That cities are often warmer than the surrounding rural areas is a well-known fact. How important is this effect on the global temperature trends, is the object of heated debate.
The skeptic side contends that the UHI effect constitutes an important part of the warming observed in recent times; the counter-arguments are that the UHI effect is minuscule, or it has been accounted for, or that even if cities are warmer than rural areas the long-term trends are not affected.
The argument that UHI has been accounted for is probably the weakest: the methods use to correct for it vary from the horrible homogeneization
used by GISS, to a simple increase of the uncertainty
associated with temperature data as done by CRU and NOAA. Moreover, it is hard to account for something of unknown magnitude: the UHI for each urban station ought to be measured before it can be filtered out.
The other two counter-arguments are in fact interlocked, and part of a more general issue: it seems to me that both sides are often forgetting a fundamental detail, that it is not sufficient to consider the magnitude of UHI at one single point in time, but its temporal evolution must be considered.
I think that the magnitude of UHI will reach a plateau when an area sufficiently large around a weather station is completely urbanized, and not change much with subsequent expansion of the urban area. Then again, all types of urbanization are the same? The coal-powered cities of the late 800's have the same microclimate of modern cities with their glass and steel highrises?
The UHI effect is most important for stations that start rural then become engulfed in expanding towns.
Also, this issue is connected to the microsite effects - the effects of the temperature sensor type, enclosure and immediate surroundings.
The temperature record of a station should be compared with station metadata (type, location, surroundings) and distribution of urban areas around the station in order to correct for the various effects influencing temperature and hopefully isolating the signal due to global (or at least regional) climate changes.
John Goetz was able to plot the temperature record for the station of Burlington, Vermont
and add markers for the historically known station moves and other changes, and found that temperature trends changed dramatically in sign and magnitude for each period of time.
Correlating temperature records and metadata for each station would be a monumental work even with the will to do so, but in the field of climate science reliance on abstruse statistical techniques and mathematical models seems to have priority over experimental and field work. Moreover, for many stations metadata are missing or insufficient for this kind of comparison.
This leaves us with the question, it is possible to extract a meaningful signal from the surface temperature records we have now?
Etichette: Science, Society