Australian Maximum Temperatures 1910-2010

Ken Stewart, March 2011

In this analysis I have looked at maximum temperatures across Australia including 2010.  Why Maximum temperature?  Well, that’s the temperature we talk about when we think of how hot it was today.  As well, maxima should be least affected by Urban Heat Island effect, and should respond to El Nino/ La Nina conditions.

Before I go any further, let me agree once and for all that 2001-2010 was definitely the warmest decade we have records for.

So, here is an alternative analysis of Australia wide temperatures- with a few surprises.

My data source was the raw maxima for the 100 non-urban High Quality sites, and from other nearby sites when  records needed to be combined (typically up to 20 km away, sometimes up to 35km)  as published at Climate Data on the BOM website.  I later included data from the 34 Urban sites as well.

I used the original (and much stricter) criteria of Torok and Nichols (1996) and only used stations with at least 80 years of data,  and similarly Della-Marta et al.’s requirement to have stations with at least 1 year of data overlap if it was necessary to combine records into a splice.  Consequently, a lot of sites were excluded (29 non-urban, 5 urban) because they did not have enough data, or because no true comparison could be made between modern and historical data.  What remains is a better quality record but which still is less than perfect.

A surprising number of sites had no 2010 data yet, because 1 or 2 months (usually December) had data not yet quality controlled ( over two months later).  Possibly the lower than usual temperatures flagged a warning that manual checking was needed.  I estimated annual data by finding the average of daily maxima for the whole year.

What’s With W.A.?

It quickly became apparent that West Australia behaves differently to eastern states.

This may be partly due to the data bug in WA data discovered by Chris Gilham and apparently not yet fixed.

Or perhaps we should regard West Australia as having a completely different climate system.

Certainly West Australia is not affected as much by the La Nina/ El Nino effects.

Here are the results.

71 Non-urban sites Anomalies, all of Australia:

Notice- the cool years 1917 and 1956; the shifts up about every 20-25 years; 2010 was cooler than any year since 2000, but still  0.11C above the 1961-1990 average.  The 101 year trend is  about 0.65C .

Here is BOM’s time-series graph of maximum anomalies:

Note 2010 is below the 1961-90 average.  My figures don’t agree with this probably because of the sites I excluded because of poor data.  The trend is similar- 0.7C.

East vs West

But look at this:- Non – urban sites excluding WA:Notice that the 2010 anomaly is now -0.22C, and that 1956 is cooler than 1917.

WA alone:

Wow!  2010 was definitely very hot in WA- anomaly of +1.04, just pipping 1994.  And 1968 is only just warmer than 1917.

City vs Country

Now include Urban sites in the mix- in fact, all Australian sites, urban and non-urban (  100 sites).  I should point out that I swapped Melbourne’s Moorabin Airport for Melbourne Regional Office to avoid UHI.

Note the step ups in climate, and a trend of about 0.5C.
Now excluding Western Australia:

Finally, here are the graphs for Urban sites only- remember, BOM excludes these from their climate analyses because of their urban warming signal.

Hmmm… as we’ve seen before, urban sites are warming less than non-urban- in fact hardly at all (less than 0.2C).  And apart from 1956, not greatly different to non-urban.  2010 anomaly 0.12 compared to 0.11.

Beach vs Bush

Now, here is a comparison of coastal versus inland ( more than 50km  from the coast) sites:  

Coastal sites (0.65) are warming faster than inland (0.35), inland sites fluctuate up and down more violently, and inland sites were much cooler in 2010.


2010 was definitely not a warm year.  Rain and cloud were associated with cooler temperatures in inland Australia.

Maxima do  not rise steadily, but in a series of step ups about 20 – 25 years apart.

There was a distinct decadal spike beginning in 2002-  coinciding with drought conditions across much of the continent.

The first decade of the 21st Century is definitely the warmest we have records for.

The inclusion of urban sites reduces the trend.

Inland sites are warming less than coastal sites, but also were cooler in 2010.  They are more volatile, warming faster and cooling faster.  Proximity to the sea moderates temperature change.

West Australia has a distinctly different climate from the rest of the country.

There is no reason for the Bureau not to include urban sites in their climate analyses.

The next few years will show whether the warming trend is continuing.

13 Responses to “Australian Maximum Temperatures 1910-2010”

  1. val majkus Says:

    Ken, great work. Just wondering is the BOM’s ‘raw data’ just that or are there adjustments, do you know?
    I’ve sent a link to Warwick Hughes

  2. Geoff Sherrington Says:

    From a background in geochemistry, we used to look for peak anomalies rather than smooting the. In some years of looking at annual temp graphs from various parts of the globe, some years have consistently higher maxima than surrounding years. Among them are 1914, 1940, 1970 and 1998.

    Of these, 1970 is the lest frequent anomaly from place to place. 1998 and 1914 would tie with the most common high peaks. I’ve never really looked at minima.

    These 4 peaks are about 28 years apart, but one cannot deduce much from such sparse data.

    I mention this because the maxima that you show on various graphs include some of these spot highs – but in contrast to global mean temperatures, where 1998 is often very strong, we see little evidence here. So, the question is, was 1998 (or any of the other nominated years) made hot because the minima were hot rather than the maxima? If so, any ideas on mechanisms?

    • kenskingdom Says:

      Yes Geoff, these plots are for maxima only, and higher minima have an equal effect on the means. However while means for Australia non-urban sites do show a spike in 1998, this is less than 2002, and nowhere near as high as 1988. More study needed.

  3. Warwick Hughes Says:

    Ken – if you are using BoM “High Quality” data – then I think that is adjusted. IMHO not worth a cup full of warm spit.

    • kenskingdom Says:

      Hi Warwick. No- I use the “raw” data, i.e. before BOM adjustments. I wouldn’t touch HQ data in a fit. I excluded 34 out of 134 HQ sites (25%) because of their poor records for a start. I’m still not sure how raw the raw data is as well. This is just looking at raw maxima from long record sites and making a few inferences from them.

  4. Brian Gunter Says:

    I have done an independent analyses of temperature trends in Australia that may be of interest.

    My anlayses concentrate on 13 long-term non-urban stations and the results are similar to Ken’s. I analysed over the 101-year period, 1910-2010.

    Most temperature indices are increasing throughout the 101-year period, with some evidence of a cyclical pattern at some locations. Interestingly, summer daytime temperatures in inland northern Australia have been steadily decreasing over the past century. But, what is really important is that there is NO evidence of any acceleration (hockey-stick effect) in recent years. The present temperature trends are more-or-less the same as they were a century ago! It really makes one wonder how the BOM and CSIRO come up with such dire predictions when the historical data based on thermometers indicate otherwise.

    • kenskingdom Says:

      Excellent Brian! That is a much more detailed analysis than mine. The trends in summer max in the north and minima in the southern inland are very interesting.

  5. Australian Maximum Temperatures 1910-2010 | Cranky Old Crow Says:

    […] Australian Maximum Temperatures 1910-2010 Like this:LikeBe the first to like this post. […]

  6. Mike Jonas Says:

    I’ve only just come across this (from WUWT via JoanneNova).

    Ken (if you’re still there) – it might be a good idea to re-classify the BOM HQ sites yourself. After looking at the WA sites, I suggest it is a bit dubious calling the following HQ sites non-urban (as in: no UHI effect. Some are obviously urban, some are non-urban but have dubious development around them):

    Katanning Comparison 10579 10092
    Kellerberrin 10073
    York 10311

    In addition to those, there were 10 airports and a post office:
    Derby Aero 3032
    Broome Airport 3003
    Halls Creek Airport 2012
    Port Hedland Airport 4032
    Newman Aero 7076
    Carnarvon Airport 6011
    Meekatharra Airport 7045
    Geraldton Airport 8051
    Kalgoorlie-Boulder Airport 12038
    Bridgetown Post Office 9510
    Albany Airport 9741

    That’s 60% of the 25 WA HQ so-called “non-urban” sites. And 2 others have closed:
    Marble Bar Comparison 4020 closed 2006
    Southern Cross 12074 Ceased temp. obs. 2007

    Note – I haven’t taken google maps for all of them.

    The others are probably OK:
    Roebourne 4035
    Giles Meteorological Office 13017
    Rottnest Island 9193
    Wandering 10917
    Cape Naturaliste 9519
    Jarrahwood 9842
    Esperance 9789
    Cape Leeuwin 9518

    And I don’t know which stations have been moved at some time.

    For UHI, I understand it’s not the size of the urb that matters, but how it changes.

    • kenskingdom Says:

      Gday Mike

      Thank you. BOM classifies sites in towns with a population of less than 10,000 as non-urban, and admit they cannot satisfactorily account for UHI. All sites have some impact from their changed surroundings, it’s just a question of minimising it and allowing for it in any analysis. Generally the record is very messy.

  7. Mike Jonas Says:

    typo – space needed between Katanning.jpg and Merredin 10092

  8. ACORN-SAT: A Preliminary Assessment « kenskingdom Says:

    […] northern and inland sites have much larger swings than coastal and southern sites (see MAN analysis here and here), therefore any warming trend will be […]

Comments are closed.

%d bloggers like this: