An oldie but goodie – Microsite and UHI in 1952

Well before the current debate over the value of the near surface temperature record and its many possible biases, and well before Parker’s UHI studies sought to minimize the effect based on windy -vs- non windy days, J. Murray Mitchell published a paper in 1952 titled: On the Causes of Instrumentally Observed Secular Temperature Trends which was a quality study on the numerous possible effects of localized micro-site effects, as well as broader UHI effects related to population growth in cities. He created a tree chart of the known influences at the time:

Diagram of known effects on weather stations, from 1952

He looked at a variety of possible influences, and attempted to quantify them, both for rural and urban stations. Curiously, he discovered an effect that I’ve never heard of before, the day of the week effect:
Days of the week effect
While not a fully comprehensive study, it did hint at the fact that in the USA, there was a greater percentage of the population and business at the time that observed Sunday as a day of prayer and rest.

This paper is actually a summary of three different studies, examining New Haven, CT, and the Blue Hill Meteorological Observatory for UHI related issues, plus a broader study of 77 stations examining the effect of UHI on those stations then.

It’s a good read, and provides some grounding for the current discussions on the issues. Note: A hat tip to Roger Pielke, Junior, for bringing this paper to my attention.


  1. Bob Meyer
    Posted Aug 21, 2007 at 1:10 PM | Permalink

    “While not a fully comprehensive study, it did hint at the fact that in the USA, there was a greater percentage of the population and business at the time that observed Sunday as a day of prayer and rest.”

    So the prophets were right? By failing to attend church we are creating hell on earth?

    Seriously, if simply the level of human activity around a temperature station can change the temperature readings then the urban heat island effect is pretty strong. I wonder if anyone has tried to correlate temperature indications with major electrical blackouts like New York in 1965 or when disasters interrupt human activity like the Northridge earthquake in 1993. If there are any good proxies for human activity they would almost have to be electrical consumption, traffic, etc.

  2. Larry
    Posted Aug 21, 2007 at 1:20 PM | Permalink

    That’s quite striking. I think you’ll find that it was also more common to work 6-day weeks back then. I think the 5-day work week started to become the standard as we got closer to the ’60s. That provides some robust verification of the UHI by looking at what happens when you roll the streets up for one day out of the week.

  3. paul graham
    Posted Aug 21, 2007 at 1:22 PM | Permalink

    All very interesting, I like the idea of temperature during blackouts; however it’s only major blackout that last days that would been useful even then you would want them to be regular.

  4. Larry
    Posted Aug 21, 2007 at 1:33 PM | Permalink

    3, that’s right. The only way you’re going to be able to determine the effect of shutting down activity is to do it regularly, because that’s the only way you can tune out the natural variation. I think this is as good as it gets.

  5. John Hekman
    Posted Aug 21, 2007 at 1:46 PM | Permalink

    Sundays were quiet not only because people observed the day of rest. In addition, far more stores were closed on Sundays.

    Another change since the 1950s in many cities is that burning of coal for home heating as well as in factories has been replaced by the use of natural gas or electricity. I’m sure that the smoke from the coal burning had a cooling effect, so that the elimination of coal as an urban fuel may have imparted an upward trend to measured surface temperatures in cities.

  6. steven mosher
    Posted Aug 21, 2007 at 1:52 PM | Permalink

    UHI study.. modelling. wind tunnels, tracer studies in
    urban areas..

  7. Nathan Kurz
    Posted Aug 21, 2007 at 1:58 PM | Permalink

    Hi Anthony —

    It’s interesting to see some numbers attached to the day of the week effect. You probably missed it, but I asked you about the same in an earlier thread. I still think it would be great if you could spot some clear difference in patterns for ‘good’ versus ‘bad’ sites.

  8. SteveSadlov
    Posted Aug 21, 2007 at 2:05 PM | Permalink

    Excellent Ishikawa Diagram! I love this stuff!

  9. Peter Hartley
    Posted Aug 21, 2007 at 2:22 PM | Permalink

    The difference between the city-airport differential on Sunday and the other days of the week is an example of what is known in the economics literature as a “difference in differences” or a “natural experiment”.

    The basic idea is that the factor causing Sundays to differ from other days (in this case a social convention) has nothing to do with what is being measured (here temperature differentials), so the Sunday versus non-Sunday effect can be taken as truly exogenous — hence, a “natural experiment”.

    You might then speculate — why not just look at average Sunday temperatures versus average other day temperatures? The problem is that other factors influening airport and city temperatures in each location might contaminate that test. By looking at the city minus airport differential, however, many of the confounding unmeasured factors for each city/airport pair are eliminated.

    Bottom line: By looking at the difference in the differences across days of the week, we get a much stronger test of the effect of different levels of economic activity, and the conclusion seems to be very strong evidence of an effect of “urban activity” on measured temperature. What other possible explanation could there be?

  10. DavidH
    Posted Aug 21, 2007 at 2:24 PM | Permalink

    This is fascinating and I recommend you also spend $9 on Bäumer and Vogel
    Statistical analyses of data from 12 German meteorological stations meeting WMO standards in the period 1991–2005 are presented. These stations represent different local climate conditions in terms of both meteorology and pollution situation. For the average over data of all stations, we identified significant weekly periodicities in many variables such as temperature, daily temperature range, sunshine duration, cloud amount, precipitation, and precipitation frequency. Not only data of stations situated in congested urban areas, but also data of remote stations as e.g. on Mount Zugspitze 2960 m above sea level in the Alps showed significant in-phase weekly cycles. These weekly periodicities cannot be explained completely by local pollution effects or local heat emissions. We tend towards the hypothesis that the anthropogenic weekly emission cycle and the subsequent aerosol cycle interact with the atmospheric dynamics on a larger scale which leads to a forcing of a naturally existing 7-day period among the spectrum of atmospheric periods.

    My question is what happens to the trends at Christmas when quite often we have a 14 day weekend? Do these cycles trend towards the top or the bottom of the cycle, or put another way when we all sit indoors and go comatose is it warmer, wetter, sunnier, cloudier or what?

  11. DavidH
    Posted Aug 21, 2007 at 2:31 PM | Permalink

    Or save your $9 and look here:

    Click to access acpd-7-11545-2007-print.pdf

  12. R DeWitt
    Posted Aug 21, 2007 at 3:27 PM | Permalink

    Seems like the peak-to-trough weekly variation could be regarded as a lower bound on the UHI effect. I don’t see how one could argue against it.

  13. Slevdi
    Posted Aug 21, 2007 at 4:11 PM | Permalink

    #10 “My question is what happens to the trends at Christmas when quite often we have a 14 day weekend?”

    Only in Europe. Last time I was in the US at Christmas, I was appalled at how little holiday they take!

    But it might be interesting to add Israel with Saturday off since mid 1900s and Islamic countries with Friday off to the mix.

  14. bernie
    Posted Aug 21, 2007 at 4:26 PM | Permalink

    DavidH #10.
    Did I miss something: aren’t they simply saying air pollution is heavier during the week than on the weekends and possible Monday holidays? The pollution that they are talking about, if true, would tend to suppress the UHI, would it not? Presumably these stations also have temperature data – I wonder if they will write that up?

  15. Larry Sheldon
    Posted Aug 21, 2007 at 4:35 PM | Permalink

    Sorry about the name duplication (every Tom, Dick, and Harry is named “Larry”)–I’ll try to highlight any posts I make here that I am and Larry-who-knows-little-but-tries-to-fix-that.

    I have never seen the word “secular” used in any science context before.

    What does it mean in this context? (The really is an agendum-free, pure question.)

  16. DavidH
    Posted Aug 21, 2007 at 4:51 PM | Permalink

    I don’t think UHI could have an intrinsic weekly cycle as it is mostly to do with buildings, asphalt and sunshine. If the cloud cover has a weekly cycle from human activity then there may be an effect on UHI, which in turn might affect temperature.

    Re#13 and 14
    I thought from a first look at these papers we were seeing an aerosol effect. The question about what happens when we are not all driving to work was a way of asking if aerosols cool as the IPCC say or warm. Clearly the human effect is complex as it shows up in so many measurements including temperature, precipitation and cloudiness.(I don’t have the papers in front of me)

    I first saw a reference to this paper saying something like “it really is more likely to be wet at weekend!” and the paper shows it is.

  17. Gary
    Posted Aug 21, 2007 at 4:57 PM | Permalink

    #15 These are the Merriam-Webster definitions:

    1 a : of or relating to the worldly or temporal b : not overtly or specifically religious c : not ecclesiastical or clerical
    2 : not bound by monastic vows or rules; specifically : of, relating to, or forming clergy not belonging to a religious order or congregation
    3 a : occurring once in an age or a century b : existing or continuing through ages or centuries c : of or relating to a long term of indefinite duration

    The third definition is applicable here.

  18. Larry
    Posted Aug 21, 2007 at 4:58 PM | Permalink

    15, look here:

  19. bernie
    Posted Aug 21, 2007 at 5:13 PM | Permalink

    The problem is that there is a secular secular trend – some call it the culture war.

  20. Larry Sheldon
    Posted Aug 21, 2007 at 7:24 PM | Permalink

    {_that_ Larry) Thanks for the patient answers–clear that I should have know that.

    But it _is_ an interesting term in this context since so much discussion in the area seems to be religious in nature. Bernie’s (19) is on to something with the suggestion of a religious secular trend, a secular secular trend…. Wonder if Hansen made a clerical error…I’ll just grab my coat, I hear there is a chill in parts of the US.

  21. Posted Aug 21, 2007 at 10:14 PM | Permalink

    Obviously the temperature differential is the first step.

    Then you take measurements to find the cause.

  22. Curtis
    Posted Aug 21, 2007 at 10:34 PM | Permalink

    Have you seen this site:

    “Because urban areas both generate and trap heat, a bubble or “urban heat island” forms around the city. The temperature in Atlanta is 5 to 8 degrees Fahrenheit higher than outlying areas, and this excess heat produces increased rainfall and thunderstorms.”

    Talking about the specially intense thermal island around Atlanta – Couldn’t the entire global warming trend be blamed on increased global urbanization?

  23. Curtis
    Posted Aug 21, 2007 at 10:40 PM | Permalink

    Maybe this site is more to your interest:

  24. Hoi Polloi
    Posted Aug 22, 2007 at 7:17 AM | Permalink

    Any reference to cloud formations/coverage cq sun hours?

  25. DocMartyn
    Posted Aug 22, 2007 at 7:38 AM | Permalink

    A tiny little aside, about 5 time a year I have to reset halk the clocks in my house because there has been a break in the power during the night. Only half of my clocks have a battery back-up.
    I just wondered if the post-1990 electronic sensors were prone to give, let us say, misleading data output after a power cut?
    Are there sudden spikes in the temperature profiles during stormy weather?

  26. Anthony Watts
    Posted Aug 22, 2007 at 8:45 AM | Permalink

    RE25 All of the MMTS and NIMBUS systems used by COOP/USHCN observers have battery backup in them, so I don’t think that’s an issue. I don’t know though about the ASOS systems.

    see this for more info on the makeup of the USHCN network.

  27. kim
    Posted Aug 22, 2007 at 9:12 AM | Permalink

    Fewer vapors escaping from Hell at night?

  28. kim
    Posted Aug 22, 2007 at 9:12 AM | Permalink

    Uh, Sunday nigts.

  29. kim
    Posted Aug 22, 2007 at 9:33 AM | Permalink

    OK, more vapors. Or Hell has frozen over. Yeah, that’s the ticket.

  30. gdn
    Posted Aug 22, 2007 at 9:51 AM | Permalink

    I became familiar long ago with the “weekend effect” in the form of precipitation tendencies.

  31. gdn
    Posted Aug 22, 2007 at 10:11 AM | Permalink

    I don’t have any references handy, but as I recall, at the time, the tendency for rainy weekends was explained basically as the effect of particulate cloud-seeding. This would clearly change the temperature at ground level both through albedo level from increased cloud cover, and from the rain itself.

  32. Douglas Hoyt
    Posted Aug 22, 2007 at 1:10 PM | Permalink

    I worked for Murray Mitchell for one year. He was a very careful worker and very much a gentleman. He is still missed by some of us. I believe it is his reconstruction of climate change for the Northern Hemisphere that is the right hand figure at

    I bet he was careful to exclude UHI effects as best as he could in his temperature reconstruction.

  33. SteveSadlov
    Posted Aug 22, 2007 at 1:19 PM | Permalink

    RE: #32 – Mitchell’s reconstruction has one interesting quirk. Instead of a 1930s peak, his has a mini peak there, and a larger peak in the 1950s. (speculation)The 1950s peak was UHI/anthropogenic direct dissipation/land use mod related. It was finally overtaken by the inarguable overall 1940s – 1970s cooling. Sodden note – both world wars were preceded by cooling.

  34. Murray Duffin
    Posted Aug 22, 2007 at 2:23 PM | Permalink

    Re: 22 and 23.I have been living in Atlanta for about 3 months now, and can confirm that most days the News climatologists report Atlanta temperatures higher than the nearby towns. Can be from 1 or 2 degrees to as much as 8 degrees. I have had the feeling that the difference is less on weekends, but have not recorded data. Will have to start recording. Have no idea where the temp. stations are, but believe Atlanta is at the airport. They are not in the USHCN network.
    Georgia is having a drought this year. We had a cooler than normal summer through late july, but have been warmer than normal during August, 11 degrees F warmer yesterday. Late afternoon clouds with some rain were frequent in late July, but it would be hard to say more frequent than in the surroundings. This effect has been almost non-existant in August. Trees are already shutting down and loosing their leaves due to the heat and lack of rain. Could be that convective heat causing rain is an inverted V. We have been sitting under a stationary high for near 3 weeks, and I would guess that the lack of convective rain is from a lack of moisture in the air because nothing is being introduced from outside the high.
    Metro Atlanta population passed 5 million people in Q1 ’07. Murray

  35. Curtis
    Posted Aug 22, 2007 at 2:44 PM | Permalink

    The data the Atlanta thermal island study talks about was obtained by remote sensor (aboard an airplane) in 1996, a similar study was going to be done in Huston TX in 1997 but I havent seen anything about that study.

    What I found most interesting was this quote “In Atlanta, commercial and suburban development dramatically increased between 1973 and 1992, and nearly 380,000 acres of forest were cleared to accommodate that growth.” I couldnt help but think of how many other cities could make the same claim. In 19 years, adding 380 000 acres of urban development.

  36. Greg Murphy
    Posted Aug 22, 2007 at 3:00 PM | Permalink

    Great study, Thanks for posting it.
    Greg Murphy

  37. John F. Pittman
    Posted Aug 22, 2007 at 3:51 PM | Permalink

    #30, #31

    I could not google it, but remember a paper in the 1970-1980 range that identified changing precipitation by seeding from exhausts of airplanes.

  38. Sam Urbinto
    Posted Aug 22, 2007 at 4:42 PM | Permalink

    I’m not sure what I’m looking at.

    Around 2 readings 2 times per hour? (46 pairs of observations) Every half hour?

    Or is that they took them for 46-48 weeks twice per day?

    Anyway, you also have to take into account some things, I think it’s very difficult to tell what’s going on without temp, wind, humidity and sunlight figures.

  39. Curtis
    Posted Aug 22, 2007 at 4:54 PM | Permalink

    No the paired observations are for rural/urban sites compared over 46 – 48 weeks.

  40. Sam Urbinto
    Posted Aug 22, 2007 at 5:24 PM | Permalink

    Ah, thanks. Station pairs over almost a year. So that makes me wonder how often they measured. One reading every 6 hours is very different than one reading every 10 minutes.

    Although as I said, without sunlight amount for the day, wind speed and direction, humidity/percipitation (and probably readings at other heights and locations around the area) I don’t know what total value just having the temps is, at least for figuring out the specifics here in this case.

%d bloggers like this: