A Tucson climate scientist commenting on the location of the University of Arizona weather station in parking lot acknowledged that:

It’s true that situating a weather station in a parking lot is not the best location.

An understatement to day the least. He argued against locating the weather station on a nearby lawn:

The emissivity of the asphalt is higher than that of grass. That’s why it’s hotter when you stand over the asphalt compared to a grassy area. However, the native rocky terrain of the Southwest also has a high emissivity. [I tried to find a plot of the spectral emissivity of dry soil, but couldn’t locate one quickly.] I would posit that the dry soil has spectral characteristics more closely related to the asphalt than the grass.

He absurdly proposed that the parking lot wasn’t used very often, suggesting that the cars had not been moved in over 3 years. Given that he could see the parking lot from his window, this shows some inattentiveness to observational detail. His point that the nearby lawn would not be representative of Southwest terrain was fair enough, but he was wrong to merely say that the parking lot is “not the best location”. The location did not meet minimal WMO standards – a lapse thast is inexcusable in a university department.

But let’s turn to the infrared properties of asphalt – is it really like dry soil? I’ve collated a few searches below, though obviously not a full survey.

Aseada ert al 1996 in ” Heat storage of pavement and its effect on the lower atmosphere” stated:

Heat flux at the air/ground interface was observed and analyzed for various pavement materials on summer days. The surface temperature, heat storage and its subsequent emission to the atmosphere were significantly greater for asphalt than for concrete or bare soil. At the maximum, asphalt pavement emitted an additional 150 W m[-2] in infrared radiation and 200 W m[-2] in sensible transport compared to a bare soil surface. Analyses based on a parallel layers model of the atmosphere indicated that most of the infrared radiation from the ground was absorbed within 200 m of the lower atmosphere, affecting air temperature near the ground. With large difference between air and ground surface temperature at noon, the rate of infrared absorption by the lower atmosphere over asphalt pavement was greater by 60 W m[-2] than that over the soil surface or concrete pavement, a figure comparable to the absorption by turbulent transport.

These are obviously not small numbers and point to a substantial differences between asphalt and bare soil. Toudert 2005 , which has much interesting material on heat flux in an urban context, stated:

With respect to urban surfaces, Aseada et al. (1996) pointed out the importance of the pavement materials in the resulting heat fluxes and air-ground interface on summer days. They reported that an asphalt pavement emits an additional 150 Wm-2 infrared radiation and 200 Wm-2 sensible transport compared to a bare soil surface. The water content in a bare soil and thus the evaporation from it produces much lower surface temperatures. By contrast, waterproof soils such as asphalt, increasing thickness of the covering material increase the temperature and heat stored under the surfaces (Asaeda and Ca 1993). Urban surfaces with high albedos typical of light colours reduce the storage in the materials (Doll et al. 1985, Akbari et al. 1995, Taha 1997, Taha et al. 1997).

Some simple measurements were reported in a chatline here:

Old asphalt street: soles feel warm after 50 sec. Measured 141 deg F
New asphalt parking lot: soles feel hot after 20 sec. Measured 162 deg F
Steel manhole cover: soles feel very hot after 8 sec: Measured 140 deg F

By the way, I recently verified my theory about surface heat absorption on asphalt. I took my infrared thermometer out with me and measured a spot on the parking lot asphalt at 158 degrees F. Then I placed the ball of my bare foot over that spot and stood for a couple of seconds then measured the spot again. It read only 136 degrees! That means if your eyes could see infrared, you would see cool footprints left on the asphalt as you walk across it. I tried the same thing on the steel manhole cover and it only dropped about 2 or 3 degrees!

UCAR in a teaching kit for high school students (which USHCN researchers might well apply) proposed the following experiment:

Proceed to specified parking lot and cars. Take surface measurements with an infrared heat detector of each car’s roof, hood, trunk, windshield, dash board, and tire. Lastly take a measurement of the asphalt surface, parking lot line markings, and the nearest sidewalk.

Maybe Atmoz could try the experiment in Tucson and report back on whether he still considers the Tucson asphalt parking lot an adequate location for a weather station.

Fazia Ali Toudert, 2005. Dependence of Outdoor Thermal Comfort on Street Design in Hot and Dry Climate,
ASAEDA T.) ; CA V. T. ; WAKE A. ; TSUTSUMI Jun-ichiro ; BORNSTEIN R. , 1996. Heat storage of pavement and its effect on the lower atmosphere, CUTEST ’92 : conference on the urban thermal environmental studies in Tohwa, Fukuoka , JAPON


  1. steven mosher
    Posted Jul 27, 2007 at 5:45 PM | Permalink

    Funny story. My first weekend racing I pulled out my IR temperature measuring thingy.
    I shot the blacktop. It was like 85F in the morning. Now the corner of this track
    was concrete. So blacktop in the straights and entry to the corner and concrete from entry
    though the apex to exit. So, some old dog ( not a climatatologist) looks at me and shakes his
    head. must be your first race kid WHY? How hot was the blacktop? it was 85F. How hot was
    the concrete? there’s a difference? He walked away shaking his head.

    My first weekend racing on the dirt I brought out my IR thingy. Nobody else had theirs out.
    Weird. What did these gear heads know about dirt, concrete, assfault, heat content and moisture retention?

  2. steven mosher
    Posted Jul 27, 2007 at 6:27 PM | Permalink

    So, I thought it would be fun do a spot check of sorts.

    The new CRN has a station in Tucson at the SONORA DESERT MUSEUM. Its been in operation since 2002.At some pointt somebody needs to
    write a program to scrape data from CRN. I did it manually and it’s not fun.

    So, anyway. I was thinking How does the New network compare to the old? Since I can’t download the whole record for CRN
    I thought I’d just compare 1 month for a few years. Anyone who can figure out how to scrapethe data\from CRN can do amore
    complete job.

    To wet your appetite Month of June.

    University of AZ Tucson Int (airport) CRN
    2002 31.3 31.1 NA
    2003 30.3 29.7 29.6
    2004 30.0 29.5 29.
    2005 NA 29.9 29.2
    2006 NA 31.4 30.8
    2007 NA 30.1 29.5

    So, that is only one month ( we need to write a program to download this stuff)
    I picked June for obvious reasons because that is when I expect to see the
    assfault signal at its highest

  3. Sam Urbinto
    Posted Jul 27, 2007 at 7:14 PM | Permalink

    As I said before, the anomalies are showing us what the ground 5 feet below is doing, and how that is mixing with the air 5 feet up.

    But of course, that makes me insane, out of touch and a denialist.

    Wait, that is redundant; denialists are out of touch and insane. We think asphault, dirt, water, metal, grass, dirt with grass, dirt with little grass, dirt with sand, sand with dirt, plastic, wood, cardboard, styrofoam and glass have different thermal properties.

    Boy, are we stupid. Us denialists.

  4. Sylvain
    Posted Jul 27, 2007 at 8:22 PM | Permalink

    How long has this station been online?

    If it is for a short period time then isn’t enough to bias the average upward by adding higher temperatures to the data set.

    Has it always been located in that parking?

    If no, then it would surely bias the average upward.

    I concede that actual temp isn’t that important when the data is compare to itself. It becomes a huge problem when it get mixed up with other stations, then homogeneity between station is an obligation.

    This proves that we can’t rely on surface station.

  5. Chris Manuell
    Posted Jul 28, 2007 at 2:01 AM | Permalink

    I just cannot believe the comments that Atmoz has made on his blog. If it was me I would now be worried that I would be sacked by the University for incompetence.
    It would appear that as he could not find the figures for the different emissions for different surfaces and that it was a guess! Since when did a guess become part of science?
    I would call him an amateur but it seems that most amateurs use their common sense more than these so called professionals.
    It would not matter if it was just used by the university but these results are being used to create fear and to spend billions on computer models when the basic information they are feeding into them is so poor

  6. Frank K.
    Posted Jul 28, 2007 at 6:18 AM | Permalink

    In terms of the USHCN measuring *historical* temperature records, what I think matters the most is how a particular site’s environment has evolved over time. Did an asphalt parking lot exist in 1920? 1930? 1940? Probably not. So the raw temperature record has been skewed, perhaps markedly so, as the site’s environment has been developed. Supposedly, the “adjusted/corrected” temperatures published by NOAA and GISS account for UHI effects, but I haven’t located the specific algorithms they use to do the corrections. It’s very possible that in many cases, a site may not be corrected because the records indicate that it’s “rural” – even though a parking lot, building, or other structure has been constructed around the measurement site. This is why the work that Anthony Watts and the scores of volunteers at are doing is so important.

    By the way, did you know that the folks back in 1920 could read their thermometers to +/- 0.01 deg F? I have downloaded some of the raw USHCN data and they all have temperatures recorded to 2 decimal places! I think most mercury thermometers of that era were … um … a little less accurate than that (assuming of course the thermometers themselves were regularly calibrated).

  7. Geoff Sherrington
    Posted Jul 28, 2007 at 6:45 AM | Permalink

    Please can I add to the confusion?

    There was a lovely paper some months ago on temperature sensitivity to height above ground, up to a few metres. Then another one showing how max/min spreads varied according to the time of day the measurement was taken. Then we have these images of sub-standard recording stations, but the experts say that the ground is warming because the sea surface shows a similar warming (but the ionosphere and troposphere are other special cases and radiosondes do not have to agree with satellite observations, or they with each other).

    Let’s start with the sea. Not all of the sea is warming up, it seems. It is assumed by many authors that deep ocean temperatures are constant (this is yet to be shown, because the flow of heat from spreading centres and volcanic activity is hard to quantify). So they talk in terms like there is an unconformity in the sea, below which temp is constant and above which it is getting warmer at around half a deg C a century.

    The depth of this warming layer is a bit hard to pin down in the literature, presumably because of minor complications like eddies and currents. So how can one say that surface sea temperatures are rising, when the eddies and currents actually do reach the surface? What correction is made for surface water that is under ice?

    Back to the land now. There is a lot of heat from the natural geothermal gradient and this varies from location to location for reasons not understood. It can also vary with time (short as with volcanics, longer as with sediment compaction exothermic equations, and possibly cyclic through interaction of the magnetic field with cosmic particles in the broad sense). So, just like the sea, there seems to be a concept that a skin of the earth is subject to global warming, but deeper down it’s all constant, but we don’t know where to put the divide.

    Now there are experiments like neutrino counting down quite deep old mines, where I imagine there is some form of artificial cooling for the comfort of operators. But are there deep mines where there is not artificial fiddling with the temp, where temp records are kept? I’d really like to see just how constant the temperature is at say 1000 m deep delow land.

    I suspect there are data from a Russian super deep drill hole on the Kola peninsula that went down to 12 km or so then stopped. There might even be a chance to get depth profiles of the temp over the last 20 years from this hole and other deep oil holes in the Gulf of Mexico.

    Does anyone else think that too much emphasis is placed on surface sea and the land temp at the standard height above ground of a weather station sensor? Why do we have a gridded global onion-skin model that is two dimensional, when heating is felt in 3 dimensions? Surely we have to model the whole globe, so we can have super tweaking possibilities and a whole new dimension of forcings. Think of the research funds that would call for. (Yet we can’t even put a thermometer inside a city and another outside, to measure if they differ over a year).

    Come back, Alice, from Wonderland.

  8. Posted Jul 28, 2007 at 8:04 AM | Permalink

    re:# 6

    Frank K., it seems that specific algorithms are in short supply for all calculations associated with Global Climate Change. A most unusual situation, in my opinion. The models, equations, and solution methods are always of primary importance. Vague statements about conservation of mass, energy, and momentum tell us nothing. The all-important parameterizations are simply glossed over, and specifics of numerical solution methods completely ignored.

    Several will insist that every piece of desired information is available in the peer-reviewed literature. My experience has shown that trying to track down a specific model/equation/solution method always leads deeper and deeper into the past in an extremely dispersed and diffuse manner. Dispersed in the sense that the articles are all over the place in many different journals and diffuse in the sense that they are few and far between; especially the actually useful articles. The articles in many of the journals are not available at no cost to the general public. Some publishers charge up to 30.00 USD for a single article. And in several cases having sprung for the costs, I find only citations to additinal previous articles.

    It is a picture that I have found to be unique in all my career.

    All suggestions that anyone can find all the information about all aspects of all calculations are completely disingenuous, at the very least. Similar suggestions that anyone can download a copy of complex computer codes and correctly apply them to a specific analysis of inherently complex physical phenomena and processes are likewise disingenuous. Well, I guess I should speak only for myself in the latter case.

  9. Posted Jul 28, 2007 at 9:47 AM | Permalink

    Currently asphalt usage is growing at 2.3% a year. See – Presumably it has been growing at something like that rate for the last fifty years or so. Unless special efforts were made to shelter surface boundary layer temperature sensors from proximity to asphalt surfaces we should expect that the record would show a false rising.

    Anthony Watts’ photos seems to testify that temperature sites have not been spared from the general asphaltization of the earth. Therefore there would seem to be need for a general correction or adjustment. Is there in fact an asphalt correction factor routinely applied to the surface temperature data?

  10. David Smith
    Posted Jul 28, 2007 at 10:02 AM | Permalink

    Asphalt and concrete tend to change albedo over time (asphalt lighter, concrete darker). Also, and this is sometimes overlooked, the soil beneath them tends to change too.

    Soil dries, air voids formed by construction disappear, soil compacts, soil pulls away from the pavement in spots, nearby vegetation sends roots, slopes and drainage shift, the list goes on and on. These are slow-motion events happening over years.

    All of these trends can affect the ability of concrete and asphalt to conduct heat from the underlying soil, which affects the temperature of the asphalt and concrete, which affects the air temperature above them. When one is looking for tenths of a degree change, this is important.

    Bottom line: take measurements in places unaffected by human and tree/shrub activity.

  11. steven mosher
    Posted Jul 28, 2007 at 11:03 AM | Permalink

    For grins have a look at the photos of potential CRN sites ( the good network)

    Lots here:

  12. JerryB
    Posted Jul 28, 2007 at 11:07 AM | Permalink

    Re #6,

    Frank K.,

    The fractions of F that you see in USHCN files may be monthly, or annual, averages.

    The USHCN daily files that I have seen indicate whole degress F.

  13. Posted Jul 28, 2007 at 11:13 AM | Permalink

    Obviously what we need to do is to pave around all existing weather stations to create uniformity in measurement.

    There’s no need to adjust the readings, because all stations would exist under the same conditions, and certainly everything will even out with time anyway.

    If the temperature readings do not rise according to model predictions, simply park a few large trucks nearby and allow their engines to idle.

  14. Steve Moore
    Posted Jul 28, 2007 at 11:40 AM | Permalink

    Atmoz writes:

    “We can see that the USHCN data has a positive bias when compared to the airport data. However, this does not necessarily mean that it is wrong.”

    No, and Charles Manson was responsible for many deaths, but that doesn’t necessarily mean he was a bad person.

    “It is entirely possible that the temperature difference seen is natural, possibly due to elevation differences, or other means. But the two curves follow the same patterns over long time scales.”

    Yep. In an urban environment, things such as parking lots, airports, asphalt, concrete, etc, are NATURAL. Mann’s HS followed the same patterns as one would get with random data, too, so I suppose we should accept it.

    “I would argue that the changes in temperature seen are not due to the location of the USHCN surface station because the airport station is not located in a parking lot.”

    Well, where is the airport station? Certainly nowhere near asphalt, concrete, vehicles, AC outlets, etc, I’m sure.


  15. steven mosher
    Posted Jul 28, 2007 at 11:45 AM | Permalink

    re 14.

    He can see the same bias if he looks at the CRN data.. which is going to be the Reference Network.
    I posted a sample of the University versusthe airport versus CRN in comment 2. He should go pull the
    entire dataset and do a proper study

  16. steven mosher
    Posted Jul 28, 2007 at 12:12 PM | Permalink

    RE 12.

    JerryB is right. If you look at USHCN daily data it is recorded in F. No decimals.
    So observers report out the temperature in whole degrees F.

    Now, the interesting thing is if you compare Tmin, Tmax, Tmean for daily data, recalling that
    Tmean for the Day is (Tmax+Tmin)/2. Tmean is also reported out in Whole F. If you hand calculate
    the Tmean you will get fractional results. If you take Tmean from the file you will get Tmean rounded

    Let me be clearer. The file will say Tmin =9F Tmax = 12F Tmean = 11F . If you calculate Tmean for
    yourself you will get (9+12)/2 or 10.5.

    I suppose it does not matter when averaged over the month.
    JerryB you know this USHCN stuff so much better than I do. Am I making a bone head mistake here?

  17. JerryB
    Posted Jul 28, 2007 at 12:27 PM | Permalink

    Re #16,


    I don’t know where you are getting the USHCN daily files which include a Tmean.

    The last time I looked at they did not
    include a Tmean, but I just now noticed an update since the last time I looked.

    More anon.

  18. JerryB
    Posted Jul 28, 2007 at 1:38 PM | Permalink

    Still Tmax and Tmin, but no Tmean, at that CDIAC USHCN daily site.

  19. SCP
    Posted Jul 28, 2007 at 1:52 PM | Permalink

    Don’t I remember that you are not mathematically allowed to get (9 + 12)/2 = 10.5 because of significant digits? If your data have 2 significant digits, your calculated mean may not have 3.

    I seem to think that (9 + 12)/2 should be 10 or 11, depending upon whether you use round-up or round-even.

    (I think round-even avoids an upward bias which is inherent in round-up… Wouldn’t that be more appropriate for temperature observations? Do the files really contain Tmean rounded up? That strikes me as problematic.)

  20. steven mosher
    Posted Jul 28, 2007 at 1:58 PM | Permalink

    Links and steps here

    Step 1:

    Step 2: select Daily

    Step 3: click blue arrow thingy

    Step 4: click California.

    Step 5: click on tahoe city

    Scroll to the bottom of the page. You will instructions and check boxes

    “Write a comma-separated file to a download area
    (Year, Month, Day, JD, State_id are included automatically)”

    Select, temperature mean, max, min.. and you download a CVS file .

    That’s I how I did it.

  21. steven mosher
    Posted Jul 28, 2007 at 2:20 PM | Permalink

    RE 19.

    Well, that’s why I asked the question and jerryB has always been a sharp source on this stuff. So, I’m
    not drawing conclusions I’m just asking.

    My sense is the report of Tmax and Tmin ( historically) has been a visual observation of Liquid level
    in a piece of glass with markings on it. ( until the 1980s) so, reports where probably made on the whole
    degree, rounding up and rounding down, by the observer.( speculation, I’ve never seen a paper report)
    So, your going to get errors on Tmax and Tmin. Some sharper tool in the box will have to figure that out.
    Then, these two estimates are added and divided by two and the result APPEARS
    ( I await JerryBs confirmation and hope I didnt wast his time) to be rounded up.

    Now, ASSUMING, this is the case, I’m not saying it is a huge problem. It’s just one of those odditys.

  22. JerryB
    Posted Jul 28, 2007 at 3:21 PM | Permalink

    It has been a while since I compared USHCN daily numbers with USHCN monthly
    numbers, but my recollection of my impressions is that, when daily tmax and tmin
    are available:add up the daily data for the month, then do the divisions to
    one tenth degree, and you get numbers very close to the USHCN monthlies. The
    rounding error is much smaller by doing it at the monthly level, and in tenths
    of degrees, rather than at the daily level, and in whole degrees.

  23. steven mosher
    Posted Jul 28, 2007 at 4:06 PM | Permalink

    RE 22.

    I was going to do that. It won’t amount to anything. I was just curious.

  24. Frank K.
    Posted Jul 28, 2007 at 5:31 PM | Permalink

    Jerry B and Steve M,

    You’re right – the data I downloaded were monthly averages. It was still amusing to see an average temperature to two decimals, but it probably doesn’t matter as the numbers are subjected to further “adjustments” down the line.

    Frank K.

  25. steven mosher
    Posted Jul 28, 2007 at 8:21 PM | Permalink

    RE 22.

    I’m finding weirdness JerryB I need to step back. Daily’s and monthlies are not cross checking
    I looked at Orland CA for jan 1990 in daily and monthly. I ned some sleep

  26. JerryB
    Posted Jul 29, 2007 at 4:15 AM | Permalink


    Below are Orland numbers from 1990: my calculated monthly averages based
    on the USHCN daily min/max numbers, followed by the numbers taken from
    the USHCN (raw) monthly means. Only October misses by more than .05 F.

    Daily Monthly
    1990 01 45.07 45.10
    1990 02 46.30 46.30
    1990 03 55.66 55.70
    1990 04 63.78 63.80
    1990 05 65.84 65.90
    1990 06 73.05 73.10
    1990 07 80.42 80.40
    1990 08 77.58 77.60
    1990 09 73.87 73.90
    1990 10 66.11 66.20
    1990 11 53.35 53.40
    1990 12 42.15 42.20

  27. bradh
    Posted Jul 29, 2007 at 4:56 AM | Permalink

    Atmoz tells us nothing about him/herself (hereinafter, “It”). The About Me section is blank – couldn’t be bothered filling it in.

    However, It doesn’t need to expand further. It is either corrupt, or grossly incompetent. There is no other possible interpretation for It’s post.

    It is an embarrassment to It’s institution and, whilst I would normally suggest a demotion to janitor, I’d be concerned that It wouldn’t see the difference between mopping with bleach and with paint-stripper.

    Posted Jul 29, 2007 at 5:27 AM | Permalink

    #26 JerryB …
    We “Steves” back up each other…
    If you take “new” termistor thermometers with
    virtually no response time at all!! Here in Stockholm at the Observatory (see SMHI with pics by my collegue Rolf Roslund…Asplund
    was the architect of the town library in the background, another thread adress..)Tmax is
    very often 0.5-0.7 C higher than the hourly taken temperature, hardly a response to GW or
    even regional very temporal warming…could
    it possibly be the black tin roof 5 meters
    away or a parking car or both…But in the
    early morning the icecream van never comes to
    lower temperatures…And am I wrong that old
    mercury thermometers wouldn⳴ have bothered
    to move …Why do you need to take your fever
    for 3 minutes (wherever) with a mercury thermometer?? So am I WRONG when stating that
    Tmax+Tmin/2 is no…hum…mean temperature!!??
    It could be mean to us taxpayers though…
    PS Rolf Roslund is my colleague in the news-
    paper distribution business, but HE is also
    the climate station responsible some 5 days
    a month! DS

  29. fFreddy
    Posted Jul 29, 2007 at 5:42 AM | Permalink

    Re #27, bradh
    According to Atmoz’s FAQ :

    I’m a graduate student in an atmospheric sciences department. My current area of study is aerosol impacts on stratocumulus microphysics.

    From responses in the comments I think Atmoz name is N Johnson.

    Is there any justification for this assumption that the presence of asphalt leads to a step change in temperature that is independent of temperature ? (And hence does not affect trend.)
    I would have thought that the hotter it gets, the greater the effect of the asphalt relative to the “natural” temperature-i.e., asphalt is an amplifier of temperature change.

  30. bradh
    Posted Jul 29, 2007 at 6:26 AM | Permalink

    Re: #29

    Thanks fFreddy:-

    My current area of study is aerosol impacts on stratocumulus microphysics.

    Hmm..where do I start? You can study this on a “microphysics” level? That’s pretty flash!

  31. bradh
    Posted Jul 29, 2007 at 6:37 AM | Permalink

    Can anyone think of an example in physics where the temperature at midday above a bumpier, less-densely packed area would be hotter than (or the same temperature as) a dark, dense, flat surface?

  32. kim
    Posted Jul 29, 2007 at 6:41 AM | Permalink

    He may need to work on his macrophysics a little. In the discussion of his 7/10 post about Gerlich and Tscheuschner’s paper he seems confused about the 2nd Law of Thermodynamics.

  33. JerryB
    Posted Jul 29, 2007 at 8:34 AM | Permalink

    Re #28,


    We have gotten far from the asphalt path here, so I will just
    mention that Tmin may often be lower than hourly readings.

  34. captdallas2
    Posted Jul 29, 2007 at 11:03 AM | Permalink

    All of Arizona’s stations in populated areas seem to have a UHI bias which is understandable. Year round sunshine, rapid growth, and low humidity, what more could you ask for? Temperatures in these desert areas are indications of anthropogenic local warming at its best (worst).

  35. steven mosher
    Posted Jul 29, 2007 at 3:24 PM | Permalink

    Here is something Ironic. When the issue is the accuracy of the Historical Network
    we hear a variety of defenses.

    1. Anomaly method fixes everything
    2. Microsite issues are not that important.
    3. It’s already been adjusted for.
    4. The network is oversampled
    5. Ashphalt is normal.
    6. there are plenty of good sites

    Now, as the historical network gets Audited people cannot forget that a new network is being
    created: the CRN. CLIMATE REFERENCE NETWORK. This network is planned for 100 year operation with
    no station changes.

    At some point this network will be “correlated” with the historical network. Some of that work
    has started. I point this paper out to illustrate a couple things.

    1. changes in how data is reported.
    2. The effects of small differences between the sites.
    3. Wind and cloud differences between sites.
    4. Warning about homogeneity adjustments.

  36. steven mosher
    Posted Jul 29, 2007 at 3:28 PM | Permalink

    Here is something Ironic. When the issue is the accuracy of the Historical Network
    we hear a variety of defenses.

    1. Anomaly method fixes everything
    2. Microsite issues are not that important.
    3. It’s already been adjusted for.
    4. The network is oversampled
    5. Ashphalt is normal.
    6. there are plenty of good sites

    Now, as the historical network gets Audited people cannot forget that a new network is being
    created: the CRN. CLIMATE REFERENCE NETWORK. This network is planned for 100 year operation with
    no station changes.

    At some point this network will be “correlated” with the historical network. Some of that work
    has started. I point this paper out to illustrate a couple things.

    1. changes in how data is reported.
    2. The effects of small differences between the sites.
    3. Wind and cloud differences between sites.
    4. Warning about homogeneity adjustments.

    Click to access 84831.pdf

  37. bernie
    Posted Jul 29, 2007 at 4:29 PM | Permalink

    Steven (#36)
    This was written up apparently in November 2004, is there anything more recent? They said they would publish again as more data became available. The article suggests (a) raising issues about the nature of the shield are valid and (b) changes in site factors can have significant impacts especially on Tmin. Pity they did not include pictures of the two sites. In fact, it appears that the authors have no real idea as to the nature of the paired sites yet that is the topic of their paper. It is puzzling, to say the least!

  38. steven mosher
    Posted Jul 29, 2007 at 5:00 PM | Permalink

    RE 37.

    the papers on CRN are mostly “forthcoming” Even the ones on site selection! although sites
    have been selected. I just happened upon this one because of some “measurement” side discusions
    we were having ( I bet most CA guys balance their checkbooks. I bet most RC guys say “in theory,
    trends show that there should be enough money in the account”….Ok that was unfair, )

    I Think everyone on CA should read this little study. First a quote
    from the study

    It is practically impossible to separate instrument
    biases from siting or exposure biases. USCRN sites are
    generally located in open and rural areas, however the
    microclimatic environments at paired USCRN sites still
    differ from site to site. As a result, the data from these
    paired stations provide an ideal resource to analyze the
    effect of the siting differences, since the paired stations
    employ identical temperature sensors, radiation shields,
    and common sampling periods of the raw data.

    I think there are some interesting threads in here. See what you think

  39. Bob B.
    Posted Jul 30, 2007 at 5:53 AM | Permalink

    I would suggest that some one who has access to the sites where the temperature monitiors are located
    get a hold of an infrared camera. You could take an image when the air conditioners are running
    and display heating effects around the sensor.

  40. MrPete
    Posted Jul 30, 2007 at 6:53 AM | Permalink

    #19 says

    Don’t I remember that you are not mathematically allowed to get (9 + 12)/2 = 10.5 because of significant digits? If your data have 2 significant digits, your calculated mean may not have 3.

    Almost. Addition can generate more significant digits, but not shift the location of the result’s least-accurate digit. Division by an exact number (# samples) does not change sig-figs. So from a sig-figs perspective:

    76+81+86+83 = 326
    326 / 4 = 81.5

    12+18+28+16 = 74
    74 / 4 = 18.5 rounds to 18 (70 has 2 sigfig’s; round 5’s to even numbers)

    To me, this is where we see how much better it is to track real confidence/uncertainty values rather than relying on sig-figs.

  41. MrPete
    Posted Jul 30, 2007 at 6:54 AM | Permalink

    whoops, that should be “74” not “70”

  42. Sam Urbinto
    Posted Jul 30, 2007 at 1:22 PM | Permalink

    #13 Putting them all above the same type of surface may seem like an interesting way to make the data the same, but you’re still just measuring how the thermal characterists of the substance acts in mixing with the air 5 feet up. You’re measuring asphault and air. Or whatever and air.

    Take for example a few different locations. They aren’t all going to have the same amount of sunlight (how much the material warms, how much heat will be released), the same age or condition (and if not planned correctly or using the same material there versus new, thickness) of the material (rate of absorbtion and release, which would change depending on the amount of sunlight – say 10 hours a day versus 2, the material ages differently), the type of soil underneath and the amount of moisture in that soil (also affecting the material’s condition and properties), and the temperature of the air and the amount of water vapour in it (which is in turn affected by air windspeed and direction).

    The thermometer needs to be located away from human and environmental influences, and in a place indicitive of the area.

    #29 Correct. N Johnson looking at how clouds act with aersols. And the surface is responsible for the temps according to its thermal properties.

    #37 I think they said have 42 normalized CRN sites as of Dec 2003, so there’s not been an update in a while. Probably just whomever runs the site not updating it, but that’s what they got.

    #39 That’s a separate thing that could be done at the same time as the other parts of the survey, infrared camera, good idea. It would have to be done at the same time as the temp readings and over time though. (Unless one could be installed in the location on timelapse or with infrared video.) In some cases, if the observer is conveniently nearby to gather the data, a separate thermometer that automatically records things and holds some time worth of data could be put somewhere nearby the network one, also.

  43. Jaye
    Posted Jul 31, 2007 at 7:42 AM | Permalink

    mosher what do you race…I’m race sprint karts.

  44. steven mosher
    Posted Jul 31, 2007 at 7:59 AM | Permalink

    re 43.

    You ever race QRC? I was quarter midget (pit crew, not driver obviously) Thinking about Focus, Legends, or
    Micros. Outlaw sprint karts are very cool

  45. Jaye
    Posted Jul 31, 2007 at 8:34 AM | Permalink

    No I’ve never raced anything but spring karts on road courses. These sprint karts are so extreme (3g’s pure mechanical lateral grip in a flat 180 deg sweeper) that I don’t know if I could step down to a Legends car. 😉

    Now a Silver Crown or Oswego style super-mod car would be fun.

  46. steven mosher
    Posted Jul 31, 2007 at 10:07 AM | Permalink

    re 45.

    You are officially insane. At one track we raced at they had a Shifter Kart track co located.
    I wanted to see a pile up or somebody launched into low earth orbit, but alas the race was in consequential.

    Given your low Cg, have you ever come close to flipping? My sense is given the low profile you mostly have
    to worry about side force and tight/loose issues. Do you guys mess with roll centers at all?

  47. Jaye
    Posted Jul 31, 2007 at 12:53 PM | Permalink

    Flipping usually doesn’t happen and when it does its usually because of wheel to wheel contact, bumper into a sidepod maybe, and a rear wheel getting briefly stuck against some curbing or a rut just off track. The karts have so much grip that its very hard to flip them. Once you get comfortable driving them you can suddenly turn the steering wheel full right or left whilst going 80 mph down an empty straight and it will spin around maybe 1.5 times. They scrub speed at a fantastic rate.

    Karts don’t have suspensions so the typical roll center that you calculate from your A-arms doesn’t really apply. You are usually trying to get just the right amount of grip (it is possible to have too much in one of these things…too much grip and they hop or bicycle around a corner) by adjusting the frame stiffness, tire pressures, front and rear track width, seat height and stiffness, axle stiffness, and usually a combo camber/caster adjustment in the front spindles. Even messing with the amount of ackerman (the tie rods don’t have the same pivot point wrt the steering wheel shaft…which helps with grip since turning the wheel “lifts” the kart picking up the inside front and putting pressure on the outside rear) will change the way they handle.

  48. steven mosher
    Posted Jul 31, 2007 at 3:51 PM | Permalink

    RE 47.

    Ok. I wondered if you guys ever “biked” from being too hooked up. When we bike, its a study in non linear
    dynamics. Car comes around and lifts the left front inside 2 inches. Next lap it lifts 6 inches, next lap the kid goes
    ass over tea kettle as the tires gain grip in a somewhat non linear fashion. Spectacular crashes!

    We have A much higher roll center so we mess with raising and lower the roll center with panhards, plus al the usual
    stuff ( track width, ride height, PSI, ) nobody messes too much with caster/camber.. but since we only turn left
    folks mess about with lead and lag ( running front and rear end out of square) and spring rates and shock rates
    also become part of the black magic.

    I look at it as an analog nightmare. Their are boundary conditions. and when you hit your set up perfectly
    nobody can come close to you. And if your off by a 1/4 inch here or there, you throw out an anchor and move quickly
    to the rear.

    I cannot imagine running without suspension, are the tracks that smooth? how do you stay hooked up? you’ve got no downforce
    or do you?

  49. jaye
    Posted Jul 31, 2007 at 11:06 PM | Permalink

    how do you stay hooked up? you’ve got no downforce
    or do you?

    Most karters stay hooked up because the tires just have so much grip and the karts are setup to transfer as much weight as possible to the outside tires. We also trail brake a lot and tend to late apex…these two techniques are used because the even the 30-40hp two strokes can be over come by the grip so we like to keep the motors spinning and “on the pipe”.

    No down force unless you count my 225 lb frame. Super karts have some aero.

    when you hit your set up perfectly
    nobody can come close to you. And if your off by a 1/4 inch here or there, you throw out an anchor and move quickly to the rear.

    True dat. Helps one appreciate the upper levels of the sport, when you know that a bad setup is very hard to drive around and sometimes impossible. I broke a seat connection tab on the bottom of my frame which freed up the kart putting me at the front of the pack for the rest of the year. Now I use thick rubber washers to keep the seat from binding the frame.

  50. David Smith
    Posted May 16, 2008 at 2:48 PM | Permalink

    Recently I placed a recording thermometer in an open field (“A”) and another next to a parking lot (“B”). The Google Earth view is below:

    This is a progress report. The weather has not cooperated, as what I want to explore are sunny days and clear nights but the local conditions have been mostly cloudy to cloudy, with high humidity. Nevertheless, here’s a plot of the temperature difference between the two points, covering five days:

    (For reference, I’ve tested the two thermometers side-by-side and gotten very close readings, averaging no-difference, so the difference indicated in the chart is something other than instrument bias. I’ll post that comparison at a later time.)

    What the chart indicates is that, on mostly-cloudy days and nights when the wind is blowing from the parking lot towards “B”, the temperature of “B” averages on the order of 1F higher than the air in the open field. The spread is somewhat higher during daylight than at late night (this data isn’t shown here).

    My expectation is that the spread will be notably higher when we finally get a sunny day and clear-sky night.

    Those results are pretty much as expected. But, there is something which surprises me. There is a period when the wind blew from the north, from a second field towards the parking lot. “B” experienced wind from that second field. When that happened, “B” showed lower temperatures than “A”.

    The differences between the fields are that the field near “B” has rather tall grass (12″ or so) and is moister than “A”, which has short (2″) grass. Does the height and type of grass make a difference? Does soil drainage make a difference? Anthony’s efforts have documented the obvious human factors like asphalt and air conditioners but is there a large, subtle role played by ground cover and drainage? Have these slowly but broadly changed at USHCN sites over the decades? In what direction?

    There are many questions to be explored.

  51. Bob KC
    Posted May 16, 2008 at 4:11 PM | Permalink

    There is a period when the wind blew from the north, from a second field towards the parking lot. “B” experienced wind from that second field. When that happened, “B” showed lower temperatures than “A”.

    Could the parking lot be pre-heating the north wind that passes over “A”, and blowing the parking lot heat away from “B”? Or possibly the trees provide a wind break for “A”, so that it cools more slowly than “B”. Or maybe all of the above;)

  52. Posted May 16, 2008 at 6:10 PM | Permalink

    A little contribution to this thread. Coordinates: 25º 48´ North latitude and 100º 19′ of west longitude Altitude: 513 m asl:

    Asphalt in red, air in light blue, sand in dark blue and Incident Solar Radiation in purple. 8)

  53. Posted May 16, 2008 at 6:11 PM | Permalink

    # 52


    Well… you can visualize it here. 🙂

  54. David Smith
    Posted May 16, 2008 at 6:14 PM | Permalink

    Re #51 Bob, your conjecture could well be correct. I think I’ll develop an experiment or two to explore those, perhaps a series of runs with B located in the middle of the north field.

    This is something of an interactive blog science experiment, one where I’ll periodically post results and readers can interpret and suggest things for me to check. I’ll also be posting all the raw data online, so anyone can slice-and-dice it.

  55. Posted May 16, 2008 at 9:57 PM | Permalink

    I have three interesting charts. The database is real, man… I think I won’t be posting here for a long period, so these will be my last posts for awhile, although I think I will be like an observer. Thanks for all! 😉

    Trends of temperature of sand and asphalt and Solar Radiation:

    Notice that the trend of the sand is more sensitive to the intensity of solar radiation than asphalt.

    Trends of sand and asphalt and relative humidity:

    In the graph above these lines, the RH seems to respond to the asphalt and/or the sand temperature, but in the opossite way, as if the WV was “expelled” by the temperature of the soil. Is it real?

    Temperatures of sand and asphalt (normal) and relative humidity (inverted):

    Yes, it is real… The WV absorbs heat from the soil. The conditions of the atmosphere depend of the energy absorbed by the surface. The asphalt absorbs more heat and also emits more heat than sand. It’s obvious, although some people want to erase the physics, as had erased the holocene concentrations of CO2, the WMP and the LIA.

  56. Geoff Sherrington
    Posted May 16, 2008 at 11:58 PM | Permalink

    Quick suggestion about grass length. Somewhere on CA is a paper showing that the height of the T sensor above ground is rather critical wrt inversions, time of day etc. Is longer grass similar to lowering the height of the sensor above ground? One might expect it to affect wind flow, which varies at different heights.

  57. David Smith
    Posted May 17, 2008 at 6:44 PM | Permalink

    Geoff, one of the experiments I have in mind is to measure temperature in the grassy field at different heights above grade. Several of the USHCN MMTS stations I surveyed were of slightly different heights. I’m wondering if those slight differences (on the order of 5 to 10%) make a difference.

    As you mention, the grass height difference may indeed account for the temperature difference.

    Nasif, thanks for the plots. I’m trying to absorb what they’re saying. Interesting.

    Another plot I’ll occasionally report is this:

    It’s a simple comparison the standard daily average based on (min+max)/2 versus the average of the twenty-four half-hour readings. Questions to be explored include –

    1. Are the slopes different for the two locations? If so, then why?

    2. Does min/max tend to over-estimate temperature (I’m assuming that the half-hour readings are closer to the true average)?

    3. Is the relationship between min/max average and half-hour average linear over the temperature range? If not, then why?

  58. Bob Koss
    Posted May 18, 2008 at 12:42 AM | Permalink

    Here is a time series of one year of mean hourly temperatures versus (Tmin+Tmax)/2 for Miami and Boston airports.
    In addition to sensitivity to amount of clear sky, I think the farther you get from the equator the discrepancy likely increases due to fewer hours of sunlight. Generally wider day/night temperature range at higher latitudes.

  59. Geoff Sherrington
    Posted May 18, 2008 at 4:32 AM | Permalink

    David #57 re minmax versus half hour average

    Maybe depends on the response time of the detector, e.g. does a sudden gust of hot air of a few seconds raise the max reading, or is there instrumental smoothing? If it indeed raises the max, then that is one explatation of points being off the graph. Vice versa a burst of cool air. Note that the air might in theory come from a car exhaust or an air conditioner for some locations Anthony has documented.

  60. John F. Pittman
    Posted May 18, 2008 at 5:25 AM | Permalink #55 Nice post. I hope you will continue with that study. The sand sensitivity suggests grass, concrete, even water could make an interesting read.

  61. Posted Sep 13, 2008 at 2:31 PM | Permalink

    It is practically impossible to separate instrument
    biases from siting or exposure biases. USCRN sites are
    generally located in open and rural areas, however the
    microclimatic environments at paired

  62. Paul
    Posted Jul 16, 2012 at 11:02 AM | Permalink

    Heat transfer coefficient for dry soil is 1.5 w/m/k.
    Heat transfer coefficient for asphalt is .75 w/m/k
    Heat transfer coefficient for aggregate is .7 w/m/k

    According to the Virginia Highway department asphalt gets 60 degrees F. warmer than ambient – in Arizona it would be worse.

    Total conductivity loss of usual 4 inch asphalt/8 inch aggregate base is 70 w/m2 (at 60F temp differential).

    Dry soil (and it has been my experience) gets up to 50 degrees warmer than the ambient temperature(better heat transfer and lower albedo than asphalt).

    Plants almost by definition are at or slightly below ambient temperature in warm weather.

%d bloggers like this: