the six series tells us several things. Small datasets of 40 stations show the same result as the 728 station dataset. In other words, for a region of this size, the average can be constructed from a limited number of sites, implying that for the 728 station network there is considerable redundancy.
As the scale increases, the overall impact of homogeneity adjustments diminishes. This might be a bit heartbreaking for those hard working boffins who spend hundreds of hours pouring meticulously over station data, ensuring the data is all homogenised (but of course, they don't do it just to calculate global trends).
And of course, the most significant finding: the trend is the same for both urban and rural groups over any of the periods. Even in the case of developing urban areas, when averaged out over large areas, urban heat island has little impact on the warming trend.