Almost without fail when discussions turn to the weather, media commentators or scientists will report that the particular weather phenomena they are talking about is the most extreme that has ever been observed or recorded, but the frequency of these reports (and even a basic understanding of statistics) leads to some serious questioning about the validity of these claims.
In early December, the Weather Bureau confirmed that Tasmania had its driest spring on record in 2015. In support of that claim, reference was made to the fact that weather records for Cape Bruny lighthouse and Bushy Park go back more than 140 years, and the amount of rainfall recorded in Tasmania in the spring of 2015 was the lowest ever recorded over that period. To compare current rainfall with all records for the past 140 years in making the 'driest on record' claim seems to be pretty reasonable and few would argue with that.
But difficulties arise in situations where a claim is made, but it is not clear what the particular climate or weather event is being compared with. Take the example of the Bureau's Annual Climate Statement for 2015. The first few sentences of the Overview section of the statement were as follows;
2015 was Australia's fifth-warmest year on record (national observations commence in 1910). Above average temperatures were persistent throughout the year, with several periods of record warmth. The Australian area-averaged mean temperature for 2015 was 0.83 °C above the 1961–1990 average. Maximum temperatures were 0.96 °C above average, and minimum temperatures were 0.69 °C above average; both the sixth-warmest on record respectively.
Looking at recent years more broadly, eight of Australia's ten warmest years on record have occurred since 2002. The 10-year mean temperature for 2006–2015 was the second highest on record at 0.53 °C above average (and just behind 2005–2014). Only one year in the past ten was cooler than average: 2011.
What is obvious from a careful reading of the text is that comparisons are being made with several different baseline temperature references, one being the record dating back to 1910 when records commenced, and the second being the average of the period from 1961-1990. Leaving aside arguments about issues like which weather stations are included and the 'heat island' question, it certainly seems appropriate to compare current year average temperatures with the long-term record and make a statement about how the 2015 average temperature compared with those 115 years of records.
However, the very next sentence then makes a claim about how much higher the 2015 average temperature was than the average temperature recorded over the period from 1961-1990, a different reference period. And it is not clear from the text included on the accompanying map whether the temperature deciles being referenced in that map are based on 1961-1990 temperatures, or the entire temperature record.
A quick look at the plot of Australian average annual temperatures (based on data published by the BOM) reproduced below, highlights why it is important to be clear about the baseline being used in any comparison.
The graph shows that temperatures over the period from 1961-1990 were more frequently lower (the red line is a five year rolling average) than the 100+ year average, or than was the case from 1910 to 1930, or subsequent to the year 2000. This leads to suspicion that the 1961-1990 reference period is being used to make the 2015 average temperature increase sound more significant that it would be based on a comparison with the long-term average temperature record.
The BOM may well have legitimate reasons for using the 1961-1990 period as a reference period, but the fact that it not clear what reference period is being used - especially in maps and graphs or in claims about 'highest' or 'driest' years - leads to questions about the legitimacy of the claims being made. It is also evident that the media does not bother with the fine detail of what benchmark is being used for comparison data.
A little more clarity and consistency, especially when it comes to maps and graphs, might help to remove a lot of the confusion and uncertainty around this contentious issue.