No, Australian climate records are not being erased

^{By David Campbell}

Social media posts have misleadingly claimed that Australia’s national weather bureau is hiding evidence of historical temperature extremes.

“Since 1995, the Australian Bureau of Meteorology has erased many of Australia's hottest temperatures which occurred before 1910,” says a post, viewed more than 16,000 times on the social media platform X.

The post, which was also shared to Facebook, highlights an apparent discrepancy between maximum temperature records published by the Bureau of Meteorology and those published by the Australian Bureau of Statistics (ABS) in its official yearbooks.

The weather bureau’s website shows the highest temperature recorded in Australia was 50.7 degrees Celsius at Onslow Airport in Western Australia in January 2022, and at Oodnadatta Airport in South Australia in January 1960.

Meanwhile, the ABS’s 1995 Year Book names five instances before 1910 where maximum temperatures were the same or hotter.

They are:

• Bourke, New South Wales (52.8C in 1877)
  
• Mildura, Victoria (50.8C in 1906)
  
• Cloncurry, Queensland (53.1C in 1889)
  
• Winton, Queensland (50.7C in 1888)
  
• Eucla, Western Australia (50.7C in 1906).
  

Why are these records excluded?

A spokesperson for the Bureau of Meteorology told RMIT Lookout that its analysis of Australia’s temperature record begins in 1910 and that earlier observations could not be relied upon “due to the suspect quality and availability of temperature data”.

“These Mildura, Bourke, Cloncurry, Winton and Eucla observations are excluded from Bureau extremes tables because they are from non-standard instruments and/or measurement practices pre-1910,” the spokesperson said.

Australia became a nation in 1901 and the Bureau of Meteorology was formed in 1908, meaning early temperatures were recorded by colonial governments and even amateur weather watchers, the bureau’s website says. 

“There is little or no information available regarding the types of instruments used, their calibration and exposures. This makes it difficult to align these early-era observations with the official record that commenced in 1910”, it says.

Introducing a national approach

In the 1880s a device called a Stevenson screen was introduced to Australia to help record temperatures. It is essentially a box with slats that shades and allows air to circulate around a thermometer contained inside. The outside is painted white to minimise heat absorption. 

Shortly after the bureau was formed, it introduced a nationally consistent approach to recording temperatures by installing the Stevenson screen in locations around Australia. By 1910 the screens had been installed in all weather stations, with a few exceptions.

Before 1910, “thermometers were located in various types of shelter, as well as under verandas and even in unheated rooms indoors,” the ABS website says.

“Because of this lack of standardisation, many pre-1910 temperatures in Australia are not strictly comparable with those measured after that date, and therefore must be used with care in analyses of climate change within Australia,” it says.

A weather bureau spokesperson confirmed to RMIT Lookout that the “extremes tables” on its website exclude non-Stevenson screen data and undigitised data.

Why older records might be exaggerated

Climate scientist Ailie Gallant, an associate professor with Monash University’s School of Earth, Atmosphere and Environment, told RMIT Lookout it was “very reasonable” for the temperature series to exclude measurements not taken using Stevenson screens because other screens “give a high temperature bias compared to the Stevenson screens when measured at the same location,” she said

If there is no information to show what shading was used to take the measurement, “there is every chance it was measured in the sun [or] some other non-compliant shade”, Dr Gallant added, noting one historical example where a thermometer had been placed in a tin can.

Such measurements grossly overestimate the air temperature because of inadequate ventilation or warming directly by the sun's rays, she said.

David Karoly, a University of Melbourne professor who formerly led the CSIRO’s Earth Systems and Climate Change Hub, agreed that excluding non-standardised observations was “entirely appropriate”. He too said the observations “may be biased” due to poor thermometer placement or lack of ventilation and would only be useful “if more information or analysis can determine what sort of instrument shelter or location was used".

A review of ABS yearbooks since 1995 shows that, from 2006 onwards, the Cloncurry observation named in the social media posts carried a footnote warning that it was “known not to have been measured in a Stevenson screen”.

The Bureau of Meteorology website notes that another of the five examples was recorded at “a particularly poorly exposed site”. Having compared January 1986 temperature data from Bourke, NSW, to that of other nearby sites, the bureau concluded that the readings were “significantly exaggerated, most likely due to the exposure of the instrument”.

The historical record

The bureau’s website says that its historical data is continuing to be digitised and that monthly data can be accessed on its website in most cases where the daily figures are not yet available.

However, the bureau spokesperson told RMIT Lookout daily data for some stations was available via its Climate Data Online database. This includes the 1906 entry for Mildura (weather station 076077), named in the social media posts.

Experts from the University of Melbourne have published a detailed analysis of observations recorded in southeast Australia between 1860 and 1910. After correcting for bias, they found that increases in temperatures since 1960 represented “the largest and most significant trends in southeastern Australian temperature in the last 152 years”.

Australia’s climate has warmed by an average of about 1.47 degrees Celsius since 1910, according to the 2022 State of the Climate report. 

In 2024, the World Meteorological Association declared 2023 the warmest year in the 174-year observational record.