Australian extreme temperatures are falling

The annual range of temperature extremes in Australia is reducing as the continent warms.

Annual temperature Range (MaxT -MinT) averaged over all 1800 stations in Australia compared to the temperature anomaly

The annual temperature range is the maximum recorded temperature for that year minus the minimum recorded temperature. The annual continental average is the area weighted average over all 1800 stations. This was a surprise, so I thought it could be an artefact of the changing distribution of stations. However the ACORN-SAT data, which is selected for its long term coverage, shows the same effect.

Annual extreme temperature range compared to temperature anomalies for ACORN-SAT stations.

Are maximum temperatures falling or are minimum temperatures rising? The answer seems to be both, but minimum temperatures are rising slightly faster than maximum temperatures are falling.

Area averages of annual a) Maximum, b) Minimum temperatures and c) Extreme temperature ranges for all 1800 stations in Australia (far-flung islands excluded)

So what is happening? Is this an artefact of some kind or can this be a real effect ?

The argument to always work with temperature anomalies when studying climate change is to remove any natural station temperature offsets due things like different elevations. Yet we  know that urbanisation changes temperatures in or near cities, and that this also increases over time. I have shown how this then results in cooler temperature anomalies before the baseline normalisation period. Likewise UHI affects temperature ranges.

During hot days convection dominates cooling from the surface, whereas cold clear nights cool mostly radiatively. Therefore an enhanced greenhouse effect is more likely to increase minimum temperatures than to increase maximum temperatures. Maximum temperatures on the other hand are more affected by any increase in pollution or aerosols thereby reducing temperatures by reflecting sunlight. These are anthropogenic effects unrelated to CO2.

Are there any specific examples where this is hapenning? Well here is one: Richmond (QLD) from the homogenised ACORN-SAT

Average Monthly & yearly temperatures from Richmond. The top trace are the maximum recorded temperatures each year and the bottom trace are the minimum.

So I think this is a real effect, although I can’t yet be sure there is not a coverage bias as well.


The overall warming of 0.8C to 1C since 1910 seems to be associated with a fall in temperature extremes. Minimum temperatures have risen slightly faster than the rate at which maximum temperatures have fallen.

This entry was posted in AGW, Australia, Climate Change and tagged , , . Bookmark the permalink.

7 Responses to Australian extreme temperatures are falling

  1. Bryce Payne says:


    You stated, “Maximum temperatures on the other hand are more affected by any increase in pollution or aerosols thereby reducing temperatures by reflecting sunlight. These are anthropogenic effects unrelated to CO2.”

    I initially presumed in the last sentence your actual intention was to speak in a restricted sense of effects of agents rather than sources of agents, but then realized that the statement is in fact a declaration of sources of agents, and, by default, their effects, and is fundamentally flawed in that regard. The global increase in CO2 is an anthropogenic effect no less than the other forms of increased pollution or aerosols you refer to. Indeed, the human exploitation and application of fossil fuel energy that has resulted in increased atmospheric CO2 has also, with just as certain a cause/effect relationship, caused the increase in other forms of pollution and aerosols. A more subtle and presumably important point that you seem to be getting at is that there are anthropogenic global cooling (e.g., sunlight reflecting) and warming (e.g., reduction of radiant heat loss) agents/effects and how the dynamic balance among them is what determines where the climate is headed.

    There is an interesting and disheartening policy implication of the narrowing max-min temperature differential and your explanation. If pollution and aerosols are providing a net cooling effect that could counter the warming effect of CO2, then why not increase pollution and aerosol levels faster? I presume there is no need to discuss the foolishness inherent in such approaches, which, by the way, are being pursued under various guises.

    • Clive Best says:


      I was just speculating. I have no firm evidence. However there is no doubt that humans are affecting and will likely continue to affect the environment in the future. Whether that be CO2 emissions, pesticides, industrial farming, over-use of fertilisers etc.

    • Ron Graf says:

      Hi Bryce,

      I think every scientific mind focused on the climate change issue realizes that humanity is inadvertently running an uncontrolled 150-yr experiment on the atmosphere.

      I believe the difference in those minds is that some were made up that the impacts were shameful and likely catastrophic before there is evidence to understand the effects. This bias is similar to the appeal of buying anything that is advertised “natural” on the label. Actually, it’s better yet if there is no label; only a tag that says “hand made by indigenous peoples.”

      Most of those chanting and carrying signs have no idea of actual climate history, or the effects of orbital, solar, volcanic tectonic uplift, ice albedo, land use, sea and air energy transport. All they know is that there is a clear 97% consensus that personkind has warmed the globe in some way to some degree unnaturally. Never mind the 2.5 million year-old Quaternary Ice Age we have been serendipitous sheltered from temporarily for the last 10K years.

      The US government’s climate assessment report last year was put out in draft form in a post on Judith Curry last year. One of the main points it contained was to expect continued temperature extremes to become more extreme in the coming years. I commented that is was absolutely wrong, which started a heated debate thread here.

      Thanks Clyde for exposing very relevant information. I understand that you are using the highest and lowest temperature reading from each whole year’s population of readings. This should be minimally affected by UHI since the coldest nights I think are clear as well as windy, the later erasing UHI effect. It will be interesting if we can see the average daily range, which would contain UHI effect. If the trend is steeper than the trend in annual extremes that would point to UHI contamination in the record.

  2. paulski0 says:

    Hi Clive,

    Could you confirm what particularly the third graph is showing? Based on the numbers what I think you’ve done is taken the highest daily (as opposed to monthly average) Tmax reading within a year at each station then averaged those together. Is that right?

    What you’ve shown for Australia would then be quite similar to what I’ve seen demonstrated for US stations. The best explanation there, as in Australia, then seems to be that it’s an artifact of changing technology and siting of thermometer stations – an inhomoegneity. You may counter that this effect is visible in homogenised records, which I’ve also seen argued with regards the US. However, those homogenisation methods are performed on the basis of averages, not extremes.

    A fairly common reason, particularly in a place like Australia, for inhomogeneity between past and current records is that older stations were less well sheltered from the influence of direct sunlight. The magnitude of bias will vary from day to day – a clear day will impose a large bias, an overcast day much less so. In standard homogenisation practice those variances in bias are averaged. That’s fine when we’re interested only in monthly or annual averages, but when we want to look at extremes a large part of the bias remains untouched.

    A key problem for explanations of this feature being real is that those would also require a similar decline in average Tmax, not just the extremes, but that isn’t observed. Tmax average is going up, both in the US and Australia. The best explanation for most of the effect appears to be a time-evolving non-linear bias which expresses itself most strongly on extreme hot days.

    • Clive Best says:


      Yes it is the annual extreme temperatures. So for example, Tmax is the record high temperature recorded in 1978 and Tmin is the record low temperature recorded in 1978 for one station such as Cairns. Range is then simply the extreme temperature change for that year ( Tmax(record) – Tmin(record)). Then I make the weighted average for these extreme values over all stations in Australia in exactly the same way as for temperature anomalies.

      Your explanation could well be true. However in that case, I would have expected to see a sharp transition as equipment is replaced, or when modern Stephenson screens are
      introduced. BOM actually state that they curtail past temperatures before 1910 precisely for this reason.

      “A key problem for explanations of this feature being real is that those would also require a similar decline in average Tmax”

      I assume you refer to monthly averages. Yes as far as I can work out everyone (NOAA, CRU, BEST) all use monthly average values of Tmax and Tmin to calculate Tavg, and then normalised temperature anomalies based on different baselines. All three are rising although Tmin(monthly) seems to be rising slightly faster than Tmax(monthly).

    • Nick Stokes says:

      “The best explanation there, as in Australia, then seems to be that it’s an artifact of changing technology and siting of thermometer stations – an inhomoegneity”
      Plus just defective readings. 1800 stations for Australia is a lot, and goes a long way down the quality hierarchy. Many would be post offices, where reading the thermometer is just one of the postmasters duties. The odd erratic reading won’t affect averages much, but it does affect extremes.

      • Clive Best says:


        That may be true but it is also pure speculation. The only way to tell for sure would be to ask postmasters to restart measuring temperatures and then compare those results with modern automated ones.

        This is what I get if I simply plot the daily (Tmax-Tmin) for Townsville (ACORN-SAT from BOMs website). Black is 30 day average, while red is annual average. Annual only shown when 365 consecutive days available.

        I chose Townsville randomly.

Leave a Reply