By Chris Gilham,
Did you know that Australia’s rate of temperature warming per decade since 1910 has increased by 23%?
No? Neither does the Australian public, despite the Bureau of Meteorology several weeks ago releasing a new ACORN dataset of daily temperatures over the past 109 years that significantly rewrites Australia’s climate history.
ACORN 1 (Australian Climate Observations Reference Network – Surface Air Temperature) was released in 2012. The bureau has released ACORN 2 dailies, a total revision that significantly increases the warming trend calculated from the average temperatures since 1910 at 112 weather stations across the country.
The BoM hasn’t yet issued a press release to announce ACORN 2 and there’s been no media coverage. However, the bureau has published a report explaining the differences between ACORN 2 and ACORN 1, the dataset that got it wrong.
In a nutshell, the rewritten dataset means the national “area averaged” maximum from 1910 to 2016 at the 112 stations increased by 0.116C per decade, up from the 0.090C per decade calculated by ACORN 1. The ACORN 2 minimum increased 0.130C per decade, compared to 0.109C in ACORN 1, and the mean temperature was up 0.123C per decade in ACORN 2 compared to 0.100C in ACORN 1.
That’s a 23% increase in the warming rate of Australia’s mean temperature since 1910 (28.9% for max and 19.3% for min). Compared to the per decade increase in unhomogenised Australian Water Availability Project (RAW) temperatures, it’s a 54% increase.
The bureau has been claiming for several years that Australia’s mean temperature has risen by 1C since 1910, and the 23% increase per decade will presumably mean it’s risen by about 1.3C. Don’t be surprised to see a slight increase in the southern hemisphere’s historic land area temperature trend.
But it gets worse. ACORN 2 estimates that from 1960 to 2016, maximums increased by 0.202C per decade (0.179C in ACORN 1), minimums increased by 0.219C per decade (0.148C in ACORN 1) and mean temperatures were up by 0.200C per decade (0.165C in ACORN 1).
So Australia’s mean temperature trend from 1960 to 2016 has had an extra 21% of heat written into it (26% compared to the original raw temperatures).
There are 112 ACORN stations, 57 of which have temperature observations since 1910 and daily datasets in ACORN 1, ACORN 2 and RAW.
Prior to and including 1980, ACORN 2 cools the original raw maximum at these 57 stations by an average 0.16C (25.02C > 24.86C), whereas ACORN 1 cooled raw by an average 0.03C (25.02C > 24.99C).
Prior to and including 1980, ACORN 2 cools the raw minimum by an average 0.45C (13.50C > 13.05C), whereas ACORN 1 cooled raw by an average 0.07C (13.50C > 13.43C).
Comparing annual maximums in the averaged first half of the record (1910-1963) with the averaged second half (1964-2017) at the 57 stations, ACORN 2 warmed 0.49C, ACORN 1 warmed 0.39C and raw maximum warmed 0.32C.
Comparing annual minimums in the averaged first half of the record (1910-1963) with the averaged second half (1964-2017), ACORN 2 warmed 0.71C, ACORN 1 warmed 0.51C and the unhomogenised raw warmed 0.39C.
According to the bureau report, there were 966 adjustments made in version 2 of ACORN (463 maximum, 503 minimum). This compares to a total 660 in ACORN 1. That’s not 966 days with their temperatures changed. That’s 966 sweeping changes that each affect consecutive days in different blocks of years among the 112 weather stations.
For example, ACORN 2 includes new metadata on weather station moves, particularly recent moves in eastern Australia, which cause a warming trend. The bureau found a coding error in ACORN 1 and the corrected data may cause a 0.1C difference in monthly temperatures for individual stations. Another ACORN 1 coding error was fixed, causing a 0.09C increase in maximum, minimum and mean temperature trends since 1910. Many stations had an ACORN 2 adjustment averaging -0.05C in maximum and +0.05C in minimum because of the shift from large to small Stevenson screens in the 1990s.
There were numerous other homogenised adjustments and it’s obvious that although plenty of the homogenisation in ACORN is justified, a substantial majority of new adjustments in ACORN 2 just happen to result in an increasingly warmer temperature trend since 1910 – as was the case in ACORN 1 when compared to RAW.
Measuring the long-term weather stations
It’s not easy figuring out how the bureau includes stations such as Western Australia’s Learmonth, which opened in 1975, to calculate temperature trends since 1910. However, 57 of the 112 stations were open in 1910 and have temperatures available since then to 2017 in ACORN 1, ACORN 2 and RAW, so some averages can be figured out over the 108 years.
1910-1963 – v1 24.98C / v2 24.83C / raw 25.03C
1964-2017 – v1 25.37C / v2 25.32C / raw 25.35C
v1 warmed 0.39C / v2 warmed 0.49C / raw warmed 0.32C
57 Australian stations 2000-2017 – v1 25.79C / v2 25.76C / raw 25.78C
1961-90 – v1 25.09C / v2 25.02C / raw 25.08C
The first decade, 1910-1919, averaged 25.00C in v1, 24.87C in v2 and 25.06C in raw
The final decade, 2008-2017, averaged 25.84C in v1, 25.79C in v2 and 25.84C in raw
Prior to and including 1980, ACORN 2 cools raw maxima by an average 0.16C (25.02C > 24.86C), whereas ACORN 1 cooled raw by an average 0.03C (25.02C > 24.99C)
1910-1963 – v1 13.38C / v2 12.98C / CDO raw 13.48C
1964-2017 – v1 13.89C / v2 13.69C / CDO raw 13.87C
v1 warmed 0.51C / v2 warmed 0.71C / raw warmed 0.39C
57 Australian stations 2000-2017 – v1 14.12C / v2 14.03C / raw 14.05C
1961-90 – v1 13.67C / v2 13.39C / raw 13.70C
The first decade, 1910-1919, averaged 13.43C in v1, 13.01C in v2 and 13.56C in raw
The final decade, 2008-2017, averaged 14.21C in v1, 14.14C in v2 and 14.16C in raw
Prior to and including 1980, ACORN 2 cools raw min by an average 0.45C (13.50C > 13.05C), whereas ACORN 1 cooled raw by an average 0.07C (13.50C > 13.43C)
A detailed analysis with charts of all 57 weather stations spanning 1910-2017 can be viewed at http://www.waclimate.net/acorn2/, including spreadsheet downloads with minimum and maximum calculations for each site.
Three wrongs don’t make a right
None of the three datasets, ACORN 1, ACORN 2 and RAW, is accurate because they all have various undocumented, unadjusted or questionably scaled influences such as surrounding infrastructure, rapid response times in Automatic Weather Stations, urban or airport heat islands, reduced smog, and a majority of Fahrenheit temperatures being rounded at x.0F before 1972 metrication.
For example, with AWS response times the bureau calculates these electronic thermometers, mostly introduced since 1996, only affected national average maxima by +0.01C and minima by between zero and -0.01C, but maximums by up to 0.08C in arid areas like Alice Springs. No adjustments for these artificial influences were made in ACORN 2.
About 60% of all temperatures recorded at the 57 long-term stations from 1910 to 1971 were rounded .0F without a decimal, an unknown proportion truncated rather than evenly rounded, and BoM testing confirmed a 0.1C artificial Australian mean temperature warming in 1972. ACORN 2 didn’t even consider a compensatory adjustment because ACORN 1 had already decided the warming may have been caused by major La Ninas and the heaviest rainfall and cloud cover in Australian history during the early to mid ’70s.
The bureau has suggested that the new ACORN 2 temperatures will replace the existing ACORN 1 data throughout its website early this year, and public announcements thereafter will presumably be based on the warmer revised dataset. The ACORN revision is sure to be promoted as a more detailed and accurate measure of Australia’s climate warming that confirms Australia is roasting more rapidly than thought, and the media will respond accordingly.
When the media and public learn about Australia’s new warmer climate history there are likely be interesting political repercussions, particularly with a Federal election due within months. The left wing should be licking its lips.
The ACORN 2 revision has been underway for several years but the question remains … if the world-class ACORN 1 temperature dataset has been self-evidently wrong for the past seven years, why should ACORN 2 be considered any more accurate or reliable?