Review of;
Easterling, D.R. et al., 1997, "Maximum and minimum temperature trends for the globe", Science, 277, 364-367.

This paper  authored by eleven members of the pro-IPCC climate establishment,   is one in a decade long rich  tradition of obfuscating and minimising true Urban Heat Island (UHI) contamination of datasets by clever misuse of the data aided and abetted by comradely referrees and sleepy editors. The authors were, David R. Easterling, Briony Horton, Philip D. Jones,Thomas C. Peterson, Thomas R. Karl, David E. Parker, M. James Salinger, Vyacheslav Razuvayev, Neil Plummer, Paul Jamason, Christopher K. Folland.

Trends  are defined for Daily Temperature Range (DTR) for the globe for the 1950-1993 period.  DTR is maximum temperature minus the minimum.  The occasion for this paper is the availability of the newly completed GHCN (Global Historical Climate Network) dataset of 5400 stations developed by  global climate giant  NOAA / NCDC of the USA.

It has been known for years that DTR has been decreasing in many regions due to minimum (or night-time) temperatures increasing faster than maximum (daytime) temperatures (Karl 1993).
Greenhouse sceptics have over many years often drawn attention to the fact that most of the century long 0.6 degrees C  "Global Warming" is in fact at night and at high latitudes and hence is fairly benign ( Balling 1992, Michaels 1992).

A striking feature of this paper is that if it was read by a "climate scientist from a galaxy far away", he would have no clue that there is an  "enhanced greenhouse effect" which is the focus of certain debate in climate science circles on this planet.
Yet if  the  enhanced greenhouse  is warping global temperatures upwards then the signature of this process must be present in the data analysed by Easterling et al.
The fact that Easterling et al could not find a space for the words  "greenhouse effect" in this paper shows that either the enhanced greenhouse effect is having a very small and difficult to quantify effect on global climate compared to climate variation from natural processes or global temperature data are so pervaded with errors and deficiencies that the GHCN dataset is not capable of  interrogation to the degree required to define the elusive greenhouse signature.

Easterling et al is condemned by deficient methodology, deficient datasets, straw clutching, excursions off into irrelevancies, pathetic attempts at salvage, in the end  with  none of its creators paying attention  it  features the most astonishingly  error ridden  colour plate (Figure 2) to appear in a modern  climate Journal.   There are many signs that this paper would have presented more informative conclusions to the readers of Science if refereeing had been firmer and if the editors of Science were a little more awake on the day this paper dropped into their intray.

Comments in order of appearance:

[1]   GHCN Station Population Biases: At the bottom of column 1 on page 365 it is stated, "We examined urban effects on global and hemispheric trends using a metadata set developed at the U.S. National Climatic data Center. These data indicate whether a station is in an urban or non-urban environment, where urban is defined as a city of 50,000 or greater population (16)."
Now (16) is a reference to the GHCN and those of us who analyse temperature data know that the second sentence above is not correct.
In fact the GHCN dataset generated by Easterling's NOAA / NCDC colleagues & co-authors uses a tripartite classification of stations by population,  with less than 9,000 classed as (R) =Rural, 9,000 to 49,000 classed as (S) = Small Town and over 50,000 classed as (U) = Urban.
One would hope that a referee / editor might have said to the effect, " If you choose to use the GHCN data and then deviate from the station classification by population as set out in the GHCN dataset, then readers of Science would like to know why.  Furthermore, the Editors will allow you space to present your analysis  using the tripartite GHCN station classification by population."
Putting aside for a moment the nonsense of  Easterling et al claiming that stations from populations up to 50,000 are "non-urban", the GHCN station / population is seriously flawed  in that there is systemic understating of populations right through the GHCN station inventory file.
Some examples will show what I mean.
To keep this review short I will just use some examples from New Zealand,  Australian, SE Asian, Mexican and stations to illustrate how out of touch the GHCN population figures are.  As a more up to date data source I have used the web site World Gazetteer (see refs.)and have checked figures with MS Encarta 2000 Atlas in many caes.

Examples of New Zealand Station Populations from GHCN Inventory File compared to World Gazetteer 1991

Station GHCN population World Gazetteer 1991 Population
Invercargill 49,000 55,700 
Christchurch 165,000 289,100 
Wellington 136,000 148,400 
Napier 48,000 51,300 
New Plymouth 44,000 67,200 
Auckland 145,000 306,200 

NB: The Auckland figures and maybe others, highlight a problem to keep in mind when finding city populations. The 306,200 could be the Auckland City Council area which is very much smaller than greater Auckland which MS Encarta puts at 970,000.

Examples of Australian Station Populations from GHCN Inventory File compared to World Gazetteer

Station GHCN Population World Gazetteer  Population
Launceston 31,000 96,000 (1996)
Geelong 35,000 146,200  (1996)
Ballarat  36,000 64,980  (1991)
Bendigo 32,000 57,441 (1991)
Albury 35,000 77,800 (1996)
Coffs Harbour 16,000 58,000 (1996)
Bundaberg 33,000 65,800 (2001)
Mackay 35,000 69,900 (2001)
Rockhampton 50,000 64,200 (1996)
Cairns 49,000 122,000 (2001)
Kalgoorlie 10,000 28,100  (1996)
Mandurah 11,000 35,900 (1996)

Examples of South East Asian Station Populations from GHCN Inventory File compared to World Gazetteer

Station GHCN Population Up to date Population
Sandakan (Malaysia) 42,000 70,000 (1980 Web Gazetteer)
Kota Kinabalu (Malaysia) 41,000 56,000 (1980 Web Gazetteer)
Kuantan  (Malaysia) 43,000 131,500  (1980 Web Gazetteer)
Kupang (all Indonesia below) 49,000 129,300 (1990 Web Gazetteer)
Tarakan 31,000 75,500 (1990 Web Gazetteer)
Sibolga less than 9,000 71,600 (1990 Web Gazetteer)
Tanjung Pinang less than 9,000 89,800 (1990 Web Gazetteer)
Sinkawang less than 9,000 79,300 (1990 Web Gazetteer)
Jatiwangi less than 9,000 46,300 (1990 Web Gazetteer)
Cilacap less than 9,000 206.900 (1990 Web Gazetteer)
Kalianget less than 9,000 21,300  (1990 Web Gazetteer)
Sorong less than 9,000 79,700 (1990 Web Gazetteer)
Manokwari 20,000 33,800  (1990 Web Gazetteer)
Biak less than 9,000 37,500  (1990 Web Gazetteer)
Tual less than 9,000 31,600   (1990 Web Gazetteer)
Merauke less than 9,000 31,800   (1990 Web Gazetteer)

Examples of Mexican  Station Populations from GHCN Inventory File compared to World Gazetteer

Station GHCN population 1990 Population from Web Gazetteer
Cuauhtemoc 27,000 69,900
Piedras Negras 21,000 96,200
Montemorelos 19,000 35,000
La Paz 46,000 137,600
Guanajuato 37,000 73,100
Rio Verdes 17,000 42,100
Tepatitlan 29,000 54,000
Tuxpan 34,000 69,200
Manzanillo 21,000 67,700
Tlaxcala 10,000 50,500
Cuatla 14,000 110,200
Chetumal 24,000 94,200
Salina Cruz 22,000 61,700
San Cristobal 26,000 73,400

To sum up the tables above, in New Zealand,  Australia, Mexican and Malaysian stations there are frequent cases where the GHCN understates station population having the effect of pushing an Urban station down to a Small Town station ( non-urban of Easterling et al), all of which constitutes a further bias in their data and weakes their conclusions.
In the case of  Indonesian records,  the understating of populations is more serious with many instances of  GHCN Rural stations (less than 9,000) actually being Urban (over 50,000). More UHI bias in the Easterling et al non-urban data.

Because the relationship between station population and UHI effect is not linear, Delta UHI is strongest in lower populations, say under 30,000 and then flattens off as populations climb into those of  big cities.  Easterling et al  use of 50,000 as a non-urban cutoff (plus the biases due to mistaken GHCN station populations),  cleverly puts them just above this zone where Delta UHI is steepest and allows them to make the utterly specious claim that UHI effects in global datasets over 100 years is only 0.1 degrees C.

[2]    GHCN Homogeneity ?:  The fact that Easterling et al only produce graphs from 1950-93 data raises suspicions about the homogeneity of the GHCN pre-1950.

[3]   Spatial Coherence Incoherence: Mid way down column 1 on page 366  there is talk of "....less spatial coherence on the DTR map..."
Can I just say that if  the eleven authors are so lacking in perception  that they present a colour plate grossly affected by a software or data glitch, then they can not be surprised when there is "..less spatial coherence...".
See analysis [6] below.

[4]   Seasonal Cycle Joke:  At the end of the first paragraph in column 1 page 366 they draw attention to the fact that the  largest  DTR changes [decreases] are in the boreal winter and the smallest in the boreal summer then conclude ( and I kid you not ) "...., suggesting that there is an element of a seasonal cycle in the changes."
Where are the referees or Editor who could say, "What you describe here is identical to the well documented  influence of the UHI on DTR,  and UHI's  surround thousands of your data points. If you choose to ignore the UHI and allude to an unspecified "seasonal cycle" at this point, then you are simply making a circular reference, something so silly that it should not appear in Science.  At this point can I suggest you state what factor in climate science  is displaying this "seasonal cycle" that relates to the very profound seasonal variation in DTR."
It's wishful thinking I know.

[5]  1976 Jump in Temperature:   Mid way down column 2 on page 366 there is a reference to their data being affected by the well known abrupt increase in temperatures in the late 1970's.  (Kerr 1992)  Easterling et al  link this through a reference to a 1995 IPCC report to a "...fundamental shift in the El -Nino-Southern Oscillation phenomenon."   The IPCC have been trying for some years to spread the idea that greenhouse induced global warming is driving the intensification in El Nino events apparent since the 1970's. The warmers are not stupid, they can see that the media can make the link between El Nino's and storms, then we are practically at the Tony Blair position where "global warming" causes any weather / storm bad news.
However it is much more likely that both the temperature jump in 1976 and the Southern Oscillation changes are driven by ocean circulation related events.

[6]   Errors in Figure 2: The DTR panel below is labelled with letters A to H  to assist readers to pinpoint  specific errors which are described to the right of Figure 2. There are many more.

Just above A the prominent red dot indicating a strong increase in DTR, is in fact the site of warming MIN and cooling MAX. Clearly a nonsense.

Just west of B, DTR rises yet MIN warms while MAX cools. Clearly another nonsense.

Just west of C, DTR rises while MIN cools and MAX cools more. The DTR dot west of C should be blue not red.

Just south of D, DTR is falling yet MAX is rising more than MIN is rising. Another error.

At E over S Portugal there is a small DTR rise yet MAX is cooling and MIN shows a v small rise. Illogical.

Due east of E just south of Majorca is a very prominent warming in MIN. The MAX warms much less yet the DTR dot is only very tiny blue. Does not look big enough.

Just SE of F the prominent red dot shows a rise in DTR, yet both MIN and MAX cooling looks equal in magnitude.

Just east of F the DTR rises yet the MIN warms more than the MAX. No logic there.

Just N of Scotland DTR falls yet the MAX warms more than the MIN.

Just west of G over S Zimbabwe, DTR falls yet MIN cools and MAX warms.

North of H one cell in from the Great Australian Bight, is a very large warming in the MIN yet no change in the MAX and no change in the DTR.

Just east of Lake Baikal is a prominent increase in DTR yet the MIN warms more than the MAX.


Balling, R.C. (1992) The Heated Debate: Greenhouse Predictions Versus Climate Reality. San Francisco, California: Pacific Research Institute for Public Policy, xxxvi + 195 pp.

Karl TR,  et al, (1993) Asymetric trends of daily maximum and minimum temperature,. Bull Amer Met Soc   74:1007,

Kerr, R.A. (1992) Unmasking a shifty climate system, Science, V 255: 1508.

Michaels, P.J. (1992) Sound and Fury: the science and politics of global warming.  The Cato Institute.

World Gazetteer :

You read it first here

Posted 9, November, 2001

© Warwick Hughes, 2001

Back to Review comments on  climate papers by  key "IPCC Supportive"  scientists main page
Back to front page
Back to 20 Anniversary Review of Jones et al 1986
Back to Met Institutes Page
Back to Global Warming   Main Page