How not to measure temperature, part 25

This picture, taken by volunteer Don Kostuch is the Detroit Lakes, MN USHCN climate station of record. The Stevenson Screen is sinking into the swamp and the MMTS sensor is kept at a comfortable temperature thanks to the nearby A/C units.


The complete set of pictures is here

From NASA’s GISS, the plot makes it pretty easy to see there was no discernible multi-decadal temperature trend until the A/C units were installed. And it’s not hard to figure out when that was.


But hey, thy can “fix” the problem with math and adjustments to the temperature record.

About these ads

22 thoughts on “How not to measure temperature, part 25

  1. Back in the 1980s, when Reagan was President the NWS was asked to cut their budget. Of course like all scientists NWS was in love with high technology, e.g., faster computers but especially more satellites. To fund their high technology they had to make a choice between what would be funded and not. So they reduced the number of weather stations that they supported. The data could all be collected by satellite. Other weather stations were moved. Stations that had collected data for long decades no longer existed and others were combined.

    Apalachicola and Tallahassee are good examples. Apalachicola had been a weather stations site for a long time. Tallahassee’s weather station was in a different location. Apalachicola, a coastal town only now seeing urbanization is gone as a station. Tallahassee’s new location has a significant bias from the previous location.

    As this blog has documented, many weather stations are now surrounded by urban sprawl and whether we were seeing increased greenhouse gases or not these stations will have a significant upward bias.

    Of course the problem with the whole global warming issue is that the data and science are hyped, misreported or poorly reported. Scientific staff, especially working for the government, do not stand up and question the data.

    Almost all of the hype from the politicos is based on seven climate models that none of these folks could ever possibly explain. All the models have significant difficulty since they do not adequately model two significant and closely related components of our climate, water vapor and three fourths of the earth’s surface, the oceans.

    Without adequately modeling those two components and without having accurate terrestrial data going forward to verify these models no one can guess what the climate will do a hundred years from now.

  2. Clearly swamp gas is an insufficient counter agent to well placed air conditioner exhaust.

  3. Re Edwin Irby’s : “Without adequately modeling those two components and without having accurate terrestrial data going forward to verify these models no one can guess what the climate will do a hundred years from now.”

    I would expand on your point — no one can guess what the climate will do next year or in ten years.


    I have finally made a significant contribution to the science of measurement with my introduction of the carbon footprint unit, the Madonna.

    “David Walton’s contribution to the science of measurement, the Madonna, is the most important new unit to be introduced since the Smoot.”



    I have no idea how the Australians picked up on this.

    *** REPLY FROM MODERATOR: Madonna’s eh? Next we could measure smell in “Crowe” units.

  4. Wow, just … wow. This would be absolutely hysterical if it weren’t so sad. Anyone with a 6th grade education should be able to look at that graph and tell something’s not right. I think this one is almost as bad as #24

  5. A crude eyeball comparison to surrounding GISS stations suggests about a 2 C increase instead of 4-5 C as posited. The exact amount of change will take a little more effort to extract from the historical data.

  6. While air conditioners cannot be good for the integrity of the data, it is premature to associate the large increase in the annual temperature with the air contidioning move. A check of the station history reveals a series of issues in the late 1990s. For instance, there is a recorded observer change in July, 1998. The equipment changed at that time too. The station closed in 1992, and it restarted in mid-1994 with an equipment change logged at that time. In 1997, there was an additional equipment change log. (That could be why 1997 appears to be missing from the annual data plot). Apparently the MMTS was installed in Aug, 2006–too late to be associated with the temperature rise.

    Then, there is the question of when the air conditioning is on. An educated guess would put the bulk of the bias in the hotter months–unless the equipment in the building is constantly generating more heat than is lost, thus requiring year-round air conditioning. In that case, the air conditioner could increase the local snow melt causing the winter months to be biased warmer more than the summer months. Any winter pictures of this site?

    From a plot of monthly average data (which I made myself using the daily coop data from the NCDC COOP CD-Rom), it looks to me the temperature increase started around late-1998–before the air conditioner move. For what it is worth, there is no strong summer signal in the monthly data. In 1999, 2000, and 2002, the warmer months relative to the long-term mean are definitely the winter months.

    A 4 degree C shift would be huge and definitely noticable if in fact it was there and concentrated in the summer months. My present guess is max 2 deg C shift warmer relative to the neighboring stations starting in late 1998.

  7. Goinggreen:

    The answer lies in the site survey form taken by the volunteer, which can be read here (PDF):

    According to the station manager, they moved the air conditioners off the roof and to the current location in 1999, and that clearly lines up with the graph. Even more disturbing, the NOAA MMS records indicate that the new MMTS electronic sensor (on the pole) was not placed until 2006. That is done by the local NWS COOP program manager. So that begs the question, how could they in good conscience place it next to A/C units? Is that incompetence, indifference, or intentional?

  8. >A crude eyeball comparison to surrounding GISS stations suggests about a 2 C increase instead of 4-5 C as posited. The exact amount of change will take a little more effort to extract from the historical data.

    Hmm. It seems closer to 3 than 2.

    Also, if 1/4 of the land stations are showing a 2-degree bias, that means overall land temp is biased 0.5C above normal.

    Here is the GHCN land data:


    Time series: Temperature January-December , 1880 – 2006
    GHCN Land Surface Data Set
    Selected Region: Longitude: -170 to -20 Latitude: 87 to 18

    Trend: 0.06°C/decade Significance: 100.0%


    Note that a -0.5C adjustment would change everything.

  9. Note that the temp jump comes after the sensor has been removed from the roof, and a rooftop alone would (if I’m not mistaken) have biased the readings right off the bat.

    So we may have to adjust for two significant biases, not merely one.


    The SENSOR was NEVER on the roof, only the air conditioners. They were removed from the roof and made ground based on 5/5/99 according to management at the radio station.

  10. Anthony, you should know by now that I am of the opinion that siting is a serious issue, the lack of good metadata is serious, that more documentation is needed, and that the present grassroots effot to document these types of stations is wonderful. I cannot answer for the COOP program manager, and I have the same questions you have, particularly regarding the MMTS siting, although perhaps with a bit less emotion.

    Evan, we can quibble (I still like my estimate of 2C), but the net result is the same–spurious microclimate warming in the temperature data, I agree with you on that.

    HOWEVER, when I look at the monthly data for Detroit Lakes, it appears to me that the warm bias starts BEFORE the reported move of the air conditioners. That’s all I am saying.

    If you look at the Tucson site from Anthony’s previous post, the station move and the series cooling line up pretty well with the station move. That is not the case in the Detroit Lakes time series, for the change there appears to occur in the Fall-1998 time period. I’d look deeper for another explanation. When, for instance, were the air conditioners put ON the roof?

  11. I wasn’t quibbling, just pointing an error (sensor not on roof) so that it was not seized upon by others.

    The date given is anecdotal, based on a person’s memory. In absence of a document showing when the A/C were moved, that recollection provided to us may be faulty.

    Again it illustrates the random nature of the observing environment.

  12. Ah. Sorry; I misread.

    “incompetence, indifference, or intentional”

    One wonders. I’d like to think the former two. Yet one also wonders why anyone would have the nerve object to your survey. (And why, oh, why are the Spacehounds of the IPCC not falling all over themselves to correct for bad NOAA data?)

    Anyone who would object to what you are doing immediately becomes an object of my suspicion.

  13. “it appears to me that the warm bias starts BEFORE the reported move of the air conditioners”

    Right. Of course it may not be a bias at all; it may have just been warmer. However your question about when the AC was installed on the roof seems relevant.

    (I’ve been running around like an idiot, placing my hands near handy street-level AC exhaust, and yes, I can feel the hot air very plainly.)

  14. Regardless of placement of the A/C units, the fact that the temp sensor is within the microclimate provided by the building seen in the photograph is sufficient to corrupt the readings.

  15. There is a factor about these outside air conditioner units which I have not seen mentioned. While probably not applicable to MN, some of these contain heat pumps, which are artifically heated in the winter to defrost the heat exchanger coils when they freeze over. Thus, the effect of these units is pretty unpredictable unless the nameplate data is recorded and the type identified.

    And, as far as value of these photographs and the ongoing data debate, I have the following logic exam:
    1. Can we assume (hope) that these stations were originally properly sited back in ancient history?
    I submit there are only two possible answers, yes or no.

    If yes, then these gross deviations from acceptable siting negates the value of current data, back to some undefinable previous time.

    OTOH, if no is the answer, then NOAA, NWS, and all the apologists have no leg to stand on regarding any data credibility.

  16. Just guessing, but probably they probably were well sited (or at least better sited than they are today): it was much easier to do so. (It seems obvious.) More open space, less concrete, and no asphalt at all.

    But one must also consider that the US government–ALL governments, for that matter–from the 1880 era were so incredibly corrupt that it defies modern imagination. (Agnew, for example, is a paragon of virtue–by the 1880 standard.)

    So the corruption factor may somehow have had an impact on the original sitings.

  17. Thus, the effect of these units is pretty unpredictable unless the nameplate data is recorded and the type identified.

    Are they? It seems like both in the summer and in the winter the AC/heaters give off heat, thus raising the nearby temperatures. It should push the winter month’s lows up higher and the hottest summer months even hotter.

    for the other months of the year they probably aren’t in use so it shouldn’t affect the temp in the spring or fall.

  18. Another long term effect not yet mentioned, I believe, has to do with automobile cooling systems. Since the photos show sensors mounted next to parking spaces, this is another artifact. Back in the 1950s, it was a rare automobile with a cooling thermostat set higher than 160° F, and many had none at all. Since then, the automakers have changed that to ever higher temperatures, sometimes exceeding 200° F, seeking higher fuel economy, etc. So, there is another subtle long term change not usually considered.
    I have not carefully read the details of processing, but do these graphs show the seasonal highest high and lowest low? Surely they must be averaged somehow, or one good pass nearby with a hot weed whacker on a hot day would make a really good peak summer reading.

  19. The real mystery to me is this: How do the glaciers know to read the faulty reports and melt? How does the Gulf of Mexico know to heat up, based on faulty data?

    Do the robins know they’re changing their migration and nesting based on this faulty data?

    And how do the insurance companies get hoodwinked into following the data and paying out more money for weather disasters made more dramatic due to warming? Shouldn’t they know that the damage isn’t as bad as their paid adjusters see?

  20. Ed, the evidence is clear that some warming has happened, and even that some of it is due to man. I am skeptical that the future effects of global wsrming will exceed, for example, the current effects of unsafe drinking water. As the Jewish people say, “why borrow trouble?”

  21. Pingback: Where Thermometers Go To Die - How not to measure temperature, part 80 « Watts Up With That?

Comments are closed.