The Australian article (linked below) reports a forensic audit finding extensive anomalies in the Bureau of Meteorology's online "raw" temperature records, including duplicated blocks, statistically implausible identical sequences, and large gaps. The Bureau admits a small number of duplicated historical values, describes their effect as negligible, and rejects claims of deliberate manipulation.
Investigator Tom Berger (former computer engineer specialising in data-based deception analysis) and retired scientist Geoffrey Sherrington used pattern-recognition software on the Bureau's online temperature records. Key results included:
Blocks of readings copied across years and seasons. The clearest case was Sydney Observatory August 1907 minimum temperatures appearing identically in the August 1867 online record. A National Archives check of the handwritten Daily Meteorological Observations confirmed the 1867 online version was wrong.
Statistically impossible sequences, such as 17 consecutive days of identical daily maximums or whole weeks matching to one decimal place.
"Broken seam" patterns: 14 days repeated, a two-day gap exactly where source records were missing, then another 15 days of duplicates.
Large voids, including 647 consecutive days with no observations at an ACORN-standard station and 408 consecutive days lacking decimal precision at another.
More than 40% of flagged anomalies dated after 1990.
At Sydney Observatory, maximum-temperature adjustments were positive in every year from 1910 to 2020 (110 years of one-way upward changes with no net cooling adjustment).
Berger described the patterns as obvious signs of human intervention that any competent quality-control system should have flagged, calling them database-level duplications that persisted across systems and software versions rather than simple observer transcription errors. He argued the anomalies are so unlikely they equate to "winning the lottery every week for a year," that the dataset is no longer reliable for serious big-data analysis or machine learning, and that none of the findings required climate-science expertise, only data-integrity standards. Sherrington noted the problem appears network-wide across stations from tropical Queensland and the Northern Territory to Western Australia, southern Victoria, and Macquarie Island. Analysis of the 50 largest duplicate temperature-block events across 17 stations showed 31 involved copying from one calendar month into a different month, which imports the wrong seasonal signal.
The results were raised with the Bureau by independent Tasmanian MP Andrew Wilkie. Chief executive and director of meteorology Stuart Minchin thanked Berger for the effort, noted the value of public scrutiny, and stated: "The bureau rejects allegations of fraudulent or deliberate manipulation of Australia's temperature record, there is no evidence to support this assertion, and it has no basis in fact."
In answers to The Australian, the Bureau said it works continuously on the integrity, quality and transparency of its datasets. Its analysis of the relevant dates confirmed "a small number of duplicated historical daily temperature records, namely four post-1910 months relevant to ACORN-SAT." These account for fewer than 200 daily values in a dataset of more than seven million. The Bureau assessed the effect on Australia's long-term temperature record as negligible. It said it will correct the identified historical records before the next scheduled release (later in the month) and will continue normal quality-assurance processes. It did not explain how the duplications occurred.
Berger responded that the Bureau had "acknowledged a few trees but denied the existence of the forest," reducing a systemic data-integrity issue to a handful of errors while ignoring the broader statistical patterns, the copy-paste evidence, the identical-temperature runs, and the one-way adjustment bias.
The article frames the findings as raising questions about the Bureau's data-handling and quality-assurance processes, especially because long-term temperature series inform climate research and policy (including net-zero). Scientist Jennifer Marohasy, who has long examined Bureau adjustments, highlighted the Sydney Observatory case as involving an entire month in the series the Bureau presents as unadjusted, not an obscure single day. The chance of two Augusts 40 years apart matching on all 31 minima to a tenth of a degree is described as nil, and the Archives check confirmed they did not match in the handwritten originals.
The piece stops short of claiming the anomalies reverse Australia's overall warming trend; its emphasis is on data integrity, the presence of non-independent measurements presented as independent, and the adequacy of the Bureau's quality controls relative to the standards expected in other scientific fields. The Bureau's position is that the confirmed duplicates are few, limited in impact, and will be fixed, with no evidence of fraud. Of course; how could I have thought otherwise?!
https://www.theaustralian.com.au/nation/bureau-of-meteorology-admits-errors-after-audit-reveals-duplicated-historical-temperature-data/news-story/cadebb66bd1f4a8b40f9af1b6ff0b1a3