The Australian Temperature Record-Part 3: Western Australia

Ken Stewart, June 2010

Western Australia, the largest state in Australia ( 33 % of Australia’s land area) is the source of much of Australia’s mineral wealth.  It is physically isolated from the eastern states, and during the politicking for the Federation of the colonies in the 1890s, the vote for Federation was only carried after the Trans Australian Railway was promised.  After seeing what the Bureau of Meteorology has done with their temperature figures, Sandgropers may well want to secede.

BOM develops its climate Trend Maps and Time Series graphs form 100 sites nationwide, comprising the High Quality Australian Site Network.  25 of these are in Western Australia. 

Here is a map of the sites:

Here’s the Trend Map

 And Time Series Graph:

As you can see, Western Australia has a warming trend of 0.09 degrees C per decade, or 0.9 degrees per 100 years.

 As part of my survey of the data from the whole High Quality series, I have analysed the Western Australian data.  I averaged maxima and minima for all stations at each site, then compared with the High Quality means.  A spokesperson for BOM has asserted that: 

“On the issue of adjustments you find that these have a near zero impact on the all Australian temperature because these tend to be equally positive and negative across the network (as would be expected given they are adjustments for random station changes).”

Here are the results. 

 44% warming bias!  Even rounding conservatively,  the warming bias is 28.57%.

Here is a Graph of the differences in trend at each location:

Notice that there is a strong positive bias.

This is another way of showing this bias:

The HQ data has been subjectively and manually adjusted, and makes no allowance for Urban Heat Island effect, apart from not including (some) Urban sites.

However a closer examination of the WA HQ dataset reveals even more basic flaws:

  • 4 of the 25 sites (Port Hedland, Roebourne, Geraldton, and Albany) were listed as Urban by Torok and Nicholls in 1996, and should therefore not be included, as BOM says that

“Stations classified as urban are excluded from the Australian annual temperature timeseries and trend map analyses. Urban stations have some urban influence during part or all of their record.”

  • 4 sites (Giles, Newman, Meekatharra, Jarrahwood) have many years of data missing.  High Quality sites are supposed to have a long record of at least 80 years, with few gaps.


The official Trend Map and Time Series Graph for Western Australia cannot be relied on as an accurate record of the Western Australian climate over the past 100 years.  This is because they are derived from the High Quality Annual Temperature Network.  This is plagued by subjective, manual adjustments that in many cases cannot be justified, sites with years of missing data, sites that should not have been used because of Urban contamination, and a large warming bias.  If this is High Quality, I’d hate to see Low Quality.  What does it say about the rest of the Australian temperature observations?

Progress report on the Australian High Quality Site Network:

Sites checked:             55 out of 100

Raw trend:                  +0.73 degrees C/ 100 years

High Quality trend:     +0.96

Average difference:    +0.23

Warming bias:           30.77%

No doubt the remaining 45 stations will show a large cooling bias!

Appendix:  WA Sites and comments


 Missing data for the first 47 years, and no sites around for hundreds of km!

 Halls Creek

Old Halls Creek and Airport sites

HQ adjustment seems to follow the airport and warms the older data a bit.

I spliced the two datasets by adding 0.7C to Old Halls Creek figures.

This graph shows BOM did the same but made the first 15 years colder.

Derby again has two sites.

Note the missing data.  The splice would show 0.6 C warming but the adjustment corrects this.


Note the four datasets (Bidyadunga is 93km away, Anna Plains is 163 km away) are very close for most of the period with a trend of   0.2C:

HQ adjusts this to produce

Port Hedland is listed as Urban by Torok and Nicholls and should not have been used, and you can see why:

An “interesting” splice with a small adjustment.

Roebourne was also listed as Urban but is still used:

Note the small cooling adjustment, but the airport shows the possible UHI effect.

Marble Bar, the hottest town in Australia. For 161 consecutive days to 20 April 1924 the (maximum) temperature in the town never dropped below 100°F (37.8°C).

The trend of 0.2C has been adjusted to… 0.9!  Where does that leave the record?

The dodgiest of the lot- BOM supposedly only use sites with a long continuous record (>80 years).  Newman is a mining town whose record starts in 1966.  How do you get a meaningful trend from that?   

Easy, splice it with Mundiwindi, 71 km away!  Then you get 1 degree of warming!




Geraldton (also listed as Urban):

A sensible splice!

Kalgoorlie-Boulder- Two diverging trends:

Adjusted vs raw splice:

Now a splice using earlier data minus 0.5C :

Gives a trend of 0.3C changed to 1.0C.

Southern Cross:

BOM doesn’t use the airport at all:

0.2 to 0.8!


1.2 to 1.1

Kellerberrin: Flat

Adjusted to 1 degree of warming!  But nearby Cunderdin and Corrigin show no such steep trend- perhaps 0.4.


From 0.1 to 0.8!  But nearby Northam is about 0.2.


Given an extra 0.4C of warming, though Narrogin is almost identical to Wandering.

Rottnest Island, off the coast from Fremantle:

From 0.65 to 1.2C.

Cape Naturaliste Lighthouse:

From 1.1 to 1.5C.

Cape Leeuwin Lighthouse:

From 0.6 to 1.0C.

Jarrahwood itself has only about 30 years of data, and has huge gaps.  How to make a trend from that?  Use Donnybrook instead, and Busselton Shire which has large gaps.

Why not call it Donnybrook?  Because Donnybrook has a rating of 4 (Poor) due to poor Stevenson screen siting during some of the record.  That’s how they get around that little problem.  Fortunately, the trend has been reduced by 0.4C.


Warmed again.


Albany is also Urban.  It also has two records differing by 0.7C on average.  Notice that Eclipse Island agrees with the early PO data.

Adjusted with an extra 0.3C warming.

Here’s a modified splice (-0.7):

Esperence has 3 records.  Note how the two main records have a similar trend, and the second is about 0.3 above the airport.

I reduced the second town record by 0.3 to match the airport.  Notice where the BOM adjustment matches this splice and where it is different- they do a very rough splice then warm the first few years.  The result is a 0.1C increase in warming, and a dodgy record.

13 Responses to “The Australian Temperature Record-Part 3: Western Australia”

  1. janama Says:

    Great work Ken. The HQ data record is total BS. I did all the NSW stations and came to a similar conclusion.

  2. Frank White Says:

    Well done. I had no idea BOM’s methodology was so ropy. Although “Harry” the programmer for the Hadley Climate Research Unit said the data was driving him round the bend. (See the Harry_README file released by the whistleblower.)

  3. kenskingdom Says:

    Frank White:
    Thanks. Yes I’ll be quoting from Harry-Read-Me in the final post (Australia as a whole). It’s a mess all right.
    I could only find a couple of the sites I need in the batch you emailed me. Do you have your analysis of the NSW sites in a form I could use, if so, may I? and where? Many thanks!!

  4. janama Says:

    No worries Ken – I’ll post the for you.

    Give me a day or so.


  5. Geoff Sherrington Says:

    These might help with some lesser known sites:

    Giles: 25 02 34S 128 18 08 E definitely rural
    The Giles Weather Station was established in 1956 by the Weapons Research Establishment (now known as the Defence Science and Technology Organisation), a division of the Department of Defence. The first weather observations were transmitted from the station by radio on 2 August 1956.
    The purpose of the station was to provide weather data for the UK atomic weapons tests at Emu Plains and Maralinga. It was also used to support the rocket testing program based at Woomera, being conveniently positioned at the edge of and about halfway down the testing range. Giles Weather Station was transferred from the Department of Defence to the Bureau of Meteorology in 1972.
    The station has a staff of four who run it on a six-month tour. The staffing is arranged by the South Australian Regional Office of the Bureau of Meteorology and the Observing staff come from various Bureau offices throughout Australia. The station has satellite and telephone communications facilities, and hydrogen gas is produced on site for weather balloon flights.”

    Esperance: Confusion.

    The BoM metadata sheet online gives position 33 49 48S, 121 53 31 E
    This shows a plausible set of instruments on Google Earth, off Fairfields St, in the middle of a wooded paddock. The airport at Esperance since 1979 has been some 20 km to the NNW of the city centre since about 1969-70. The airport surrounds are quite rural. Some records (shipping) are taken from the city site, perhaps even another one by the sea front. The Post Office site is quoted for temperatures before the airport was used, starting 1910. It is not clear which station provides official temperatures for what purpose from which site of about 4 possible. See
    There are discrepancies in lats and longs.

    One could go on and on.

    • kenskingdom Says:

      Thanks, thanks, thanks!
      But I tried togo back to get metadata on other stations but got a message “forbidden from accessing this page”. What’s the secret? How did you get in?

  6. Geoff Sherrington Says:

    Well, Ken, If I tell you then you’ll have to kill me. You do a search of “Australia BoM metadata sheet Esperance”. Or another town. I have not checked all the 100+ RCS, but I suspect that most get a mention. Unfortunately, the write ups have an enormous amount remaining to be useful and I do not know the budget or priority of the BoM to do this. I suspect that if you pay some money you can get more data for some stations. There is a greater available volume of recent data than earlier data, say pre 1950.

    The information on the changeover date from daily to half hour recordings is partially useful, but it does not give the day on which the system data recording was switched, or whether the records were tapered together or butt-joined.

  7. Calem Says:

    Last March, after seeing a lot of scrutiny about station locations in North America, I was thinking about where the actual data came from which BoM was using to show temperature change in W.A.

    Just from observation of a few graphs I had an inkling that maybe things had been exaggerated. These are the over lays I put together.

    I noticed that BoM has different stations for measuring annual
    mean temperature then what stations they use for daily and monthly

    There seems to be a few more used for the annual stations then daily.

    With what knowledge I have of math, using station locations I ended up estimating that according to their annual network, W.A has warmed at a rate of 1.24 deg C per Century since 1970 and, since, 1910 W.A has warmed at a rate of 0.99 deg C per Century. This is basically what BoM is also saying.

    (Note. Using the daily network, it comes out at 1.14 for since 1970 and 0.97 since 1910)

    I tried something different.

    I used the trend map instead. I’m not sure how this map is generated but I thought I would use the surface area map rather then the stations map to figure out the warming trend.

    Using Photoshop to count pixels this is what I came out with on my spreadsheet.

    W.A has warmed at a rate of 1.16 deg C per Century since 1970 and, since, 1910 W.A has warmed at a rate of 0.96 deg C per Century.

    (Note. At the time I had no idea that Port Hedland, Roebourne, Geraldton, and Albany should have been discluded like Perth was)

    Have I done a whole heap of math for nothing or are the numbers I’m coming up with realistic?

    Is the Map i used generated from the Annual so called high-quality network?

    Like anyone else, I’m really interested to see what the real trend is and your the only person I’ve seen who has even attempted W.A’s.

    I would really like to see what you, Ken, and everyone else who reads this thinks about the matter.

    • kenskingdom Says:

      Look Calem I haven’t got the time right now to go through this in depth, can you remind me again in a few weeks please? I have no idea how the trend maps are generated I’m sorry, only the time series.

  8. Calem Says:

    That’s ok, not a problem. Cheers anyway, I was just curious.

  9. Julian Braggins Says:

    Do you think it would be worthwhile to inform the Shadow Minister for Climate of your findings, pointing out perhaps that the Government is being fed either dubious, spurious or doctored temperature trends ? 😉

Comments are closed.

%d bloggers like this: