Skip to content

Muller’s problem with UHI

October 23, 2011
tags: ,




According to the BEST findings, “urban warming does not unduly bias estimates of recent estimates of recent global temperature change”.

Richard Muller admits that UHI effects exist. For instance, he quotes the example of Tokyo where “excess warming of almost 2o /100 years has been estimated”. He goes on to describe the mechanisms by which UHI can increase temperatures, citing absorption of sunlight, increased heat storage of manmade surfaces, obstruction of re-radiation by buildings, absence of plant transpiration, differences in air circulation and other phenomena.

Currently GISS allow for this effect by making adjustments which result in a reduction in global temperatures of about 0.01C over the period 1900-2009. HADCRUT’s adjustment is slightly larger at 0.055C per century.

According to BEST, their own calculations can find no statistically significant UHI effect at all. Indeed they actually find a negative trend! Richard Muller explains this away by saying “This is not surprising; the fraction of the Earth’s land area denoted as urban by the MOD500 analysis is only 0.5%. Even if all these urban stations areas had heat island effect as large as that of Tokyo, roughly 3C per century, the contribution to the world average once properly weighted for land area would be only 0.5% of that, or 0.015C per century”.

This all sounds very believable. Until, that is, you read his own comments in the introduction :-


“Urban areas are heavily overrepresented in the siting of temperature stations: less than 1% of the globe is urban but 27% of the Global Historical Climatology Network Monthly stations are located in cities with a population greater than 50000.”


Under Muller’s own logic UHI would add 0.81C per century not 0.015C.


The full BEST paper can be seen here.

  1. Mike Davis permalink
    October 23, 2011 12:46 pm

    All of the stations used to measure land temperature data are located in inhabited regions of the globe that to some extent are effected by land use / land change to some extent. Think of the stations on Antarctica and the increase of research personnel since measurements began. The waste heat from industrialization and advances in life style comforts has got to have a drastic effect. To determine the effects it would require more than a record of what happened in the acre surrounding the instrument. I recall driving out of Las Vegas and noticing where the temperature dropped. The city sits in a basin / valley with an outflow region to the east. Over the years the location where the temperature changed moved further from the center and to a higher elevation. During Winter you could go to a higher elevation and see the inversion layer. Las Vegas is an extreme example because it grew from 30 thousand to almost 2 million in 50 years.
    his all brings up the issue, What if the earth has really been cooling since 1930 and what we are being presented as warming is an artifact of the way the monitor temperatures?
    I believe that the researchers have found warming since records began, but is it just like the tornadoes and hurricanes where most of the increase was due to improved monitoring capabilities. If we were monitoring tropical storms the same as in the 40s the year to date total for the Atlantic region would probably be three or four.
    ES is comparing the monitoring of 30 sites to the monitoring of 10,000 sites to the monitoring of 3,000 sites today and all the numbers in between as the site list was expanded and contracted.

  2. Douglas Hoyt permalink
    October 23, 2011 9:08 pm

    Population quadrupled from 1.5 billion to 6 billion between 1900 and 2000. Since land area didn’t change, population density increased by fourfold. Using Turok’s formula for the effect of population on temperatures, a 0.8 C temperature rise should be expected for a typical location. Some places it will be greater, some places less, but an average UHI effect of 0.8 C/century should be expected. The UHI could explain all the observered warming. It is just poor placement of most thermometers.

  3. October 26, 2011 6:18 am

    The great bulk of UHI effects on measurements kicks in at very low populations, perhaps under 1,000 in a village.
    The whole “rural-urban” metric actually used is smoke and mirrors.

  4. OldOne permalink
    October 28, 2011 5:14 pm

    Muller uses a population of 50,000 to define urban in his 1%area/27%stations statement.

    Peer-reviewed science has found that a village with a population of <5,000 can have a 2.2C winter UHI effect. re: Hinkel et al. 2003 'The urban heat island in winter at Barrow, Alaska'

    So what % of the stations had a population greater than 5,000? Probably an overwhelming majority of them. And they probably don't add much area either. So the UHI impact is much greater than the miniscule adjustment presently made in the temperature records.

  5. Gator permalink
    November 2, 2011 12:17 pm

    You do not need people at all for a heat island effect. It is infrastructure buildup. And in today’s world we have ‘small towns’ that have been engulfed in large metropolitan areas. The heat from heat islands drifts with the winds and can even effect surrounding rural locations.

    This is all about the defintinition of ‘is’.

    • November 2, 2011 11:19 pm

      And not only infrastructure, Gator.

      Roger Pielke’s (Snr) study shows that land usage changes can have a significant effect e.g. deforestation, irrigation etc.

      • Gator permalink
        November 2, 2011 11:31 pm

        Hey Paul! Correct. I have reclaimed farmland that I have planted with native species. Within that ‘natural’ environment I can get temperature differences of up to about five degrees. Extrapolating temperatures from an airport across my land can only give us a false impression of our world.

      • Mike Davis permalink
        November 4, 2011 11:37 pm

        Having my own forest, I can have ten F temperature difference between my house and the valley 500 ft west over the ridge that is full of mature trees. It is cooler in the summer and warmer in the winter over there. When I first moved here I set up a hammock over there for my afternoon naps. It has a wet weather spring that flows for about two weeks after a rain, but this time of year it will not stop until next summer.

  6. Fred Streeter permalink
    November 15, 2011 7:53 pm

    You are discussing the contribution made by Urban Areas (MODIS classification) to the warming of the Global Land Area (GLA).

    Assume that all Urban Areas warm at 3 degC/Century:
    – When the GLA is 100% Urban, the contribution is 3 degC/Century;
    – When the GLA is 50% Urban, the contribution is 1.5 degC/Century;
    – When the GLA is 0.5% Urban, the contribution is 0.015 degC/Century.

    The contribution would be 0.81 degC/Century, were the GLA 27% Urban.
    But it isn’t.

    • November 15, 2011 9:26 pm

      The GLA is not 27% urban, but 27% of the stations are. If these were properly adjusted for, the adjustment would be much more than the 0.01C that GISS makes.

      • Fred Streeter permalink
        November 15, 2011 10:01 pm

        “27% of the GHCN-M stations are located in cities with a population greater than 50,000.”

        They are present in the GLA, but in that 0.5% of the GLA that is MODIS classified as Urban, which are assumed to be warming at 3 degC/Century.
        Consequently, the contribution of the Urban Areas remains 0.015 degC/Century.

      • November 16, 2011 10:16 am

        Fred, that would work if stations were evenly spread across the planet, but they are not.

        Most rural areas have no station coverage and therefore urban stations cover much more than 0.5% of the area where measurements actually exist.

        Remember as well that UHI effects can occur even in small towns. How many more GHCN stations are based in towns of say 5000+ or airport sites?

      • Fred Streeter permalink
        November 16, 2011 3:56 pm

        You quote the BEST paper as saying:
        “Even if all these urban [stations] had heat island effect as large as that as that of Tokyo”

        It actually says:
        “Even if all these urban [areas] had a heat island effect as large as that of Tokyo”

        Thus the weighting by Urban area, rather than number of Urban stations.

      • November 16, 2011 7:01 pm

        Thanks, Fred. Fixed!

        The point I was making, though, is that if you take the land area where temperatures are actually measured, the urban area is much greater than 0.5%, given that the areas not measured are pretty much all rural. Are urban temperatures therefore giving a disproportionate influence on overall temperatures?

  7. Mike Davis permalink
    November 19, 2011 6:26 pm

    Temperatures are measured and represent less than 1% of the GLA. All the measurements represent a degree of modernizing. It would be BEST to discard BEST along with other attempts to measure SAT for the GLA!
    Any individual site records shorter than 80 years or with more than 1% missing data should be discarded first. Then any remaining records should be given an error range of 2C +/- anything within that range equals no change either way! Also because we are dealing with mean temperature the should stick with mean rather than attempting to create a meaningless average and it should be for the entire time period rather than a hand picked period. Using an average of the mean gives a biased result that is improperly weighted! Primarily because there is no way to properly weight unevenly distributed stations.

    • November 19, 2011 6:35 pm

      Not one rural station would qualify in Africa then, Mike!

  8. March 30, 2012 10:00 pm

    Reblogged this on Climate Ponderings and commented:
    Thank You! I needed this one for my scrapbook [ blog ]

Comments are closed.

%d bloggers like this: