The weather bureau might be underestimating Australian warming: here's why

Former prime minister’s business advisor Maurice Newman fired another attack at the Australian Bureau of Meteorology on Monday, arguing in The Australian that the weather bureau needs to be investigated for fiddling with the climate data that show Australia is getting warmer.

Technically known as “homogenisation”, the practice of removing biases from historical climate data has been well defended. In fact, you can do it yourself.

In his latest salvo, Newman claims that estimates of surface temperature trends by weather bureaus “diverge increasingly with satellite and radiosonde datasets”, and thus call into question the “integrity” of the surface data and the processing of those data.

This is a very easy claim to check, and interestingly it shows the weather bureau might actually be underestimating warming.

Surface v satellite

You can find yearly average Australian temperatures at the Bureau of Meteorology. I have plotted the mean annual temperature for Australia, expressed as difference from the 1979-2015 average, in the graph below.

Satellite data show a stronger warming trend. Neville Nicholls, Author provided

I have only plotted the data since 1979, since that is the first year for which satellite estimates of temperatures are available. The Bureau estimates of average annual Australian temperature for each year are shown as the thin blue line. The linear trend is shown as a thick blue line.

The satellite data for the temperature of the lower atmosphere, averaged across Australia, are available from the scientists at the University of Alabama at Huntsville. I have also plotted these data in the graph below (as a thin red line).

These are the latest version of the satellite data (version 6.0). As with the surface data, I have plotted the satellite data as differences from their 1979-2015 average. The linear trend of the satellite estimates of temperature over Australia is shown as a thick red line.

You can see immediately that the two graphs vary up and down quite similarly, but that the two graphs are indeed diverging. This is because the satellite estimates of Australian temperature show much stronger warming than do the surface temperatures measured by thermometers by the Bureau.

The Bureau data show warming at a rate of about 1.3°C per century, over the period 1979 to 2015. The satellite data reveal a warming rate of about 2.4°C per century over the same period.

Why the difference?

It is not surprising that the Bureau’s surface data and the satellite data are not identical. They measure two different things, and use very different data to do it.

The Bureau’s data are from thermometers at the surface. These data are adjusted to take into account possible sources of bias, such as the urban heat island effect or changes in the location of the thermometers.

The satellite data are from remote observations of radiation from the lower layers of the atmosphere, observed by a small number of satellites. These data have been adjusted for changes between satellites, changes in instrumentation, and even changes in the time of observation. (You can read a discussion of these adjustments here.)

But despite all these adjustments (to the satellite data and to the surface data), the two estimates of temperature averaged across Australia show quite similar variations between years, and both show substantial warming over the period for which we have satellite estimates.

But the stronger warming trend in the satellite data suggests that the Bureau, if anything, is underestimating the rate at which Australia is warming.

However, I suspect that the stronger warming shown by the satellite data in the graph above is incorrect.

In the graph below I’ve used a slightly older version of the satellite data still available from the University of Alabama.

Older satellite data are more similar to surface measurements. Neville Nicholls, Author provided

Using the older version of the satellite data (that has somewhat different adjustments made by the University of Alabama scientists), the similarity with the Bureau’s estimates of Australian average temperature is even more striking.

And the satellite data show a warming trend much closer to the Bureau’s estimate (although the satellite data still exhibit slightly more warming over Australia than do the Bureau’s surface measurements).

But whichever version of the satellite data we use, there is no evidence from the satellite data that the Bureau is overestimating the rate of warming, and there is some reason to believe, if we were to trust the satellite data, that Australia may even be warming (slightly) faster than the Bureau’s data indicate.