Lazarus of the Thermometers

The “Great Dying of Thermometers”, to use E.M. Smith’s (Chiefio) apt phrase has been discussed here from time to time for several years. Contrary to some misconceptions, it’s not that people mysteriously stopped taking temperatures around 1990. GHCN had said that it would update its non-CLIMAT (mostly airports) station data from time to time and it seems that such an occasion hadn’t arisen since the reign of Bush I.

The combined inability of NOAA and NASA to locate up-to-date from such locations as Wellington NZ, Dawson, Canada and the Bolivian sites whose existence Tamino denied has been a regular source of amusement here.

Bob Koss wrote in to report the resurrection during of some long-dead stations in 2009. Bob:

GHCN updated about 380 neglected stations in their database during 2009. GISS has picked up many, and maybe all of them. As of 11-14-2008 GHCN had 1233 records for 2008. Now they have 1613 of them. Seems they updated early 2009, as my June GHCN download contains them. Didn’t check until downloading it today.

In early 2008 I scraped GISS for just basic information. Below are the start/end dates GISS showed for a few of the stations. Now they are current through 2009 with many having the intervening years filled with data.

———————–Back from the dead——————
306840080000 SAN CRISTOBAL -0.9 -89.6 RURAL AREA 1951 1991
507932920000 GISBORNE AERO -38.65 177.98 30000 1962 1991
507930120000 KAITAIA -35.13 173.27 RURAL AREA 1961 1985
507936150000 HOKITIKA AERO -42.72 170.98 RURAL AREA 1880 1991


  1. Posted Jan 31, 2010 at 12:11 AM | Permalink

    Temperature recording from Kaitaia Airport has been reported daily for more than 40 years. (I grew up there.)

    I suspect the same is true for Hokitika and Gisborne. There is no reason data from such locations should not be included, though I have not seen the actual weather stations. Hokitika and Kaitaia are remote rural towns, close to large seas. Gisborne is a city on the Pacific coast.

    • Beth
      Posted Feb 1, 2010 at 8:25 PM | Permalink

      Two stations at Wellington, NZ clearly show different temperature records.
      Baring Head, an isolated rural site:
      The dataset at for Wellington NZ runs from 1880 and shows no overall increase to 1989 (when the lighthouse was automated). This site is at 41.3S,174.8E
      versus central Wellington city (~20 km away) demonstrates problems with city centre temperature measurement.

      Then read JWD Hessell “Apparent trends of mean temperature in New Zealand since 1930″ [NZ Journal of Science 1980, vol 23, 1-9], which discusses the issues identified in 1980 with urban temperature observations in Auckland, Wellington (Kelburn) and other centres. A copy of this was at
      The Kelburn site would definitely give a higher figure to be used for NZ’s recently reported so-called 1 deg rise.

  2. Bob Koss
    Posted Jan 31, 2010 at 12:13 AM | Permalink

    Here is an animation of 1998, comparing a map I downloaded 11-15-2008 with one I downloaded tonight.
    There is a drop of 0.03c in the yearly anomaly. Don’t know if that would apply consistently though.

    • Geoff Sherrington
      Posted Jan 31, 2010 at 3:34 AM | Permalink

      Bob, This might be old ground, but when revisions happen (such as a large reduction in the number of global stations over the years) is there a corresponding correction to the selection of stations, both as to their number and their average temperature, used in the reference period set? Sorry if I’m repeating a well known concept.

      • Bob Koss
        Posted Jan 31, 2010 at 4:19 AM | Permalink


        My understanding is that all stations currently in use, that have even one year in the 1951-1980 reference period, are used to establish the baseline for comparison. An noticeable example would be Byrd. It started in 1980 and it’s 1980 readings are the only baseline for many cells. It is the only station with in 1200 km. If another station were established in the area, it would also use the single year of Byrd 1980 values as a baseline.

        In the animation above Byrd is entirely responsible for the disappearing, appearing blue area near the Antarctic coast near 120W. It had no temperature readings for Jan and Feb 1980. That means you can’t generate a GISS map showing any anomaly in that area for those months. Nor for any Dec-Feb quarter year.

        The older map in that animation predates their inclusion of Byrd in calculating anomalies. So that map shows the area empty.

        There are many stations which started late 70’s and have only a couple years which define the baseline for the anomaly period. Most of them you don’t notice because their short anomaly baseline is combined with other stations nearby. The late 70’s were when temperature was bottoming out, so you can imagine the impact.

        Side note:
        Each individual station affects the 2×2 grid cell anomalies for about 0.9% of the earth’s surface.

      • Bob Koss
        Posted Jan 31, 2010 at 4:44 AM | Permalink

        Let me clarify what I mean in my first sentence about stations currently in use.

        I mean a station that has a homogeneity adjusted record in addition to the usual combined record. Just how they determine when to create that record, I am unsure about. There might be some subjectivity involved.

        I do know that a station must have a 20 year history to be considered.

  3. Ian Beale
    Posted Jan 31, 2010 at 3:35 AM | Permalink


    I put this up at

    – it might be of interest here too with the reply. Delete if not.

    “E.M. This might be a bit off thread but it seems to fit

    As a rancher from experience my first daily look at what might happen weather-wise is rather than the local BOM.

    In the face of the above post how does wxmaps get to update twice a day with what I presume is world wide data?

    And BSCH seem to update 4 times a day according to their site info???

    REPLY: [ Well, the data are clearly available if you actually want them 😉 but for questions of motivation, well, you would have to ask them… But I think it’s pretty clear that it’s a choice, not an accident of history. (Or put another way: Not choosing to do something IS a choice.) -E.M.Smith ]

    (BSCH is

  4. Posted Jan 31, 2010 at 4:42 AM | Permalink

    They will not include places like Hokitika and the Chatham Islands because they show a temperature drop.

  5. Adam Soereg
    Posted Jan 31, 2010 at 5:03 AM | Permalink

    A complete list of Bolivian synoptic stations which have at least 15 years of data between 1961-90:

    It would take about an hour to update all these sites to December 2009, by using the Global Summary of the Day (GSOD) archives and 3 or 6-hourly SYNOP reports. I can do it with a regular laptop and an Internet connection. Why the folks at NCDC are unable to update their data from non-CLIMAT sources while they have unimaginable computing capacity?

    WMOind. ICAO Station name Lat. Lon. Alt. Data availability

    85033 SLGY Guayaramerin -10-49 -065-20 130 1973 2009
    85041 SLCO Cobija -11-02 -068-52 235 1956 2009
    85043 SLRI Riberalta -11-00 -066-07 141 1951 2009
    85104 SLJO San Joaquin -13-04 -064-40 140 1956 2009
    85114 SLMG Magdalena -13-20 -064-09 140 1971 2009
    85140 SLRY Reyes -14-19 -067-22 140 1971 2009
    85141 SLRQ Rurrenabaque -14-28 -067-29 204 1951 2009
    85151 SLAP Apolo -14-44 -068-25 1415 1971 2008
    85152 SLSB San Borja -14-52 -066-45 194 1971 2009
    85154 SLTR Trinidad -14-49 -064-57 155 1951 2009
    85196 SLCP Concepcion -16-08 -062-06 497 1949 2009
    85201 SLLP La Paz/Alto -16-31 -068-11 4058 1898 2009
    85207 SLSI San Ignacio D. -16-22 -060-57 413 1951 2009
    85223 SLCB Cochabamba -17-25 -066-06 2548 1941 2009
    85230 SLCN Charana -17-35 -069-27 4054 1951 2007
    85242 SLOR Oruro -17-57 -067-04 3072 1949 2009
    85245 SLET Santa Cruz/El. -17-48 -063-10 418 1943 2009
    85247 SLJE San Jose De Ch. -17-48 -060-45 284 1951 2009
    85268 SLRB Robore -18-19 -059-45 276 1951 2009
    85283 SLSU Sucre -19-01 -065-16 2904 1966 2009
    85289 SLPS Puerto Suarez -18-59 -057-44 134 1951 2009
    85315 SLCA Camiri -20-00 -063-33 798 1951 2009
    85364 SLTJ Tarija -21-33 -064-43 1854 1966 2009
    85365 SLYA Yacuiba -21-57 -063-42 645 1914 2009

    • Posted Jan 31, 2010 at 9:20 AM | Permalink

      Excellent information, Adam.

      What link did you use to get this information? I searced for “Global Summary of the Day” and got the NOAA site, I found this site not very user friendly, and that I could only get annual data, i.e. one file for each station for each year. But as you pointed out, the data is readily available and at NOAA at that.
      Thank you.

      • Adam Soereg
        Posted Jan 31, 2010 at 10:50 AM | Permalink

        FTP access to annual files for all station files:

        If you would like to browse the daily and monthly values, I recommend you this site:

        You have to type the WMO index of a specific station. The end of the url contains 6 digits, where the first 5 is the WMO index and the 6th one is zero in most of the cases.

        See the GSOD station inventory for all included stations:

      • HectorMaletta
        Posted Jan 31, 2010 at 9:40 PM | Permalink

        Bruce, are these data corrected for urban heat island effect? In the case of Bolivia for example: La Paz-El Alto refers to a desolate and windy high plateau above the city, which was practically desert in 1898 (the start of the series) but now has an airport, a population above 1 million people, and lots of factories. Such station, not having changed position along the 20th century, must undoubtedly show an increase in temperature due to the increase in cement, asphalt, aircraft engines and factory machinery going on around it, even if climate has not changed a bit in the meantime. The same for the one in Santa Cruz, a sleepy small town of 30,000 in 1943, when its series starts, and now a sprawling city of 2 million people. I suppose the historical records for such stations must be retrospectively calibrated with other stations at places that remained rural in the same region, but having worked in Bolivia I do not remember any such adjustment when I asked for historical records.
        Even if the adjustment is made, the heat island should enter the average with a weight reflecting the small territory they cover, whilst rural stations receive weights reflecting the wider zones they represent, but again I do not think these refinements exist anywhere (not only in Bolivia, I’m afraid). In Buenos Aires, where I am now, the main station at the federal district is at the Agricultural Science school of BA University, and it routinely yields 2° to 3° C higher temperature than the next station at the airport, some 20 miles away, and 3°-4° more than rural stations a bit farther away and not close to the airport. The Met Office in Argentina gave me the raw data since 1970, but they show no signs of any adjustment. I suspect back in the 1930s with a much smaller urban area, the differences should have been much narrower, since they are all essentially in the same general geographic area (if anything, the Agric.School station is closer to the River Plate and in olden times it should have yielded cooler, not warmer temps). What gives?

        Steve: UHI is the major controversy with these records

        • Adam Soereg
          Posted Feb 1, 2010 at 4:37 AM | Permalink

          GISS version of La Paz/El Alto: shows no warming, but it contains two very dubious step-like changes. Unfortunately, I do not know anything about the station history, except that the measurements started in the city, about 3600 meters above sea level.

        • HectorMaletta
          Posted Feb 1, 2010 at 8:17 AM | Permalink

          La Paz is a unique case because there are huge variations in temperature at different parts of the city. El Alto (and airport) is at the high plateau (4100 m above sea level); the rest of the city is in a narrow valley with a vertical range of about 1000 m; the historical center of the city is at about 3600, and the lowest parts (developed only in recent decades) are at about 3100m. On a given winter day you can have snow or frost in the airport and much warmer weather 1000 m below, possibly some 20-25° C. There is also wide circadian variation.
          Under these conditions any change of location may potentially have perceptible effects on the historical record (such as the step changes Adam noticed).

    • NicL
      Posted Jan 31, 2010 at 10:14 AM | Permalink

      Adam, Thanks, that is very helpful to know.

      I see that I can manually obtain GSOD data on a one station at a time basis at (or on a year by year basis at, etc., which I could automate the download of); no doubt with a bit of effort I can write a script in R to convert the resulting GSOD files into monthly summaries in the same format as the ghcn files. But may I ask if you have found an easy way to automate the download of all this station data and/or to freely obtain monthly average rather than GSOD data?

      If you are using a quite different source, could you specify the url, please?

      I am unsure why one needs 3 or 6-hourly SYNOP reports (which I haven’t located) to obtain monthly means, as well as GSOD data?

      I am also slightly confused as to your source of data availability – may I ask at what URL you found this? Availability seems to be given for all stations in However, for, e.g., 85033 SLGY Guayaramerin, data is listed as being available from 1943 on, not 1973 on as you state.

  6. Adam Soereg
    Posted Jan 31, 2010 at 5:09 AM | Permalink

    2 out of 24 Bolivian stations are not reporting since 2007 and 2008, 22 of them can be updated to December 2009.

  7. Posted Jan 31, 2010 at 7:06 AM | Permalink

    Thanks to the great work by you, Steve, Anthony Watts and many others, we know of the condition of the temperature records and measuring stations. I would like to write letters to my congressman and senators and I would like to propose that the data archiving and storage be taken out of the hands of the academics and placed in the hands of professionals. The National Bureau of Standards has various statistical data functions, and is a part of the Department of Commerce. Whenever I start to put my thoughts down I realize how much I don’t know about the subject. I would appreciate if you, Steve, would put out a position statement on what your recommendations are, that I and your readers could put in letters to their governments.

    • MinB
      Posted Feb 2, 2010 at 6:59 PM | Permalink

      NIST (Natl Institute of Standards & Technology) shares a campus with NOAA in Boulder, with NCAR just about 2 miles away. Possibly a good fit too.

  8. Posted Jan 31, 2010 at 9:55 AM | Permalink

    this could be maybe OT, but if find it extremely important. Somebody on the internet has performed an experiment by plotting together just rural stations in the USA and comparing the trend thus obtained with the urban stations trend.

    The result was astonishing: for 32 selected rural stations there was no trend 1900-2009 whatsoever, while the temperature rose by 1,5 degrees C in urban stations. Interestingly enough, the rural record in the USA closely parallels the record from Greenland and Arctic, by having 1930s and 1940 as warm or warmer than recent two decades.

    It seems that rural-urban bias (the old good UHI) is a much stronger factor producing the artificial warming than micro-environment, as Anthony and others originally hypothesized.

    The experiment this guy performed was quite simple and easy to do. I am wondering, why nobody pays attention to this problem? If the entire rural network in the USA covering 1900-2010 period shows little warming overall and higher temperatures in 1930s and 1940s then the very idea of AGW is extremely weakened. One can then plausibly assume that since Arctic, Greenland and the USA rural networks show the similar pattern there is no much reason to believe in different “calculations” and adjustments made by NOAA, NASA and CRU for the rest of the world. Especially in the context of this post.

  9. Street
    Posted Jan 31, 2010 at 11:23 AM | Permalink

    Because it’s not ‘peer reviewed’….
    If it is peer reviewed, well then it wasn’t published in a ‘reputable’ journal….
    If it was published in a reputable journal, well then the study was ‘flawed’….

    I think it’s pointless to wait around for GHCN to update the stations they dropped. Not only are they not resourced (so they say), but they have no motivation to do so. It appears they were very selective when they decided which stations to keep updating. Bringing the other stations back in will only expose the biases of the current record.

    Now, I’m assuming many of these stations are available by web these days? Perhaps we need another project like to collect all this ‘lost’ station data and set up feeds to a central database…..

  10. Posted Jan 31, 2010 at 1:23 PM | Permalink


    it is not even necessary to launch any new extensive campaign similar to previous Anthony’s campaign. That is the most astonishing thing here! It is quite sufficient to go to the GISS website, and download their official temperature data – data of rural stations that have been kept working all the time since the beginning of the 20th century. And you automatically see, in the 10 minutes time, that something is really fishy there, because the rural stations (that SURVIVED all changes during the 20th century) give you a completely different pattern of warming and cooling then the urban ones (and for that matter, different than the official GISS product as well), as well as that the overall amount of warming was practically negligible in the EXISTING rural network! Why nobody pays attention to that?

    Instead to waste months and years to extensively monitor the micro-siting, station by station, and try to disentangle Hansens’s adjustment procedures for urban stations, why not simply point out that Hansens’s own rural stations show NO WARMING? If global warming is a climatic trend, then it must be fairly visible in the rural network. Since it is not visible, then, well – there is no “global warming”, but only UHI in the USA at least. The end of story.

  11. Posted Jan 31, 2010 at 1:32 PM | Permalink

    I think that John Christy already published a paper pointing out the complete absence of warming in the rural network of California, as well as in Alabama.

  12. Rob R
    Posted Jan 31, 2010 at 1:34 PM | Permalink

    RE David White (first comment).

    NIWA is still reporting monthly and annual data for more than 50 climate stations in NZ. The data is freely available via the cliflo database which you can get to on the NIWA website.

    It is a mystery why this data can’t seem to make the leap offshore into the supposedly primary global databases.

    By the way there is nothing unusual about the Hokitika station (which I have photographed). It is at a small regional airport with the usual building, carpark and runway issues, but it is at least out in the opern about 100 m away from most structures and at least 30 m from the runway. The record goes back to the 1860’s with only one significant station move from “Hokitika South” to “Hokitika Aero” around 60 years ago.

    There is little warming at Hokitika since the mid 1950’s.

  13. Street
    Posted Jan 31, 2010 at 3:49 PM | Permalink


    That is one line of attack and proving the extent of UHI is certainly important. However, with the scarcity of data (geographically), we still end up with alot of ambiguity. What it comes down to is this: we can poke holes in GHCNs methods and results all we want, but in the end, they can just ignore us. If you want to calculate world surface temp from thermometers, they are the only game in town.

    Producing an independent and transparent dataset is the only way to change the game. I don’t think the project would be all that massive if we just focus on aggregation of data from available sources with appropriate metadata – the same thing GHCN is supposed to be doing.

    In the end, we’d have a reliable dataset that allows us to analyze the real surface temperature trends. We could do world-wide what the Russians did with their dataset – definitely prove that the GHCN dataset is biased and inadequately small for surface temp tracking. We could move beyond poking holes in their work to producing competing scientific results.

    At some point we have to sieze the initiative and provide an alternative to these institution that have been captured by the warmists. I’m sure with a solid cadre of volunteers, we can out-do NASA’s 1/4 person effort, can’t we?

%d bloggers like this: