Grose Michael R., Boschat Ghyslaine, Trewin Blair, Round Vanessa, Ashcroft Linden, King Andrew D., Narsey Sugata, Hawkins Edward (2023) Australian climate warming: observed change from 1850 and global temperature targets. Journal of Southern Hemisphere Earth Systems Science

A neat little paper that looks at estimating an 1850-1900 baseline for Australian average temperatures.

The Paris Agreement sets to limit global temperature change from pre-industrial conditions. There were few thermometers back then, so it’s not practical to measure change in this way. The 1850-1900 period is often used instead as a sort of “eh!-close-enough” estimate for global temperature. Hawkins et al. looked at a number of lines of evidence and concluded that the period is probably a little warmer than the actual pre-industrial was.

Of course, once we have this target, people start asking questions like, “How much has Australia warmed since 1850-1900?” This paper by Grose and company sets out to answer that question. I note here that the Paris Agreement only talks about global temperatures in this context. The local stuff is a consequence of making sense of that and accrues additional uncertainty.

Australia is fortunate in having a relatively long instrumental temperature record. It’s unfortunate in that the earliest part of the record is affected by thermometer screen biases and the biased thermometers are largely confined to the southeast of the country, otherwise the task would be considerably easier.

They use six different methods

  1. What data there are (back to 1860)
  2. Paleo data from things like stalactites, corals and what not)
  3. global datasets
  4. climate models
  5. bias corrected climate models and…
  6. a statistical model based on global mean temperature1

The existing data and the paleo data underestimate the trend in the 1910-present instrumental series. In so far as we believe that that is the truth2, then these estimates underestimate the actual overall warming. The paleo record is for a wider area so incorporates marine areas too and thus the underestimate is to be expected (it also has less variability, which again, we’d expect). The global datasets don’t indicate much change between 1850-1900 and 1910-1930, but one wonders what’s informing the reconstructions at that point. As previously noted, the earliest data are biased and data from further away – ocean data and data from neighbouring land areas – will perhaps have similar problems to the paleo record. Before about 1880, the local estimates from the global datasets diverge. Models come in with a difference around 0.2°C between 1850-1900 and 1910-1930 but as you might imagine, it varies from model to model.

Sticking this all together they choose 0.2°C (from the models because models are complete) as a reasonable central estimate and then use a range around that bounded by the estimates from the models and global temperature datasets. This comes out as 0.2±0.2°C.

Combining this with the existing Australian temperature record back to 1910 gives estimates of temperature change since 1850-1900 (and also gains a small uncertainty associated with the instrumental series to bring it up from ±0.2°C to ±0.3°C). They find that Australia has warmed 1.3-1.6 times more than the global mean (land and ocean) and about the same as the global mean for land areas. In both cases, and relatively speaking, Australia warmed faster in the real world than it does in climate simulations.

It’s a nice thing to know, but I’m a little puzzled as to what to do with this information. There’s more in the paper about relating Australian temperatures to different global warming levels, but it goes over my head. More reading needed3.


1 The description of this statistical regression model is basically, “we did a regression”. I’d grumble more about this, but (a) they don’t really make use of the results and (b) it’s an industry standard description.

2 Probably as far as the uncertainty estimates allow.

3 Always more reading needed.