climate change wa

Has ACORN robbed Marble Bar of its world record?

Temperature adjustments by the Bureau of Meteorology to weather stations in the Australian Climate Observation Reference Network haven’t just cooled Australia’s history of very hot days (see No more extreme hot days in Australia than 100 years ago).

It seems ACORN may also have robbed Marble Bar of its world record heatwave back in the 1920s.

Australians have known for decades that Marble Bar in the north of Western Australia is world famous for the 160 consecutive days in which it recorded maxima at or above 37.8C (100F or a “century” in the Fahrenheit days).

Those relentless scorchers happened every day from 31 October 1923 to 7 April 1924, and nowhere else on earth is known to have recorded 160 century days in a row without a break.

Marble Bar is now a runner-up

Back in 2009, the BoM website had a Climate Education page explaining Marble Bar’s heatwave record. It was such a good page that the National Library of Australia considered it to be of national significance and has archived it for posterity.

In 2020, the BoM website still has a page that explains: Marble Bar, in the Pilbara, holds the Australian record for the longest sequence of days over the old century mark (100°F or 37.8°C). This occurred during the period from 31 October 1923 to 7 April 1924 when the maximum temperature equalled or exceeded 100°F for 160 days in a row.

The Australian Bureau of Statistics even adds an extra day to the world record:

marble bar heawave record

Australians have been taught about the Marble Bar heatwave world record for many decades :

marble bar record in newspapers

However, it seems that ACORN has cooled 31 October 1923 to 7 April 1924 so much that the Pilbara town can no longer boast that it had a world record 160 consecutive days above 37.8C.

ACORN 2, which is said to be a world-class homogenisation network, has reduced the 160 days to just 128 - from 1 November 1923 to 7 March 1924.

From 31 October 1923 to 7 April 1924, the dates during which the 160 days of 100F or more were recorded, there’s now 153 days at or above 100F.

Global records are sketchy but Wikipedia's Death Valley page states that "The greatest number of consecutive days with a maximum temperature of 100 °F (38 °C) or above was 154 days in the summer of 2001." This data is confirmed by the American Meteorological Society, which also references 134 consecutive days at Furnace Creek in Death Valley that were above above 37.8C during the summer of 1974.

154 days doesn't match 160 days but it's a lot more than 128 days, so it seems that America can now claim to hold the world record heatwave of consecutive 37.8C+ days at Death Valley - thanks to ACORN.

An Excel spreadsheet (499kb) with columns of daily maximum temperatures at Marble Bar from October 1923 to April 1924 in ACORN 1, ACORN 2 and RAW can be downloaded here, or the heatwave daily temperatures can be viewed in this pop-up.

Daily cooling adjustments

To quote the BoM from its archived Climate Education page: "The highest temperature recorded during the record spell was 47.5°C on 18 January 1924."

That’s correct in the original RAW temperature dataset, but ACORN 1 cooled 18 January 1924 to a maximum of 46.4C, and ACORN 2 restored a bit of dignity by nudging this highest temperature back up to 47.3C.

In late 2019, politicians were scorned for suggesting that the BoM adjusts temperature data to fit a global warming agenda, and SBS News reported that the bureau denied it has rewritten Australia's climate record.

The animation below uses the daily temperature datasets for RAW, ACORN 1 (introduced 2011/2012) and ACORN 2 (introduced quietly in early 2019 with no BoM announcement) to compare the number of days each year from 1910 to 2019 that Marble Bar recorded a very hot day (defined by the bureau as at or above 40C) :

very hot days in marble bar

This animation demonstrates that temperature data has been adjusted and that Australia’s climate record has been rewritten.

The slides show clearly that very hot days were far more frequent according to the RAW original thermometer observations in the first half of the 1900s, but ACORN 1 cooled many of these and ACORN 2 has trimmed them even further to create an upward trend in the occurrence of 40C+ days at Marble Bar since 1910.

Rain = clouds = fewer very hot days

The bureau’s archived 2009 Climate Page helps explain why Marble Bar had the world’s longest heatwave in 1923/24: "The town is far enough inland that, during the summer months, the only mechanisms likely to prevent the air from reaching such a temperature involve a southward excursion of humid air associated with the monsoon trough, or heavy cloud, and/or rain, in the immediate area."

Marble Bar averaged just 9.9mm of rain per month from November 1923 to April 1924, compared to the 1910-1964 Nov-Apr average of 43.4mm per month. Just 71.1mm of rain fell in 1924, compared to an average 325mm in 1910-1964. The town had 132 very hot 40C+ days in 1924, compared to the 1910-1964 average of 112.3.

So it seems that the frequency of rainfall strongly influences how hot it gets in Marble Bar and how often the town exceeds 40C. No surprise there. Rainfall is a proxy for cloudy days that keep temperatures below 37.8C or 40C.

The animation below shows the correlation between annual November to April rainfall at Marble Bar and the number of very hot days in the RAW, ACORN 1 and ACORN 2 daily temperature datasets :

very hot days and rainfall in marble bar

RAW very hot day (40C+) annual averages:

1910-1964 112.3
1965-2019 98.7

ACORN 1 very hot day (40C+) annual averages:

1910-1964 93.2
1965-2019 94.5

ACORN 2 very hot day (40C+) annual averages:

1910-1964 90.7
1965-2019 95.1

Rainfall November to April monthly averages:

1910-1964 43.3mm
1965-2019 56.5mm

The animation shows how the frequency of very hot days increased when rainfall and cloudy days were relatively sparse at Marble Bar in the early 1900s, and how very hot days decreased when rainfall and cloudy days increased from the 1970s. However, the correlation between very hot days and rainy days is ignored by ACORN 1, and even more so by ACORN 2.

Marble Bar had 6,178 very hot days of 40C or more from 1910 to 1964 in the original RAW observations, and ACORN 2 cuts this to 4,991 - a 19.2% reduction. This homogenisation suggests the bureau identified a significant recording error in the standardised equipment and/or Stevenson screen at Marble Bar in the first half of the temperature record, sufficient to diminish the ability of rain clouds and clear skies to keep the daily temperature below or above 40C.

ACORN 2 significantly reduces the frequency of very hot days in the early 1900s at Marble Bar, but both RAW observations and adjusted ACORN 2 maxima show the average temperature of the very hot days was 42.4C in 1910-1964 and 42.3C in 1965-2019.

As elsewhere, Marble Bar’s temperature history has been influenced by shifting rainfall patterns rather than CO2. ACORN homogenisation of that temperature history doesn’t change observations at the rainfall gauges since 1910, and is seemingly unaffected by the correlation between cloudy days and very high temperatures.


The cooling adjustment of Marble Bar’s early observations doesn’t just affect the frequency of very hot days. Below compares average annual maxima in the RAW, ACORN 1 and ACORN 2 datasets :

marble bar average temperatures

Average maxima

1910-1964 - ACORN 1 34.90C / ACORN 2 34.77C / RAW 35.45C

1965-2019 - ACORN 1 35.16C / ACORN 2 35.18C / RAW 35.25C

ACORN 1 warmed 0.26C / ACORN 2 warmed 0.41C / RAW cooled 0.20C

Average change per decade : ACORN 1 0.10C / ACORN 2 0.12C / RAW 0.03C

The reduction of very hot days in the historic records is typical of broader adjustments to maximum temperatures on all days in Marble Bar before the 1970s.

If rainfall isn't considered a reliable indicator of cloud cover, another way to gauge it is through the BoM estimate of the mean number of cloudy days averaged over 30 year periods at Marble Bar since 1921 (a cloudy day being when the mean of 9am and 3pm observations shows at least 75% of the sky has cloud) :

marble bar cloud cover

The months with greatest cloud cover increase had the largest decrease in average maxima, and the months with the biggest decrease in cloud cover had the largest increase in average maxima.

Note: 1921-1950 averaged 44.3 days of cloud per year but there were an average 29.8 days per year that recorded rainfall in this period. 1999-2019 averaged 72.1 days of cloud per year but there were an average 44.1 days per year that recorded rainfall in this period. This indicates that annual days of rainfall are an underestimate of how many days had substantial cloud cover each year, albeit indicative.

Not just very hot days

ACORN adjustments to historic daily observations affect not only the frequency of very hot days (40C+) but also what the BoM defines as hot days (35C+). The chart below shows annual hot 35C+ days at Marble Bar in the RAW, ACORN 1 and ACORN 2 datasets from 1910 to 2019, as well as annual rainfall :

marble bar hot days above 35C

RAW hot day (35C+) annual averages:

1910-1964 206.5
1965-2019 196.7

9.8 fewer hot days per year in 1965-2019 than 1910-1964

ACORN 1 hot day (35C+) annual averages:

1910-1964 196.3
1965-2019 194.2

2.1 fewer hot days per year in 1965-2019 than 1910-1964

ACORN 2 hot day (35C+) annual averages:

1910-1964 191.2
1965-2019 194.8

3.6 more hot days per year in 1965-2019 than 1910-1964

RAW hot day (35C+) annual average temperature:

1910-1964 40.1
1965-2019 39.9

Average hot days temperature 0.2C cooler in 1965-2019 than 1910-1964

ACORN 1 hot day (35C+) annual average temperature:

1910-1964 39.9
1965-2019 39.9

Average hot days temperature the same in 1965-2019 as 1910-1964

ACORN 2 hot day (35C+) annual average temperature:

1910-1964 39.9
1965-2019 39.9

Average hot days temperature the same in 1965-2019 as 1910-1964

Annual rainfall averages:

1910-1964 325.0mm
1965-2019 407.3mm

If the BoM still supports its 2009 Climate Education web page and believes that Marble Bar retains the heatwave world record because the original 160 consecutive days of 37.8C+ in 1923/24 were valid and accurate, it destroys the credibility of ACORN as a homogenisation process that persistently cools historic temperature observations around Australia.

If this isn’t the case because ACORN is a world-class network that produces accurate historic temperatures, Marble Bar can no longer boast that it holds the heatwave world record.

Data appendix

Urban heat has had no influence on temperatures at Marble Bar. At the 2016 Census, the town had a population of 634 living in 134 dwellings. In 1920 the Marble Bar population was about 1,200 and the town had no bitumen roads :

marble bar pictures

The table below presents Marble Bar's annual rainfall days, total rainfall, hot days, very hot days and maxima in RAW and ACORN 2 from 1910 to 2019 :

marble bar climate variables

Excel spreadsheet for table above

The charts below are sourced from the bureau's raw temperature hottest days each month per annum database (e.g. the hottest day recorded in July 1910 had a maximum of 28.3C and all other days that month were cooler). The averages of all the months each year show that from 1901 to 1959 the hottest days averaged 39.91C, and from 1960 to 2019 they averaged 39.76C - cooling 0.15C - with cooler hottest days from November to May and warmer hottest days from June to October.

marble bar hottest days

The table below is extracted from the BoM website and summaries the adjustments made to Marble Bar minimum and maximum temperatures since 1910, including the impact, causes and comparative stations used to justify the changes (within the actual temperature datasets, ACORN 2 average maxima were 0.92C cooler than RAW in 1910-1944 and 0.93C cooler in 1923/24).

acorn adjustments at marble bar

Sourced from the ACORN adjustments table above, the table below shows how the adjustments since 1910 take no account of other natural and artificial influences :

natural and artifical climate variables

Little is known about the circumstances of the Marble Bar thermometer and Stevenson screen prior to 1930. Marble Bar 4106 and the original station, Marble Bar Comparison 4020, were both reading temperatures throughout 2001 to 2005. 4106 was an automatic temperature probe placed in the same screen as 4020 from which manual observations were taken.

During this overlap from 2001 to 2005, the 4106 AWS had an annual average maximum of 35.3C and 4020 had an annual average maximum of 35.4C.

The veracity of both original RAW and homogenised ACORN temperatures since 1910 is uncertain due to undocumented thermometer, screen or location influences, as well as documented meta changes with questionable adjustments and influences such as area averaging based on neighbouring weather stations.

Note: The charts on this page are not produced by the Bureau of Meteorology but are based on the daily temperature and rainfall data provided by the bureau in its different datasets. Animation charts are free for use on any platform.

Top of page
Website home

Bookmark and Share

This website produced by Scribeworks 2008-2020


Very hot days annual

Very hot days seasonal and rainfall

Very hot days eastern Australia