One of the most important principles in science is that, when making claims, apples are compared with apples, rather than oranges. What this means is that data collection should use consistent methods and instruments across different sites at different times.
In some cases, this isn’t possible: usually when dealing with historical data. But, when inconsistent methods and especially instruments are used, the different datasets must be reconciled. Therein lurks danger. Every time you adjust raw data, the potential for bias creeps in. At worst, blatant data massaging results.
Consider this analogy: you are tasked with measuring different lengths of timber. You start with a piece of string, stretching it from one end to another, and then measuring it off against a ruler. Then you swap to an old ruler with Imperial measurements. Later, you graduate to a metric ruler, but it only marks off at 5mm intervals. Finally, you make the last measurements with a high-grade steel ruler marked off at half-millimetre increments.
But you have to adjust all of those to millimetre-perfect measurements. It should be obvious where the biases creep in: how tight did you stretch the string? Should you allow, hmm, an extra millimetre more or less? How about converting Imperial to metric? When you measured with the 5mm ruler, did you tend to creep one way or the other when estimating?
No carpenter worth their salt would claim that the string is the same as the steel ruler.
Yet, this is exactly what we are asked to believe about historic temperature records. Instrumental records are sporadic, to say the least, prior to the 20th century – and even then, instruments and data collection methods have varied wildly.
So, when we are told that Australia is having record-breaking temperatures and more heatwaves than ever before, it’s important to keep the above analogy in mind. ACORN2 is the the Australian Bureau of Meteorology’s (BoM) long-term (meaning: post-1910) temperature record for Australia. ACORN2 is a third-generation dataset, following the raw data, then ACORN adjusted dataset (which was found to be plagued with errors).
ACORN2 is also a “homogenised” dataset: that is, the raw data has been adjusted. Which makes it subject to exactly the problems I described above: comparing not just apples with oranges, but the whole fruit basket.
Comparing the raw dataset with ACORN and then ACORN2 shows an obvious trend: each one leans warmer than the last. In fact, the BoM’s raw dataset shows a distinct cooling trend. It’s only when the data has been twice-adjusted that the alarming warming trend emerges.
That doesn’t necessarily mean that it’s false, of course. But, there are hints that the warming trend is a relic of changing collection methods and a particular bias in adjusting data.
Each time the BoM changed its collection method (i.e., a new, smaller type of Stevenson screen, metrication in 1972, and Automatic Weather Stations (AWS) in the mid-late 90s), the data immediately “warmed up”. One reason for this is that AWS are able to record near-instantaneous changes in temperature, while liquid thermometers are much slower to respond to momentary fluctuations. AWS are biased toward extremes.
Analysis also suggests that the 50-plus weather stations added until the late 1970s were added in hotter locations. The new stations recorded nearly 40% more very hot days than the older stations. Data from the older stations has also been more heavily adjusted. Comparing the raw data with adjusted shows that many of those original stations have been significantly cooled after adjustment. Overall, in fact, ACORN2 cooled the period from 1910-1963 by 0.2°C.
The inevitable bias in adjusting data need not be conscious or deliberate: there isn’t necessarily an overt conspiracy to hoodwink the public. On the other hand, the BoM has systematically hidden or disguised inconvenient data in other instances: a graph of “Very Hot Days” mysteriously disappeared from its “Climate Change – Trends and Extremes” page.
“If the BOM were a bank adjusting it’s own tax receipts, the Labor Party and Greens would be demanding a Royal Commission,” as scientist and blogger Jo Nova writes. If nothing else, its claims about “record temperatures” ought to be treated with the kind of deep scepticism which ought to a basic tenet of journalism – but of which the legacy media today seem all-but-incapable.