Skip to content

UHI In Greece

August 1, 2015

By Paul Homewood




Cato report on a new study on UHI in Greece:



Perhaps no other climatic variable receives more attention in the debate over CO2-induced global warming than temperature. Its forecast change over time in response to rising atmospheric CO2 concentrations is the typical measure by which climate models are compared. It is also the standard by which the climate model projections tend to be judged; right or wrong, the correctness of global warming theory is most often adjudicated by comparing model projections of temperature against real-world measurements. And in such comparisons, it is critical to have a proper baseline of good data; but that is easier acknowledged than accomplished, as multiple problems and potential inaccuracies have been identified in even the best of temperature datasets.

One particular issue in this regard is the urban heat island effect, a phenomenon by which urban structures artificially warm background air temperatures above what they normally would be in a non-urbanized environment. The urban influence on a given station’s temperature record can be quite profound. In large cities, for example, urban-induced heating can be as great as Tokyo’s 10°C, making it all the more difficult to detect and discern a CO2-induced global warming signal in the temperature record, especially since the putative warming of non-urbanized areas of the planet over the past century is believed to be less than 1°C.  Yet, because nearly all long-term temperature records have been obtained from sensors initially located in towns and cities that have experienced significant growth over the past century, it is extremely important that urbanization-induced warming – which can be a full order of magnitude greater than the background trend being sought – be removed from the original temperature records when attempting to accurately assess the true warming (or cooling!) of the natural non-urban environment. A new study by Founda et al. (2015) suggests this may not be so simple or straightforward a task.

Working with temperature records in and around the metropolitan area of Athens, Greece, Founda et al. set out to examine the interdecadal variability of the urban heat island (UHI) effect, since “few studies focus on the temporal variability of UHI intensity over long periods.” Yet, as they note, “knowledge of the temporal variability and trends of UHI intensity is very important in climate change studies, since [the] urban effect has an additive effect on long term air temperature trends.”

To complete their objective the four Greek researchers compared long-term air temperature data from two urban, two suburban and two rural stations over the period 1970-2004. The UHI was calculated as the difference between the urban and suburban (or rural) stations for monthly, seasonal and annual means of air temperature (max, min, and mean).

Among their several findings, the authors report notable differences in the UHI’s intensity across the seasons and in comparing the UHI when calculated using maximum, minimum, or mean temperatures. Of significance to the discussion at hand, however, the authors note “the warming rate of the air temperature in Athens is particularly large during [the] last decades,” such that the “difference of the annual mean air temperature between urban and rural stations exhibited a progressively statistically significant increase over the studied period.” Indeed, as shown in the figure below for the stations (a) National Observatory of Athens (NOA) in the center of Athens and Tanagra (TAN), approximately 50 km north of the city, as well as for (b) the coastal urban station of Hellinikon (HEL) and again the rural station of Tanagra, the anthropogenic influence of urbanization on temperatures at these two urban stations is growing in magnitude with time such that “the mean values of UHI magnitude [calculated across the entire record] are not quite representative of the more recent period.”



Interdecadal variation and annual trends of the Athens, Greece UHI calculated between two urban and one rural station using mean annual temperatures over the period 1970-2004. The two urban stations were the National Observatory of Athens (NOA) in the center of Athens and Hellinikon (HEL), located near the urbanized coast. The rural station Tanagra (TAN), was located approximately 50 km north of the city. Adapted from Founda et al. (2015).


Such findings as these are of significant relevance in climate change studies, for they clearly indicate the UHI influence on a temperature record is not static. It changes over time and is likely inducing an ever-increasing warming bias on the temperature record, a bias that will only increase as the world’s population continues to urbanize in the years and decades ahead. Consequently, unless researchers routinely identify and remove this growing UHI influence from the various temperature data bases used in global change studies, there will likely be a progressive overestimation of the influence of the radiative effects of rising CO2 on the temperature record.



According to the graph, the influence of UHI has been steadily growing since 1970. Hellinikon, by the way, was Athens’ airport until 2001, since when it has been redeveloped. The rate of divergence between Athens and the rural site at Tanagra looks to be the best part of a degree since 1970. Indeed, the Abstract quotes a figure of 0.2C /decade.

This leads us to the question, how much allowance are GISS making for UHI in Athens and other locations in Greece?

Astonishingly, at Athens Observatory, far from allowing for UHI, GISS are doing the opposite by reducing historic temperatures by 0.1C. For instance, the annual temperature in 1970 was 18.66C, but GISS have adjusted this down to 18.56C.




Altogether, there are six operational sites in Greece that are used by GISS, of which none are rural:-


  Population x 1000 Airport? Allowance for UHI
Since 1970
Degree C
Athens Observatory 2567   -0.1
Athens Airport 2567 Y 0.1
Thessaloniki 482 Y 0.2
Kerkyra 29 Y 0
Larissa 72   0.39
Kalamata 39 Y 0



It is clear that the adjustments made are woefully short of what is needed to reflect the reality of the situation. As a result, and given the lack of any rural stations, the GISS analysis grossly overstates climatic warming in Greece.

Indeed, Founda et al find that UHI accounts for almost half of Athens’ warming since 1970. This statement takes on even more significance when we see that the familiar cyclical pattern, with a much colder interval during the 1970’s , squeezed in between the warm 1930’s – 40’s and recent warming.

The reality is that there has been little if any warming in Athens since the 1930’s. Take away the UHI effect, and we are probably looking at an underlying cooling trend.






It is becoming painfully apparent that the GISS system for dealing with UHI is pitifully inadequate, a fact that I suspect they already know. It is no wonder that they say:

The homogeneity adjustment procedure [Hansen et al., 1999, Figure 3] changes the long-term temperature trend of an urban station to make it agree with the mean trend of nearby rural stations. The effect of this adjustment on global temperature change was found to be small, less than 0.1°C for the past century.


despite the fact that the vast majority of global temperature stations are urban/airport based. 


The GISS dataset is quite simply unfit for purpose.

  1. Graeme No.3 permalink
    August 1, 2015 10:58 pm

    “The GISS dataset is quite simply unfit for purpose.”

    That depends – whose purpose?

    • AndyG55 permalink
      August 2, 2015 11:55 am

      The paper its printed on is too hard !

  2. Mark Hodgson permalink
    August 2, 2015 6:56 am

    Yet another story you won’t see reported by the BBC or by most – if not all – of the MSM.

  3. August 2, 2015 1:48 pm

    Doesn’t Greece have its own climate data keeping system?
    What does it say is the temperature?

  4. manicbeancounter permalink
    August 2, 2015 3:32 pm

    The standard definition of temperature homogenisation (such as used by the International Meteorological Organisation) is of a process that cleanses the temperature data of measurement biases to only leave only variations caused by real climatic or weather variations. This is done (at least by GHCN & GISS) by pairwise comparisons. It is fairly obvious to anyone who thinks about the issue is that if UHI trends are common to a majority of temperature stations in a particular area, then homogenisation will retain these trends, just as much as if those trends were naturally occurring.
    A more appropriate definition of temperature homogenisation is to make the data more homogeneous, for the purposes of creating regional and global average temperatures. It is only by checking the results of homogenisation against such studies as this one that we can form an opinion as to whether the process has reduced measurement biases. But questioning assumptions and checking data outputs are not things that the climate community engage in when it could undermine their work.

  5. August 2, 2015 3:54 pm

    There are two issues: UHI and station site problems. I eliminated the second by looking at all 14 CRN class 1 (best) stations in the database (WUWT). 4 urban, 10 suburban/rural. Clear evidence of UHI in all four urban raw. Homogenization did cool more recent urban temps. No evidence of UHI in any of 10 raw suburban/rural. Homogenization added spurious warming to 9 of the 10. Homogenization with poor sites that are CRN 3, 4, 5 and known to read warmer than a pristine site. QED. Unpublished. Essay When Data Isn’t contains other clear examples including Sulina Romania and Rutherglen Australia.

  6. Stephen permalink
    August 2, 2015 6:22 pm

    The UHI affects a wider area – a rural place could be quasi-urban if downwind of a city. Though the town 50km from Athens sounds a safe distance away.

Comments are closed.

%d bloggers like this: