The History of a Weather Station in Western Australia: Roger Underwood

I have recently made a superficial analysis of temperature trends at York, Western Australia, the nearest weather station to my place at Gwambygine. York is approximately 100 kms inland from the Indian Ocean, on about latitude 32.

The weather data for York is interesting for two reasons: (i) there has been a continuously reporting weather station here since 1877; and (ii) in 1996 the station was relocated from the rear of the Post Office in the centre of town to a farm paddock two kilometres away. The Bureau of Meteorology (BoM) publishes separate weather data for each site. Thus it is possible to compare mean daily max and min temperatures for the period 1877-1995 with those for the period 1996-2006. (The 2007 data has not yet been published).

I found that the mean daily maximum temperature for the period 1996-2007 was 0.6 degrees warmer than the mean daily temperature over the previous 119 years. However, the mean daily minimum temperature for the decade 1996-2007 was 1.0 degree cooler than for the previous 119 years. This suggests that on average, overall, York has been marginally colder since 1996. In any case there is no evidence of “catastrophic warming” for this site.

Without the actual data (which is not freely available), it is impossible to test the statistical significance of these differences. In any case, I consider it more likely that any differences are due to the relocation of the weather station. The old Post Office site was surrounded by high stone walls and heat-absorbing/retaining brick buildings and car parks, whereas the new site is beyond the town in an open paddock.

I wrote to the BoM for comments on my analysis. In reply they presented a graph showing annual maximum and minimum temperature trends with a running 11-year mean combining both weather staions for York for the period 1910 to 2006. These reveal a roughly 1 degree increase in annual maximum temperature over the last 96 years and a roughly 0.3 degree increase in annual minimum temperature.

I wrote back to the BoM and asked why they chose 1910 as the starting point for their analysis. Their interesting reply was:

“A change in the type of thermometer shelter used at many Australian observation sites in the early 20th century resulted in a sudden drop in recorded temperatures which is entirely spurious. It is for this reason that these early data (pre-1910) are currently not used by the Bureau in monitoring climate change.”

I would be interested if anyone could refer me to an authoritative paper on the history, quality and anomalies in Australian weather records and the influence of the re-location of weather stations. I am aware, for example, that the Perth Western Australia weather station has been re-located at least three times over the years, each time to an area with an obviously different microclimate. How is this taken into account in determining real long term trends? And are there other key sites in the historical record for which temperature records have been artificially influenced by changes to thermometer shelters, or other technical aspects.

Roger Underwood is a former General Manager of the Department of Conservation and Land Management (CALM) in Western Australia, a regional and district manager, a research manager and bushfire specialist. Roger currently directs a consultancy practice with a focus on bushfire management. He lives in Perth, Western Australia.

‘Déjà Vu on the ABC’ by Roger Underwood was voted one of the best Australian blog posts of 2006.

14 Responses to The History of a Weather Station in Western Australia: Roger Underwood

  1. Ender April 17, 2008 at 11:05 am #

    Roger – “Without the actual data (which is not freely available)”

    I got 10 years of raw climate data for the whole of WA for $60.00. You only pay for the time of the person extracting the data and that is only a nominal cost.

    Why did you think that it is not freely available?

  2. roger April 17, 2008 at 12:26 pm #

    In using the term “not freely available” I meant “not available for free”. BoM did advise me, when I asked, that I could purchase their weather data. As it turns out, I am not so much intersted in the temperature trends as in the mechanics of collecting temperature data, and how this has been artificially influenced over the years by technology or weather station relocation, and whether or not these things are affecting the outcomes of trend analysis. Perhaps I have missed discussion of this issue over the years, and the whole subject is Old Hat.

  3. cinders April 17, 2008 at 12:29 pm #

    Another thought provoking post from Roger, which made me have a look at the York Post office climate averages available at
    Apparently there is a web site with these type of statistics for every station in Australia, although the spreadsheet of raw data was a bit disappointing as I could graph all years average tempretures.
    Clearly the history of the tempreture record and relocation of the site is an important variable that must be considered in looking at the statistics.
    After all wasn’t it Mark Twain and/or Benjamin Disraeli who gave us the knowledge of “There are lies, damned lies and statistics”

  4. Mr T April 17, 2008 at 12:32 pm #

    Jennifer, I have been somewhat concerned over the use of dialy minina and maxima to determine the extent of ‘warming’. I know it’s pretty much all the data we have so that’s why they use it, but I fear that it will not actually show the ‘real’ warming.
    We know from the physics that the addition of CO2 will lead to some warming. It’s how much that is the question.
    So as the Earth has a pulsed heat source (diurnal variation from the sun) what we would expect to see is the surface warming quicker (due to more CO2 avaialable to absorb IR) and cooling down slower (as the IR released won’t necessarilly move out into space but cab be re-absorbed by the more CO2 available). I don’t think the day time temperatures can actually get much hotter, due to limits on the amount of IR absorbed. So the daily maxima and minima shouldn’t have an enormous change (a few degrees at most), but what will change is that it will be warmer for longer.
    I wonder how it would be possible to measure that?

    It’s not the ‘warming’ that will be catastrophic, but the subsequent changees to climate. For York this will probably mean a diminishing Winter rainfall and slight increase in Summer rain as high pressure systems in the Great Australian Bight stay further north in the winter and deflect cold fronts south of Western Australia. There is evidence of this in the decreasing rainfall patterns across southwest WA.

  5. gavin April 17, 2008 at 1:56 pm #

    Roger: “I have recently made a superficial analysis of temperature trends at York”

    I guess there is still a lot we don’t know about your weather station Roger. As a former tech I would like to know the following information about the temperature records before feeling confident in analyzing various periods in the station history.

    1) Who made the instrument?
    2) Who calibrated it?
    3) Who installed it?
    4) Who read it?
    5) Who checked it?
    6) Who authorised the records?

    Perhaps a more general case is to ask about procedures at airports, ports and other stations versus post offices, colleges etc across the states.

    Let’s introduce something on the crafty side of calibration. From first principles and without frequent reference to say national standards we could merely resort to dipping our thermometer in ice water then boiling water from time to time to ensure zero and range remained relatively unchanged. We could also get a mate up the road to bring his “calibrated” thermometer along for a comparison in the shade.

    Big instruments were easy. I once made semi portable manometers using steel tanks, precision glass tubing and a bottle of mercury. The attached scale could be hand scribed on a strip of Perspex. Zero adjustment was achieved by simply moving the scale against the liquid column. One percent accuracy over full scale was a matter of knowing G and a few minor factors like the difference in coefficients for Perspex and glass.

    I also made lots of other gauges including dial thermometers with metal bulbs & long capillaries. These were mostly liquid/vapour non linear devices. Each had to be hand calibrated by squeezing then sealing off the liquid exhaust at around mid point on the prescribed dial. Accuracy there was about the thickness of the pointer in most cases. Overall and at best say 2-5%.

    For mobile calibrations, I carried several Hg thermometers safely stored (tool box tray) in their original fibber tubes for a degree of shock protection. These selected individuals came after examining many lab instruments and finding only a few did not have bead separation after long periods of storage. Ordinary thermometers frequently grow gaps, 2 or 3 in some cases.

    NOTE: Tapping thermometer bulbs is not recommended!

  6. gavin April 17, 2008 at 2:16 pm #

    At first glance Wiki notes on mercury spills and recovery are a bit alarming. Picking up blobs with an eye dropper is one line of attack. Adding another liquid (water, kero, metho, white spirit etc) also helps avoid bead separation around dust particles. Light oil on the surface prevents most vapour rising.

  7. Ender April 17, 2008 at 2:17 pm #

    roger – “In using the term “not freely available” I meant “not available for free”.”

    Sorry about that. With all the tinfoil hat people that post here I thought for a minute that you meant to say that the BOM would not give you the data.

    BTW I found this at the BOM that has some of the information:

    and if you google “stevenson screens” you will get several links to how meteorogical instruments are mounted.

  8. Luke April 17, 2008 at 2:23 pm #

    We’ve been over some of this before.

    York doesn’t appear in BoM’s climate change network.

  9. Ian Mott April 17, 2008 at 2:42 pm #

    The past 140 years data for Rockhampton Qld are neatly bisected by a change of location in the late 1930s. The old one at the post office and the new one about 7km away at the airport. The Bom publishes both sets and, surprise, surprise, the mean temp in the older series is about 0.7C warmer than the recent series.

    The new series paints Rockhampton as a ‘hot-spot’ of global warming but a mean based on the total record shows no such thing.

    I have made numerous references on this blog to the lack of integrity in using daily max and min to derive a daily mean. The proper weighting for the temperature during the hottest hour of the day is only 1/24th of the sum of hourly records. Ditto for the coldest hour. And neither is a suitable proxy for the temperature at dawn and dusk. It is the equivalent of deriving an annual mean by taking the highest daytime maximum in summer and the lowest nighttime winter minimum. It may well be a mathematical mean of the two numbers but if the aim is to gain information for important decision making then the more data that feeds into a mean, the better.

    My earlier analysis of the UK decadal means made it clear that most of the recorded warming came in the form of milder winter nights and earlier springs and later autumns. None of which pose a threat to polar ice cover, glaciers, sea levels or the survival of any species.

    Similarly, if the impact of high CO2 levels is only manifest in lower night time minima and warmer mornings then we certainly need to find out. The current assumption that GW will produce hotter summers, extreme cyclonic activity and extended drought is already exhibiting some significant holes.

    And it follows that climate models that are unable to determine how warming could be spread throughout the seasonal and diurnal variation areclearly not up to the task demanded of them.

  10. Mr T April 17, 2008 at 3:11 pm #

    “if the impact of high CO2 levels is only manifest in lower night time minima and warmer mornings then we certainly need to find out. ”

    This is certainly my belief, however this is still very serious. It most certainly will change the ‘climate’ and affect everything form agriculture to breeding cycles. It’s a serious change and one that if it happens quickly will lead to the extinction of a lot of flora and fauna (see the PETM extinction, and other large extinction events). A lower minima at night and longer periods of ‘hot’ during the day will have serious consequences for the poles and glaciers. Also consider evaporation rates… being warmer longer means evaporation increases and plants and animals require more water which will (locally in the SW) become scarcer.

    The modelling can’t show this yet, give it time.

  11. Mr T April 17, 2008 at 3:19 pm #

    Ian, “My earlier analysis of the UK decadal means made it clear that most of the recorded warming came in the form of milder winter nights and earlier springs and later autumns. None of which pose a threat to polar ice cover, glaciers, sea levels or the survival of any species. ”

    I think you need to research this more deeply. I think you are missing some immportant details about what happens if you make winter, spring and autumn all warmer…

  12. Roger April 17, 2008 at 3:37 pm #

    Gavin’s questions about the qualtiy, calibration and maintenance of weather stations, and about the capability and monitoring of people taking weather readings are rather worrying. It reminded me that many years ago I worked for a government agency in a small country town and we ran the official weather station as a sideline to our normal business. The “readings” were made at 9 am and 3 pm every day of the year, and sent of to the weather bureau. In all the years I was there I can never remember any of our instruments being officially re-calibrated or updated/replaced. A large tree partly shaded the Stevensen Screen. Measurements during the week were made by our female admin assistant, but on the weekends there was a roster of staff. We had was no training, and no-one checked our work. Thinking back, it was all very casually done. It frightens me to think that today there are climate analysts using this data and drawing portentious conclusions from it.

    Bob R

  13. Luke April 17, 2008 at 3:56 pm #

    Tyre kickers might do well to see that BoM do actually understand some of these issues.

    A few emails or phone calls might help instead of pre-supposing what is or is not done for detailed analyses.

  14. Mr T April 17, 2008 at 4:14 pm #

    Luke, but you’re missing the point.
    All this sciency stuff should be obviously apparent without actually having to think it through or read anything. The BOM obviously have something to hide if it isn’t immediately apparent… It’s obviously the communists…

Website by 46digital