The Surface Temperature Record and the Urban Heat Island

There are quite a few reasons to believe that the surface temperature record – which shows a warming of approximately 0.6°-0.8°C over the last century (depending on precisely how the warming trend is defined) – is essentially uncontaminated by the effects of urban growth and the Urban Heat Island (UHI) effect. These include that the land, borehole and marine records substantially agree; and the fact that there is little difference between the long-term (1880 to 1998) rural (0.70°C/century) and full set of station temperature trends (actually less at 0.65°C/century). This and other information lead the IPCC to conclude that the UHI effect makes at most a contribution of 0.05°C to the warming observed over the past century.

More recently two papers have sought to show directly that the effect is minimal. David Parker in a recent paper in Nature, said:

Urban heat islands occur mainly at night and are reduced in windy conditions. Here we show that, globally, temperatures over land have risen as much on windy nights as on calm nights, indicating that the observed overall warming is not a consequence of urban development.

The reasoning behind this is that the major cause of urban heat islands is the reduced cooling that occurs at night when the “view to space” of the surface is blocked by buildings. In more rural areas, cooling can be stronger. This is more likely to occur in calm conditions, when air near the surface is less well mixed with air higher up. Since the UHI effect is reduced in windy conditions, if the UHI effect was a significant component of the temperature record, then we would see a different rate of warming when observations are stratified by calm or windy conditions. The absence of such an effect (which is what Parker finds) is, conversely, evidence of a minimal UHI effect on the record.

The Parker paper, however, is very brief and recent and has not had time to be fully tested by the scientific community. A paper by Peterson (2003) is of interest because it has been out for a while and is more comprehensive. It agrees with Parker. The paper, “Assessment of urban versus rural in situ surface temperatures in the contiguous United States: No difference found” published in the Journal of Climate finds that the effects of the urban heat island may have been overstated and that “Contrary to generally accepted wisdom, no statistically significant impact of urbanization could be found in annual temperatures.”. This was done by using satellite-based night-light detection of urban areas, and more thorough homogenisation of the time series (with corrections, for example, for the tendency of surrounding rural stations to be slightly higher, and thus cooler, than urban areas). As the paper says, if its conclusion is accepted, then it is necessary to “unravel the mystery of how a global temperature time series created partly from urban in situ stations could show no contamination from urban warming”. The main conclusion is that micro- and local-scale impacts dominate the meso-scale impact of the urban heat island: many sections of towns may be warmer than rural sites, but meteorological observations are likely to be made in park “cool islands”.

The evidence points to a warming of about 0.6-0.8°C over the past century and a neglible effect on this from the UHI. While some ‘contrarians’ appear determined not to accept this finding, the evidence they cite appears thin indeed compared with the published research.


Parker, D.E., Large-Scale Warming is not Urban, Nature 432, 290, doi:10.1038/432290a, 2004.

Peterson, T.C., Assessment of urban versus rural in situ surface temperatures in the contiguous United States: No difference found, Journal of Climate, 16, 2941-2959, 2003.

10 comments on this post.
  1. stephan harrison:

    The supposed contamination of the surface temperature record is always used by skeptics to argue against GW. However, there appears relatively little in the scientific literature on a global review of the response of the cryosphere to GW. It seems clear to me that that mountain glaciers and permafrost are: 1 sensitive indicators of changes in temperature; 2 uncontaminated by urban heat islands; 3 have short response times (no problem with lagged response to Little Ice Age cooling); have wide geographical coverage (especially in remote areas). Since the vast majority of mountain glaciers are in recession, and most studies show permafrost to be melting, this knocks the sceptics’ position re urbanisation on the head.

    Response: There is what looks like a good review in the IPCC tar Chapter 2, section “Mountain glaciers” – William.

    PS….a great site.

    Stephan Harrison

  2. Stig Hasselqvist:

    What happens when the arctic icebelts melt â?? in summer?

    We know that the arctic icebelts melt during summer in the northern hemisphere. As ice is composed by sweetwater, the cold meltwater is spread over large areas in the north.
    As cold sweetwater can be lighter than not so cold saltwater, the colder sweetwater spreads on the surface and brings a cooling of the air over those large areas.
    During summer, the northern hemisphere will mainly become cold, rainy and cloudy. This is what happened during last years, all over the northern continents. Apart from that the Gulf Stream now and then pushes a warm bubble northwards, the summer climate will mainly become gray and cold in the nearest future.
    The heated airstreams from south will be restricted to reach the same northern latitudes as before; thus a heat concentration is accumulated as happened extremely in 2002 in central Europe ( e.g. France).
    In winter the greenhouse-effect will influence the climate as it conventionally has done the last years, with mild, cloudy weather on the northern hemisphere.
    I would like to read some comment of this postulation from meteorologists.

  3. Jim Dukelow:

    A couple of comments on UHI:

    1) The effect of urban heat islands on global and regional averages could be reduced if the compilers of the temperature datasets used area-weighted averaging of the individual station temperatures. This could be done fairly easily and rigorously using the Voronoi grid induced by the station locations. The numerous stations in and around large cities would be down-weighted and rural stations would tend to be more heavily weighted.

    2) Three years ago, I tried to get a handle on whether UHI was responsible for the recent warming trend in most of the temperature datasets by comparing the trends for the UAH/MSU 2LT channel and the Jones et al. surface data for some of the world’s “empty places”.

    Location/ MSU (deg C per decade) / Jones et al. (deg C per decade)
    N Quebec, W Labrador / 0.317 / 0.327
    N Ontario, James Bay / 0.413 / 0.533
    N Alb/Sask/Man / 0.424 / 0.470
    E Yukon and Nunavut / 0.101 / 0.666
    N Alaska / 0.191 / -0.008
    SW Alaska / 0.196 / -0.013
    Arabian peninsula / 0.021 / 0.328
    Sahara / 0.105 / 0.346
    W China and W Mongolia / 0.335 / 0.328
    Outback / 0.007 / -0.057
    Amazon Basin / -0.183 / 0.171
    Patagonia / -0.013 / 0.049

    All trends from the Idso’s web-site world temperature calculator at The MSU and Jones et al. trends for these empty places are weakly positively correlated with r = 0.426.

    Notable in this data is how strong the warming in the North is and how different the Southern Hemisphere is.

    Jim Dukelow

  4. Ferdinand Engelbeen:

    I have no idea how good the urban stations are corrected for UHI by rural stations. For another discussion (where to find the effect of aerosols) I was looking for more or less reliable rural station data for North-West Russia for the period 1945-2004. In the whole Northern halve of Russia there are only two (Vytegra and Reboly), six others ceased operation between 1970-1990. Only 7 urban stations did span the 1945-2004 period, with between 102,000 and 8 million inhabitants…

    Surface data are more adequately corrected for urban island heat effect in the US and other OECD countries. But the problems can be found in non-OECD countries, especially around the equator. Look e.g. to the data for Salvador, a town of 1.5 million inhabitants. That should be compared with rural stations to correct for urban heat island effect. But the nearest rural stations are 458-542 km away from Salvador (Caetite, Caravela, Remanso). And their data are so spurious, that it is impossible to deduct any trend from them. Quixeramobin is the nearest rural station with more or less reliable data over a longer time span, and shows very different trends than Salvador. Or look at Kinshasha (what a mess!), 1.3 million inhabitants, Brazzaville (opposite the Congo stream), and something rural in the neighborhood (Mouyondzi – 173 km, M’Pouya – 215 km, Djambala – 219 km,…). East Africa is not better: compare the “trends” of Nairobi with these of Narok, Makindu, Kisumu, Garissa,… Rural data trends with some reliability on a longer time span are very rare in the whole tropics. Only expanding towns have (sometimes) longer data sets which are hardly correctable. The unreliability of the data in the tropic range is thus obvious, that one can wonder how a global surface temperature trend can be calculated to any accuracy…

    [Response: You really need to read the Peterson paper more carefully. His assertion (and indeed Parkers) is that the series *don’t need any correction*. But, I guess if you really wanted a “rural” station near the coast, you could just use the nearest ocean gridpoint – William]

    That one can make different surface trends, depending on the method and corrections used, can be seen in from the October 2003 Asheville NOAA workshop:
    IPCC global surface temperature trends: GHCN: 0.31 ºC/decade, Jones: 0.25 ºC/decade and GISS: 0.18 ºC/decade.
    For (well maintained!) Australia: GHCN: 0.18 º/decade, Jones (FDM): 0.14 ºC/decade, Jones (CAM): 0.10 ºC/decade.

    [Response: I know you’re aware of so I’m a bit puzzled by you posting that – William]

  5. Kaleberg:

    Does the increasing area of urban heat islands increase global warming? Are they in some kind of equilibrium, or have they increased the overall surface warming?

  6. Pejk:

    Some considerations regarding the surface temperature record relate to the measurement techniques themselves.

    The thermometer radiation screens affect the sensor readings. A factory-new Stevenson standard screen still allows solar radiation impact on the temperature measurement of more than one degree. Accumulation of dirt, growth of mnold, even flaking of paint, is naturally occurring over the life cycle of some 15 years that the product is used in the field. The radiation error then multiplies. Weather services try to control this by regular inspection, maintenance and replacement programs, which are more or less succesful in practice. I assume these programs have changed for the better over time. In the climate statistics such improvement should be present as a cooling trend, although it would be difficult to identify.

    Probably a larger impact is from changes of the ventilation techniques used. In the older days, and in many parts of the world still today, thermometers were installed inside naturally ventilated screens. The temperature measurement then depend on elevation angle of the sun, on prevailing cloudiness and the wind speed at the time of measurement. The forced ventilation technique, (using a blower) can remove the radiation heating effect. Correspondingly this should appear as lower temperatures in the newer data.

    In the USA the forced ventilation shields have been introduced in the whole NOAA surface observation network (maybe in the 1960’s?). As this is a network-wide action, not limited to a few select stations, it might be detectable in the climate statistics if the timing were known. Correction to produce continuity of record is not simple, though, as the radiation error is so complex.

    A substantial difference between the weather forecasting and the climate analysis requirement is that the former is often satisfied with relative accuracy of measurement. Changing biases can be handled with minor discomfort if the network remains homogeneous. Climate analysis requires absolute accuracy as a means of ensuring long term data quality if technology changes are to be made. In the upper-air field, this difference of approach has been very obvious, and great efforts have been expended to produce corrections to the older radiosonde data sets.

  7. CharlieT:

    What do you make of De Laat and Maurellis’s analysis and their conclusion/speculation that quite a lot of the warming in the temperature record is down to some kind of ‘local’ heating effect?

  8. CharlieT:

    it is here

    [Response: thanks for the reference. My initial impression is that its a bit weird and probably mixing up different effects (CO2 and T both correlate to latitude). Not mentioning the Petersen study is odd. Comparing current T trends to predicted future trends and discovering that they are different also seems curious. But I’d want to read it more carefully before venturing a firmer opinion- William

  9. Spot the recycled denial III – Prof Ian Plimer «

    […] Prof Plimer also said that measurement techniques were flawed. If you measure the temperature from a certain place, such as Los Angeles, the development of the past 200 years has made the city much warmer because of all the buildings and roads absorbing solar energy – and naturally the temperature trend is upwards. On the other hand Plimer cited evidence that rural centres used to record long-term temperatures, mostly tended downwards. […]

  10. Scientists warn of climate catastrophe - Page 3 - Political Forum:

    […] them. "This is stupidity" and that pretty much sums up your post. 6 December 2004 The Surface Temperature Record and the Urban Heat Island There are quite a few reasons to believe that the surface temperature record – which shows a […]