Deception Detection In Non Verbals, Linguistics And Data.


 
Special Report: The Vanishing Record: How the Bureau Quietly Deleted 274,000 Temperatures and Built Its Climate Data on Unpublished Files


The public cannot verify the ACORN-SAT dataset because the raw data it was built on is either missing, withdrawn, or never published.


In 2007, the Bureau sold a CD containing its raw temperature data. A colleague Geoff Sherrington kept that CD. We compared it, day by day, to what the Bureau currently publishes on its Climate Data Online (CDO) portal.

The result is staggering. Since 2007, the Bureau has quietly removed 274,722 temperature values from the public record.


 Discovery 1: The Bureau Quietly Deleted 274,000 Readings

- Most of these were flagged as "suspect" back in 2007, which is fair enough. But nearly 13,000 of them were flagged as "Y" (Quality Checked). The Bureau certified them as good data in 2007. Today, they are gone.

- The Bureau deleted 18 entire weather stations from the public record. Point Lonsdale Lighthouse lost 7,670 days of data.


 Discovery 2: 80,000 Readings missing from CDO, Used to Build the Climate Record

Inside the ACORN-SAT download, there is a separate zip file labeled "raw data and supporting information."

*This file contains the actual daily readings the Bureau used to build its official climate record. And it contains roughly 80,000 temperature readings that do not exist on the public CDO portal.*


- About a third of these values for Maximum temps were on the 2007 CD, and for Minimum Temps its half, meaning they were public once, but have since been withdrawn from CDO.

- Port Macquarie 1921–56, adjusted and published in ACORN-SAT, has never been on CDO or the CD.

The Bureau is using 80,000 data points to calculate its official climate trends that the public has never seen and cannot verify.

 

The Bottom Line

The Bureau of Meteorology has a duty to maintain a transparent, verifiable historical record of Australia's climate.

Instead, they have quietly withdrawn 274,000 readings from public view, and used 80,000 unpublished readings to build their official dataset.

This is not a clerical error. This is a systemic failure of data integrity.

The Bureau can fix this.
They can release the complete raw data (including the 80,000 unpublished values), explain why 274,000 values were withdrawn, and fix the backdating flaw in their homogenization algorithm.


Until they do, the ACORN-SAT dataset should not be used for policy or climate analysis without a clear disclaimer: This record is incomplete, unverified, and artificially warmed.


Thanks to Patrick Lloyd and his assistant Claude AI for analysis of Geoff Sherrington's CD 2007.

 


Special Report:
The Precision Collapse: 80,000 Days Where the Bureau's "High-Quality" Data Lost Its Decimal Points

The Bureau said "some stations at times" lost decimals.
The data shows 33 stations lost over 80,000 days—when the automatic system came in.


The Bureau of Meteorology claims its ACORN-SAT dataset is the highest-quality climate record in Australia, accurate to 0.1 C. But a forensic audit has revealed a massive data-handling failure.


For years, the Bureau's automated weather stations archived temperatures in whole degrees only, silently destroying the decimal precision. And the Bureau's quality control system didn't catch it.


 The Scale of the Loss

The Bureau has publicly acknowledged that "some stations at times" lost decimal precision in the early years of the automatic weather station era. But the data shows it was far worse than that.


Across 33 ACORN stations, there are 80,612 days where the temperature was recorded in whole degrees only. During these windows, the effective resolution drops from the promised 0.1 C to a blunt 1.0 C.


- Cape Otway (minimum): 3,342 days (nearly a decade).

- Katanning (maximum): 3,369 days.

- Cabramurra (maximum): 2,212 days.


These are not occasional rounded months. They are near-total loss of the decimal digit for years on end.


 The QC Failure

99.5% of these days are flagged as "Y" (quality checked). Why? Because the Bureau's QC checks for missing data, not missing precision. A value of 31 C passes the check, even if the thermometer is supposed to be recording 31.4 C.


The data loss begins exactly when the Bureau installed its Automated Weather Stations (AWS) in the late 1990s, and ends abruptly in 2003. This proves the data loss is a Bureau data-handling failure. Their systems were archiving whole degrees and throwing away the decimals for years, and they never told anyone.


 Why This Matters

BOM is correct that pure whole-degree rounding produces no systematic bias in the long-term mean. But it does affect:


- Threshold counts: Heatwave days, frost days, growing-degree days. This occurred during the 2002 droughts.

- Extremes and records: A 1 C grid cannot record a record-breaking hot day that is 0.3 C above the previous record.

- Day-to-day sequences: The flatlines and artificial runs that plagued the original audit.


 The Bottom Line

The Bureau's "high-quality" dataset contains multi-year periods where the effective resolution is 1 C, not the nominal 0.1 C. And it was never disclosed.


The Bureau's quality control system certified 80,000 days of lost precision as "quality checked." That is a fundamental failure of data integrity.



Special Report: The Manufactured Warming:
How a Flawed Correction Inflates Australia's Climate Trend

The Bureau's algorithm assumes no town ever grew around a thermometer. The data proves that assumption is wrong.
 

The Bureau of Meteorology has long claimed that its ACORN-SAT adjustments are necessary to correct for station moves and instrument changes. That claim is true. But a new forensic analysis has found that the method used to make those corrections is flawed—and it is artificially inflating the warming trend.


 The "Backdating" Flaw
When a weather station was in a town, it recorded the Urban Heat Island effect—the town grew around the thermometer, making it artificially hotter. When the Bureau moved the station to an airport, they measured the difference in temperature at the time of the move, and then subtracted that difference from every year going back to 1910!


This is the flaw.
Towns grow over time. In 1910, the town was small, and the heat effect was tiny. By subtracting the 1965 heat effect from 1910, the Bureau artificially cooled the past.
And when you cool the past, you manufacture a warming trend.

As the report puts it: "The Bureau's method assumes no town ever grew around a thermometer that moved."


 The Proof
The report proves this using three independent tests:

1.  The Drift Test: Comparing town stations to truly rural stations, the old sites were warming before the move. Immediately after the move to the airport, the drift stopped. This proves the site was getting hotter due to town growth, not climate change.

2.  The Moved vs. Unmoved Test: In the raw data, moved stations were warmer than unmoved ones (because of the Urban Heat Island). After ACORN's adjustments, the moved stations are now 0.12°C to 0.15°C *cooler* than the unmoved ones. The Bureau overcorrected.

3.  The Independent Check: An independent homogenization algorithm (ACMANT) was run on the same raw data. It added only half as much warming as the Bureau's ACORN algorithm.


 The Bottom Line
ACORN's adjustments add +0.69°C per century to the national minimum temperature trend. The backdating error accounts for +0.12°C to +0.02°C per century. This error is carried by about ten stations, including Hobart, Launceston, Walgett, and Halls Creek.

This is not a clerical error. This is a systemic methodological failure.


The Bureau is not falsifying this data; they are applying a mathematical correction that is mathematically wrong.

Until the Bureau fixes this flaw, the ACORN-SAT dataset is not a faithful record of what the thermometers recorded. It is a record that has been artificially warmed by a flawed algorithm.


Thanks to Patrick Lloyd and his assistant, Claude AI for the data.



Special Report: The 200-Value Claim: Falsified by the Bureau's Own Data

They didn't fix it, they deleted it. They didn't scan the network. And there are thousands more duplicates than they admitted.



Above: Cairns 1950 copy/paste into different month, after BoM said, "there are no further duplicate months."


What follows is the result of a special investigation into the Bureau of Meteorology's temperature and rain record. This level of independent, line-by-line verification was practically impossible just a few years ago. 

The findings you are about to read are not based on theories or speculation. They are based on the data itself, 334 million daily rows of temperature and rainfall data.


When The Australian newspaper exposed the Bureau of Meteorology’s temperature record in October 2026, the Bureau rushed to defend itself. 

They issued a statement: The problem was tiny—"fewer than 200 values in a database of seven million." They had "confirmed and fixed" the errors. And they promised there were "no further duplicate months in the entire bureau network."


These claims are falsified by the Bureau's own data.

A forensic audit of the Bureau's complete temperature and rainfall data has revealed the truth. The Bureau didn't fix the data; they deleted it. And the copy-paste problem doesnt only persist in 2 stations—it is a systemic across many stations in the network.


Here is the evidence.


The Temperature "Fix": Deleting the Evidence

The audit originally identified 978 copied days at 15 stations—nearly five times the Bureau's "fewer than 200" claim. But the Bureau's "fix" for the Sydney and Charleville stations was not to correct the copied values. It was to delete whole months of data.


*   At Sydney Observatory Hill, 571 values were removed from raw. 

*   At Charleville, 56 values were removed.

*   At Sydney it deleted both months of 7 copied pairs, so genuine observations went with the copies.


But the Bureau said there were "no further duplicate months." That is also falsified by the Bureau's own data. A network wide search found whole months still copied at Cairns, Toowoomba, Charleville, Weeaproinah, Inverell, and Bathurst Gaol. 


The Rainfall Bombshell: 1,337 Copies Left Untouched

If you thought the temperature copies were bad, the rainfall data is worse. 


A comprehensive search of the Bureau's daily rainfall records found 1,337 copied stretches at 1,191 stations, covering 61,877 days. These are not just historical relics—60 copies are from the year 2000 or later, the latest in 2026. 


*   The Scale: You can find copies in every state, every decade. There are 268 copies in the 1970s alone.

*   The Bureau's "Fix":  1,334 of the 1,337 rainfall copies were still published exactly as they were. The older records, where almost all the copies sit, had not been touched.

*   The Quality Flag: Over 97% of these copied stretches carry the Bureau's "Y" flag, meaning "Quality Checked." The Bureau's QC system is certifying copied data as genuine.


The "Late-1990s Batches": The Bureau's Database Error

Perhaps the most damning finding is the "late-1990s batches."

At 22 stations in six states, rainfall from 1998, 1999, and 2000 appears under dates in 1980, 1990, and 1997. The stations share exact offsets, several at a time. This is not an observer's mistake. This is a systematic database operation at the Bureau's end.

As the report states: "Whether overwritten or filled in, what fell on those dates is no longer in the public file."

The Bureau overwrote the historical rainfall records with data from other years, and then certified the results as "Quality Checked."


The Bottom Line

The Bureau's temperature and rainfall records are not a faithful historical account. They are a patchwork of copy-pasted values, deleted months, and overwritten databases.


The Bureau claimed the temperature copy-paste problem was limited to "fewer than 200 values." The forensic audit found 978 copied temperature days at 15 stations. But if you thought that was bad, look at the rainfall. A separate search found 61,877 copied rainfall days across 1,191 stations.


The Bureau's quality control system missed 62,000 days of copied data, and 97% of it is still flagged as "Quality Checked."

They claimed they "fixed" it. They didn't.  They removed 12 temperature copies, left 32 behind, and also left all the rainfall copies in place.

The Bureau's quality control system is not a quality control system. It is a rubber stamp that certifies copied data as genuine observations.


Until the Bureau conducts a full, independent audit of every temperature and rainfall record—and explains why it deleted months instead of correcting them—the public cannot trust any of its data.


Thanks to Patrick Lloyd and his assistant, Claude AI for the data.


Special Report: The Rutherglen Proof: When the Algorithm Overrides the Thermometer

A single chart proves homogenisation can invent cooling

The Bureau of Meteorology has a standard defense when its climate data is questioned: "The adjustments are necessary to correct for station moves and instrument changes. We use neighboring stations to verify the corrections."


For many adjustments, that defense holds up. 

But at Rutherglen in January 2007, it collapses completely. 

The chart below is the perfect visual proof of how the Bureau's algorithm can manufacture a warming trend where the raw data shows none.




The "Statistical" Adjustment

In January 2007, the Bureau applied a -0.61°C cooling adjustment to all data before that date at Rutherglen. Because there was no documented site move or instrument change, this was labeled a "Statistical" adjustment. The algorithm simply decided a break existed.


But the neighbors say otherwise.


Reading the Chart

 The Top Chart (Raw data): This shows the temperature difference between Rutherglen and its 10 nearest neighboring stations. Look at the blue vertical line (January 2007). The black line is smooth. The neighbors show a step of just-0.01°C. Nothing happened. The weather at Rutherglen was perfectly normal relative to its region.

 The Bottom Chart (ACORN-SAT): Look at the orange line at the same vertical line. There is a massive, artificial step. The Bureau's algorithm applied a -0.61°C cooling adjustment to the past. 


An adjustment the Bureau’s own method can’t justify

By cooling the past by 0.61°C and leaving the present alone, the Bureau artificially created a warming trend at Rutherglen that does not exist in the raw data or the neighboring stations. 


This is not a correction. This is the algorithm manufacturing a warming adjustment by cooling the past, despite the raw data and the neighboring stations proving no such change occurred.


The Bottom Line

The Rutherglen adjustment is one of 1,575 adjustments the Bureau has made. If the algorithm can invent a -0.61°C cooling step where the neighbors show nothing, how many other "statistical" adjustments are equally unsupported? 


This single chart proves the Bureau's algorithm is not just correcting data; it is capable of overriding the raw evidence and manufacturing trends. 

Thanks to Patrick Lloyd and his assistant, Claude AI for the data.

Possible Is Not Probable: Media Watch and the Palmerville Flatline


On 23 October, Media Watch contacted me. They said they were investigating claims from my BoM audit report. 

They also told me that “all climate scientists” disagreed with me. 

That is a big claim. But they did not tell me they who they were, whether they had read my report or only the media, whether they had examined the data, or whether they had done any calculation at all. 

Instead, I was told that a 17-day temperature flatline at Palmerville, to 1°C precision, “was possible”.

Possible here meaning that it could be a natural weather phenomenon.

Possible? 

Yes. A stuck thermometer is possible. A copied database block is possible. Missing data infill is possible. 

But that is not the question. 

The question is: is it plausible as real weather, or is it a data integrity failure?


At Palmerville, from 21 April to 7 May 2004, the maximum temperature was listed as 31°C every day for 17 consecutive days. That is not a minor oddity. That is a flatline. 


Media Watch and the Bureau want to hide behind the word "possible". But in statistics, "possible" is a remarkably low bar. It is possible to flip a coin and get heads twenty times in a row. It doesn't mean it happens, and it certainly doesn't mean it's a valid explanation for a pattern in an official dataset.


Let’s look at the actual mathematics of this "possibility".


I calculated under normal conditions at 1 C precision to account for dropped decimals, using the station's actual historical variance and autocorrelation, the probability is roughly 1 in 90 million. The point, though, is that whether it is 90 million, 1 million or 600 million matters not, it is extremely improbable.


Under normal conditions, the Palmerville flatline is effectively not a natural weather event.

But it is quality flagged as Y by BoM and passed to Acorn.
This is the crux of the issue and why it is a data integrity fail.


Media Watch never told me who this anonymous scientist was. They didn't tell me if he had read the report, examined the data, or done a single calculation before dismissing it. 

But here is the ultimate irony.

You do not need to be a climate scientist to count duplicate months. 
You do not need a PhD to find 17 identical temperatures in a row. 
You do not need a climate model to ask why a database contains copied blocks. 

You just need arithmetic, an audit trail, and a willingness to ask the obvious question.


Media Watch offered the word "possible". The data offers the math.

Possible is not probable.  

Probable is not proven.  

And a flatline is not a measurement merely because it appears in a government database.




© ElasticTruth

This site uses cookies from Google to deliver its services - Click here for information.

Professional Blog Designs by pipdig