Australian Met Office Faces Storm Over Weather Data Inaccuracies
A complete review is being undertaken of the network equipment and BoM’s temperature data handling. It is the biggest public scandal for BoM since furious debate was sparked three years ago over its treatment of historic and contemporary temperature records to compile its new homogenised national temperature data series.
Bush meteorologist Lance Pidgeon had hard evidence it was cold near Goulburn in the early hours of July 2 this year because his cold water pipes froze, bursting in the wall and breaking the toilet.
To be certain, Pidgeon checked the Bureau of Meteorology website and saw the temperature had plunged to minus 10.4C.
“But then I saw something truly bizarre,” Pidgeon says. The temperature recording on BoM’s website adjusted itself to minus 10C and then disappeared from view.
That early morning observation by Pigeon has forced BoM to admit shortcomings at an undisclosed number of cold weather locations within the automatic weather station network.
A complete review is being undertaken of the network equipment and BoM’s temperature data handling. It is the biggest public scandal for BoM since furious debate was sparked three years ago over its treatment of historic and contemporary temperature records to compile its new homogenised national temperature data series known as ACORN-SAT.
For an agency that screams from the rooftops every time the mercury nudges to the slightest record high, losing a half a degree Celsius here and there at the lower extremities is a pretty poor look.
It has led again to claims that ideology has gotten the better of good scientific practice.
The bureau’s chief executive, Andrew Johnson, is adamant the bureau has not been tampering with the lower temperature record. But the Pidgeon affair, which moved past Goulburn to also raise questions about the Snowy Mountains thermometer at Thredbo Top and also Tasmania, has left BoM open to claims it is working to ensure the dominant narrative of rising temperatures due to climate change remains intact.
In a statement, the bureau says it “holds the integrity of our weather observations and climate data to the highest possible standards”.
“The bureau rejects allegations aired in some media outlets that it has sought to tamper with temperature data,” BoM says.
Johnson says BoM is also replacing as a priority hardware in four locations other than Goulburn and Thredbo Top. They are Tuggeranong in the ACT, Butlers Gorge and Fingal in Tasmania and Mount Baw Baw in Victoria where the climate record indicates it might be reasonably expected to reach below minus 10C.
William Kininmonth, a former head of BoM’s National Climate Centre, says he is puzzled that after decades of service the bureau now claims the automatic stations are not fit for purpose at some cold weather locations.
“My understanding is a lot of testing was done before the automatic weather stations were installed in all different sorts of conditions,” Kininmonth says.
“Why this is happening now, unless they have changed their manufacturers who they get them from, I don’t know.
“I would have thought minus 10 would have been well within their scope. They take automatic weather stations down to Macquarie Island and Antarctica, I can’t understand this at all.
“It seems to me they have some sort of automatic collection system in the computer; once the data comes in, they check on it then. I don’t know why they would be doing that at that stage.”
BoM strongly rejects any suggestion of manipulation.
Nonetheless, the handling of temperature data is a red-hot issue with claims and counterclaims dogging the world’s premier meteorological agencies including the National Oceanic and Atmospheric Administration and NASA in the US, and Britain’s Met Office.
Reports of the latest controversy at BoM have quickly and widely circulated around the world.
But for now BoM has escaped a full independent evaluation of its temperature data handling. The bureau will conduct an in-house review with assistance from at least two outside participants.
BoM narrowly escaped a forensic audit of its temperature handling methods for its national temperature data set ACORN-SAT after concerns were raised. Anomalies highlighted at the time included missing data and changes to temperature trends at some stations and areas from cooling to warming after homogenisation in 2014.
Documents released under freedom of information reveal former prime minister Tony Abbott’s department had canvassed using a taskforce to carry out “due diligence” on BoM’s climate records at the time.
But the push was defeated by then environment minister Greg Hunt, who sought to protect the reputation and integrity of the institution. “It is important to note that public trust in the bureau’s data and forecasts, particularly as they relate to bushfires and cyclones, is paramount,” Hunt wrote at the time.
In the context of the latest BoM controversy, Environment Minister Josh Frydenberg says: “Public confidence must be maintained in the integrity and effectiveness of data collection and quality control of data observations.”
But he has stopped short of demanding an independent, public inquiry, insisting instead on independent outside participation in BoM’s in-house probe.
What cannot be controlled is a small army of largely amateur enthusiasts such as Pidgeon who pore over the millions of lines of BoM’s temperature data made public by the high-profile institution. Pidgeon, a freelance radio technician and citizen scientist, has found instances where thermometers accurate to a tenth of a degree were adjusted by as much as two degrees. Original cooling trends in temperature records were being revised to warming trends.
There are vast areas of the nation where identical temperature readings have been recorded over long periods and places where the daily minimum temperature has exceeded the maximum, changes that defy logic.
Records of extremely hot days before the turn of the century have been erased, in one celebrated case simply because a diligent worker had taken the observation on a Sunday, which was outside of usual practice.
Anecdotes and evidence of manipulation have fuelled a deep mistrust of BoM’s national data record among some people, as exists in other countries.
But in-house reviews and expert panels have not produced a smoking gun.
Hunt was able to head off a forensic examination of the bureau’s national data set with the appointment of a technical review panel that has taken a collegial rather than adversarial approach.
The panel is headed by a former CSIRO official Ron Sandland, and includes people with expertise in statistics and data handling. Now in its third year, the panel has praised BoM’s methods and supported the necessity of homogenisation as a tool.
In its first report, the panel concluded that ACORN-SAT was “a complex and well maintained dataset”. The forum also concluded “that there is scope for improvements that can boost the transparency of the dataset and increase its usefulness as a decision-making tool”.
It called for greater transparency, more co-operation with statistical experts and a more open system to convey uncertainties inherent in the homogenised data set. A key recommendation was that the bureau seek to quantify the uncertainty for both raw and adjusted data for temperatures.
It said prioritising the provision of explicit standard errors or confidence intervals should further inform the bureau’s understanding and reporting of trends in all temperature series.
In its follow-up report last year, the forum welcomed progress made by BoM but repeated its call for greater emphasis on uncertainty. It recommended “targeted and active consultation with expert statisticians about the bureau’s work plan on understanding and communicating uncertainty”.
“This work should recognise the disciplinary differences between meteorologists, climatologists and statisticians in describing and estimating uncertainty, with a view to optimising the bureau’s approach by adopting appropriate methods from each of these disciplines.”
Contrary to BoM’s strident defence of ACORN-SAT, the main argument against the bureau is not whether homogenisation was justified.
There is wide agreement that referencing temperature recording sites with surrounding areas is a valuable tool to build a wide area average.
It is also logical for BoM to adjust individual site readings at its more than 600 sites in the automatic weather station network over time to compensate for factors such as changes in recording equipment, relocation of the reading device or other material changes.
The major concern has been the secrecy of BoM’s methods, the value judgments involved where changes are made without documented evidence of site changes or other factors, and the inability of those outside the bureau to replicate the adjustments.
Concerns have also been raised when homogenised temperature trends have diverged sharply from the raw, observational data.
In her introduction to the Institute of Public Affairs book Climate Change: The Facts 2017, Jennifer Marohasy says it is generally stated that without homogenisation, temperature series are unintelligible.
But, she says, this is disputed by research that clearly shows there is a very high degree of synchrony in all the maximum temperature series from Victoria beginning in January 1856 and ending in December 2016.
“The individual temperature series move in unison suggesting they are an accurate recording of climate variability and change,” Marohasy says of work undertaken with Jaco Vlok from the University of Tasmania.
“But there is no long-term warming trend. There are however cycles of warming and cooling, with the warmest periods corresponding with times of drought,” Marohasy says.