Graham Lloyd, The Australian, June 19, 2015
Better data handling and statistical methods and the use of pre-1910 temperature records would improve the Bureau of Meteorology’s national temperature data set ACORN-SAT, an independent review has found.
A technical advisory panel, brought forward following public concerns that the bureau’s homogenisation process was exaggerating a warming trend, said it was “generally satisfied” with BoM’s performance.
But it said there was “scope for improvements that can boost the transparency of the data set”.
Scientists who queried BoM’s management of the national temperature data said they had been vindicated by the report.
The review panel made five recommendations and said it was “not currently possible to determine whether the improvements recommended by the forum will result in an increased or decreased warming trend as reflected in the ACORN-SAT dataset”.
The independent review panel was recommended by a peer review of the Australian Climate Observations Reference Network — Surface Air Temperature, but it not acted upon until public concerns were raised.
BoM’s technical advisory forum said ACORN-SAT was a complex and well-maintained data set. Public submissions about BoM’s work “do not provide evidence or offer a justification for contesting the overall need for homogenisation and the scientific integrity of the bureau’s climate records.”
Nonetheless, the review report said it considered its recommendations for improving the bureau’s communications, statistical methods and data handling, and further regional analysis based on the pre-1910 data, would address the most important concerns.
David Stockwell, who raised concerns, said he was “very pleased with the recommendations”.
“They largely identify and address all of the concerns that I have had with the past BoM work,” Dr Stockwell said. “When implemented, it should lead to considerable improvements.
“The panel recommended strongly that the BoM communicate the limitations and it agreed that errors in the data need to be corrected and homogenisation is necessary, as I do, although it must be communicated clearly that the ACORN result is a relative index of change and not an observational series.”
The forum received 20 public submissions which questioned;
• The 1910 starting date (although some pre-1910 records are available) and its potential effect on reported climate trends;
• The treatment of claimed cyclical warming and cooling periods in the adjustment process and its effect on reported warming trends;
• The potential effects of site selection and the later inclusion of stations in warmer regions;
• The treatment of statistical uncertainty associated with both raw and homogenised data sets;
• The ability of individuals to replicate or verify the data set; and
• the justification for adjusting historic temperature records.
Bob Baldwin, the parliamentary secretary responsible for BoM, said the bureau would work to adopt the recommendations.
“We believe the forum’s recommendations for improving the bureau’s overall communications, statistical methods and data handling, and further regional analysis based on the pre-1910 data, will help address the main concerns surrounding the dataset,” he said.
Mr Baldwin said the report was an important part of ensuring the bureau continued to provide world-class information on the climate trends affecting Australia.
The review panel said its recommendations predominately addressed two key aspects of ACORN-SAT.
They were: to improve the clarity and accessibility of information provided, in particular explaining the uncertainty inherent to both raw and homogenised datasets; and refining some data-handling and statistical methods through appropriate statistical standardisation procedures, sensitivity analysis and alternative data filling approaches.