How accurate is the official temperature history for your city or town?
Statistical analysis of 3 years of maximum temperature data for Brisbane airport shows the temperatures recorded from the probe in an automatic weather station are significantly different from the mercury thermometer. This contradicts claims by the Bureau’s Director, Andrew Johnson, that measurements from these two instruments are equivalent. The probe has been the official measuring instrument since November 1996 with these temperature values incorporated into national and international databases used to calculate climate trends. In the most extreme cases, temperatures recorded by the probe were 0.7 degrees Celsius hotter than the mercury. Analysis across the three-year period, shows the probes record higher 41 per cent of the time, the same 32 per cent of the time, and lower 25 per cent of the time.
The editorial in The Weekend Australian explains:
Given the high level of public interest in record temperatures as well as the significant financial stakes involved, the Bureau of Meteorology has a duty to be transparent about how it collects and handles its data. The belated release of data that compares temperatures recorded by traditional mercury thermometers and new resistance probes that sit alongside each other at Brisbane Airport is a welcome start.
But it has raised as many questions as it has answered. The bureau continues to insist there is no significant difference and that its recording devices are maintained to the highest standard. But analysis of the data shows, across the three-year period for which records have been made available, probes returned temperatures higher than the mercury thermometers placed alongside them for 41 per cent of the time.
Recordings were the same for 32 per cent of the time and lower for 25 per cent of the time. In the most extreme cases, temperatures recorded by the probes were 0.7C hotter than the mercury.
This raises some difficult questions about the bureau’s ability to claim new records are being broken at stations when comparing temperatures taken with the new technology with temperatures recorded using mercury thermometers in the past. Given that new records are claimed on the basis of readings that are only a tiny fraction of a degree warmer, the problem is obvious.
The bureau continues to back its technology, as it is entitled to do. But, given the results of analysis of what has been made public, it is difficult to argue all of the parallel data records the bureau holds should not be made available for similar inspection.
The lengths to which the bureau has gone not to co-operate with Freedom of Information requests for data gives the impression of an organisation with something to hide.
As we report on Saturday, more than three years after a FOI request was lodged by scientist John Abbot, the bureau released three years of data after its refusal was challenged in the Administrative Appeals Tribunal. Throughout the ordeal, the bureau had first claimed there was no parallel data as it simultaneously erected barriers to the public accessing parallel temperature data by saying there would be high processing costs involved. The bureau refused to waive the costs of satisfying the initial FOI request, claiming there was no public interest in the data sought.
The applicant correctly argued it was truly astonishing that the bureau would argue that accurately evaluating the extent of temperature change during the past 100 years by fully understanding the effect of changes in instrumentation was of no public interest. This is particularly so given the bureau was simultaneously publishing reports and giving media interviews claiming that a temperature increase of 1.5C would have devastating consequences for the planet.
The documents finally released by the bureau included 1094 A8 reports with the handwritten daily maximum and minimum temperatures from both probes and traditional liquid-in-glass thermometers recorded from instruments in the same shelter or Stevenson screen. They represent only 20 per cent of the parallel records held for the Brisbane Airport site, which is one of 38 sites that were originally requested under FOI. The bureau is believed to hold 760 years worth of parallel data from 38 locations spread across Australia.
The information it contains is of more than academic interest, given that much of it eventually finds its way into what becomes the international record for global temperatures on which climate change policy is made. Dr Abbot, the scientist who made the initial FOI application, says he will persevere with new requests to get the remaining information. He says the information should be made freely available as it was clearly in the public interest.
We agree. The Bureau of Meteorology is a public institution and the information it holds is the property of the public. Parallel data records should be made immediately available alongside all of the other data the bureau prides itself on making public.
There should also be a proper, expert appraisal of what the information says, whether it has been collected as fully as it should have been, and whether the bureau is indeed following world’s best practice in the use of platinum resistance probes, as it says it is, despite World Meteorological Organisation advice to the contrary.
The Thursday before Easter (April 6, 2023), the Australian Bureau of Meteorology provided John Abbot with the most recent three years of parallel data for Brisbane airport as 1,094 A8 reports with the handwritten daily maximum and minimum temperatures from both probes and traditional liquid-in-glass thermometers recorded from instruments in the same shelter/Stevenson screen.
I began transcribing this data the next day, and then undertook a preliminary analysis of the data.
As far as I know, this represents the first time anyone has been successful at obtaining parallel temperature data from a freedom of information request in Australia, and perhaps for anywhere in the world.
John Abbot first requested the parallel data for Brisbane airport on 12 December 2019. Dr Abbot initially requested all the available data sets of daily maximum and minimum temperatures, which the Bureau confirmed existed as parallel data for Brisbane airport back to January 2008. While confirming the existence of this data, the Bureau asked that the scope of the request be reduced in size, specifically that the request be limited to a three-month period or 3 x 1-month periods. The case eventually went before the Administrative Appeals Tribunal on 3 February 2023, and was subsequently resolved with the Bureau agreeing to provide three years of data, specifically the 3 most recent years (August 2019 to July 2022) of parallel data for Brisbane airport – that was received the Thursday before Easter.
These three (3) years of data represents just 3 of the 14.5 years (January 2008 to July 2022) of parallel data that the Bureau holds for Brisbane airport. These three years of data, represents just a fraction of the 760 years of parallel data that I estimate the Bureau holds for a total of 38 different locations spread across the landmass of Australia.
Bill Johnston was unsuccessful in his request for data from Canberra airport in 2014, and this was erroneously reported as all this data having been destroyed.
I successfully secured limited parallel data for Mildura in October 2017 and again in December 2017 following the intervention of then Minister for the Environment Josh Frydenberg.
HOW TEMPERATURES ARE RECORDED
The maximum daily temperature is measured as the hottest temperature in a 24-hour period. This was traditionally recorded using mercury thermometers; as the air temperature rose mercury was forced up the tube and as the temperature fell the constriction prevented the mercury from returning to the bulb. The height of mercury, traditionally read manually at 9am, showed the hottest temperature for the previous day and was reset by gentle shaking.
Thirty years ago, probes in automatic weather stations (AWS) began replacing mercury thermometers for the measuring of maximum temperatures across Australia and the world. These probes are generally more sensitive to changes in temperature — so they can measure extremes of temperatures that traditional mercury thermometers with slower response times cannot detect.
Most meteorological offices attempted to achieve equivalence between the probes and mercury by averaging instantaneous recordings from probes over 1 to 5 minutes. The Australian Bureau of Meteorology adopted a different policy instead taking instantaneous readings every one-second from custom-designed probes with longer time constants purported to mimicked mercury thermometers.
Equivalence is important for the construction of reliable historical temperature datasets, for understanding temperature trends and for knowing if a new record hot day as measured automatically by a probe really is hotter than what might have been read manually from a mercury thermometer.
ANALYSIS OF THE DAILY MAXIMUM TEMPERATURE DATA
Subtracting the daily temperature as recorded by the mercury thermometer from the daily temperature as recorded by the probe provides some indication of equivalence. Chart 1 is a scatterplot of this difference, and it shows that for the three (3) years from 31st July 2019 through until 30 July 2022, the probe twice recorded 0.7 degrees Celsius warmer than the mercury for the same weather. The probe often recorded 0.3 and 0.4 degrees Celsius warmer.
This has implications for the recording of new record hot days, as a mercury thermometer may have recorded 0.7 degrees cooler.
Considering the daily values as a time series there appears to be a discontinuity at the end of 2020.
MAXIMUM TEMPERATURE DATA AS MONTHLY MEANS
The daily values can be averaged as a monthly mean. When these monthly means are plotted as a time series the dramatic change in the relationship between temperatures as recorded by the mercury and the probe becomes more obvious, Chart 2.
Before January 2020 the mercury was recording on average warmer than the probe, from January 2020 the probe was on average recording warmer.
Most monthly maximum temperatures from January 2020 fall above the red dashed line, indicating that after this time the probe recorded warmer than the mercury.
I initially thought that this step-change from minus 0.28 in December 2019 to plus 0.11 in January 2020 (a difference of 0.39 °C) represented recalibration of the probe at the beginning of 2020.
The Bureau has denied this, claiming it represents a fault with the automatic weather station in December 2019, that was fixed.
Considering the maximum temperatures as monthly averages, the difference shows the probe initially recording on average nearly 0.2 degrees Celsius cooler than the mercury, and then after what appears to be a recalibration of the probe at the beginning of January 2020, the probe records on average 0.15 degrees warmer than the mercury through until the end of the record.
The step-up cannot be credibly explained by way of a one-off fault.
The difference between the two periods represents an average difference of 0.35 degrees Celsius over the three years of the available record.
The monthly mean maximum temperature falls below the red dashed line from May 2022 suggesting the probe is drifting and may need recalibration.
A similar pattern in the temperature difference between the probe minus the mercury was evident in the parallel data that I obtained for Mildura, as shown by the green series in Chart 2.
There is an article about this in The Weekend Australian, entitled Measure of Contention in Bureau of Meteorology Probe Row.
REQUESTS FOR PARALLEL DATA
I first asked for access to the parallel measurements for Wilsons Promontory lighthouse on 25th August 2015. This request was ignored.
On 16th October 2017, John Abbot first made a formal Freedom of Information request for the parallel data for not only Wilsons Promontory, but also the desert location of Giles where there should be parallel data for the last 25 years.
These requests were refused.
On 12 December 2019, John Abbot made a further request for parallel data from Brisbane airport.
BUREAU NOT CONSISTENT WITH OWN POLICIES
The Australian Bureau has a policy of maintaining mercury thermometers with probes in the same Stevenson Screen for a period of at least three years when there is a change over. This policy, however, is not implemented and mostly ignored. For example, the Rutherglen agricultural research station had a long, continuous, temperature record with minimum and maximum temperatures first recorded using standard and calibrated equipment in a Stevenson Screen back at the beginning of November 1912. Considering the first 85 years of summer temperatures – unadjusted/not homogenized – the very hottest summer on record at Rutherglen is the summer of 1938/1939.
At Rutherglen, the first significant equipment change happened on 29 January 1998. That is when the mercury and alcohol thermometers were replaced with a probe – custom built to the Australian Bureau of Meteorology’s own standard, with the specifications yet to be made public. According to Bureau policy, when such a major equipment change occurs there should be at least three years (preferably five) of overlapping/parallel temperature recordings. However, the mercury and alcohol thermometers (used to measure maximum and minimum temperatures, respectively) were removed on the very same day the custom-built probe was placed into the Stevenson screen at Rutherglen, in direct contravention of this policy.
One might think it prudent to have maintained the two methods of measurement alongside each other for some period at Rutherglen to determine whether there were any biases. There are parallel data for just 38 of the 695 locations with new Automatic Weather Stations. That is there is a record of temperatures from both an alcohol in glass thermometer used for manual measurement of minimum temperatures, and a mercury in glass thermometer used for manual measurement of maximum temperatures in the same Stevenson screen as the electronic probe. But the equivalence, or otherwise, of temperatures from these different measuring systems has never been established in a rigorous way by the Australian Bureau of Meteorology. Furthermore, only a fraction of one percent of the reams of manually recorded data appear to have been transcribed by the Bureau. Most of the data only exists in handwritten reports many of which may never be archived unless there is a campaign to secure them.
MILDURA PARALLEL DATA
In the report entitled ‘The Australian Climate Observations Reference Network – Surface Air Temperature (ACORN-SAT) Version 2’ published in October 2018, Blair Trewin provides differences in what he refers to as diurnal range between manual and automatic instruments at six sites and concludes these are small. For example, at Mildura, Dr Trewin concludes there is a difference in the maximum temperature of just –0.22°C when considering the data for the period December 1997 to May 2000. What is not explained in this key report is that this is a statistically significant, and it is negative, indicating that the probes were underestimating the extent of warming.
In September 2017, the Bureau claimed a series of new record hot days across south-east Australia, including on 23 September 2017 at Mildura. At that time, the media reported this as a new record for the state of Victoria, specifically claiming it was the hottest September day ever recorded – all the way-back to September 1889.
After the intervention of then Federal Minister for the Environment, Josh Frydenberg, following correspondence from me dated 26th September 2017, I was provided with a first set of parallel measurements for Mildura by way of undigitized photographs of A8 reports with parallel measurements for the period November 1996 to December 2000. I analysed a component of this data after manually transcribing the values and found a statistically significant difference of –0.34 °C, –0.27 °C and –0.28 °C for the month of September for the years 1997, 1998 and 1999, respectively. This is broadly consistent with Blair Trewin’s finding that the probes were recording too cool.
I received a second tranche of photographed A8 reports on 4 December 2017 with A8 reports through until January 2015. Analysis of this data showed that after December 2000, there was a change in both the size of the Stevenson Screen used to house the probes and mercury thermometers at Mildura and then later also a change in the type of probe. The screen was changed on 3rd May 2000, and a different probe installed on 27th June 2012. My analysis of this data suggests that both the change to a smaller screen and the change to a different probe produced discontinuities in the temperature data. In particular, the new probe installed on 27th June 2012 recorded much warmer for the same weather. The probe often recorded 0.4 °C warmer for the same weather.
MORE INFORMATION ABOUT BRISBANE’S TEMPERATURE HISTORY
I have a long-standing interest in the integrity of historical temperature data. I first described the problems attempting an accurate historical reconstruction for Brisbane in a blog post in November 2019.
According to the available metadata, maximum temperatures were measured at Brisbane Botanical Gardens (station number 40214) from 1887. For the period 1897 until 1996 temperatures were recorded using the same mercury thermometer in a Stevenson screen. This very long continuous record does not show a pattern of warming. Rather there was warming to about 1900, then cooling to about 1950, followed by warming again.
The more scary the so called scientists can make it look the more money they get and the more power and status they gain.
Then the media sells more units and the loser is the general public.
Daily Telegraph is reporting:
Teenager took his own life after ‘losing hope over climate change’
What are next steps are for the scaremonger ayatollahs of so called ‘climate science’ now they have martyr for brainwashed kids to venerate?
“His is thought to be one of the first deaths linked by loved ones to concern over climate change. “
EDIT FOR ACCURACY:
“… one of the first deaths linked … to
concern over climate changethe vicious, baseless, propaganda of the solar, wind, electric vehicle and other human CO2 lie-based scammers.”
Their greedy, goblin-like, barely-suppressed, glee can be heard in their: “one of the first.” Disgusting.
equal in value, amount, function, meaning, etc.
“one unit is equivalent to one glass of wine”
synonyms: equal, identical, similar, parallel, analogous, comparable, corresponding, correspondent, interchangeable, like, commensurate with, the same as, synonymous with, much the same as, amounting, tantamount, approximate, near, close, of a kind, of a piece, coequal
Sounds like a potential question of Semantics
Since both measurements are made in centigrade they are SIMILAR in FUNCTION and SIMILARLY scaled…so…they fall within the definition of equivalent… loosely
When people use a word that has multiple meanings, they are generally intending to convey just one of those meanings, the intended meaning discernable by the context of the sentence. In this case I, and I contend most people, would have assumed that equivalent meant the same in every way that was relevant to interpreting weather data. This does not appear to have been the case.
By equivalence, I was referring implicitly to the difference in measurements not being statistically significant.
“Analysis across the three-year period, shows the probes record higher 41 per cent of the time, the same 32 per cent of the time, and lower 25 per cent of the time.”
The scatter plot shows a fairly small range, and this seems like a fairly random scatter about zero. It can’t be far above the accuracy limit of the thermometer.
How about some really useful statistics for the whole period – mean difference, standard deviation of difference. Or even post the data?
The scatter plot shows a dog’s breakfast. Something that is not fit for purpose.
The scatter plot is a difference between two instruments. It should be a dog’s breakfast. Small random numbers, as they are.
No there should be a pattern. There is not. The differences shouldn’t be random.
Why not? If both instruments are accurate, the difference should be noise.
Yes Nick true but what is the noise level? For both instruments is is a minute fraction of the reading, just like all useful measurements! These differences are actually huge by comparison. Noise is completely random, these are not random are they, always a very bad sign.
They are nearby, but measuring the air about them. The difference is probably just variation in the turbulent air.
You have no idea about physical measurements do you? Noise is extraneous external information that is not part of what is being measured.
As a mathematician, you may consider temperatures outside of what you think they should be as noise, but they are simply excursions in the atmosphere that cause a widening in the dispersion of data around a mean. Nothing more.
If there were ‘random scatter about zero’, the % higher and lower would be the same, would it not, Nick? In other words, the ratio of higher:lower readings would be 1. Actually it is 1.64
41% higher – 21% lower
Bias toward reading higher
Bias? mmmm. Guess that prejudiced “data” gets carried straight into the crystal ball that is climate models…
Excellent idea. I’m sure BOM will get right on it and spend some of their massive climate change study funding to produce and release this data publicly as soon as possible. /sarc
“and then after what appears to be a recalibration of the probe at the beginning of January 2020, the probe records on average 0.15 degrees warmer than the mercury through until the end of the record.”
Looking at Chart 2, it seems that not a single month reaches 0.15° C, and most monthly differences are a lot less, below 0.05° C. That is pretty good accuracy.
What I find interesting Nick is the lengths that this public institution went to in not providing the information. You may want to blather on about costs etc. but the plain fact is that climate change remains a controversial global topic – and this is unacceptable and, indeed, suspicious behaviour.
So Jen got what she was after, how about looking at what it says?
Yes but look how long it took
Actually that is not correct. She got only a small part of the data requested. She wanted ALL the data from the inclusion of the electronic reading device.
The researchers really want ALL the side by side data, but are requesting only specific sites due to the massive push back of the TEAM.
Providing the data, all of it, would probably only mean starving a couple of thousands of the poor. The results of the CFGW fraud, which full transparency would show that Aus. temperature increases and records are a fraud, and the resultant massive expenditures for unreliable electrical generation will k!ll millions through lack of housing, “energy”, food, etc.
Why do you hate poor people so much that you push a fraud, ignore the results, and support hiding data?
Nick, there are at least five issues you are denying:
We asked for 15 years of data for Brisbane Airport, and were given 3.As for Mildura, the probe is often recording 0.4C warmer and up to 0.7. This has implications for the calling of new record hot days, and more.As I state above, the difference is statistically significant whether the analysis is of all the data, or month by month.The average difference from before to after the January 2020 step-up is 0.35. What caused the step-up? The Bureau claims it was a one-off fault in the AWS in December 2020. That is not credible.The apparent drift down after January 2020 is similar to the drift I found at Mildura, and is in the order of 1 degrees C per 100 years that is considered a significant rate of change in mainstream climate science.I could go on.I will be combining the details of this statistical analysis with what I already have for Mildura and publishing in due course.
I am not denying any issues. I am commenting on the data you have presented. Your description is very inadequate. You talk of statistical tests, but give no details. You claim the later part have an average difference of 0.15C, which is clearly contradicted by Chart 2. And all that you have posted seems to indicate not very much.
Since you battled the Bureau so hard to get the data, why not post it? Or at least properly report it?
I think I have provided much information, at a level suitable for a blog and newspaper article.
The bottomline is that there is no equivalence. The difference between the measurements is statistically significant. This should be of concern.
The 0.15 refers to the average monthly exceedance after January 2019. I agree that considering just that period of the record it would appear they have done a fair job.
Because I was provided with only a few months of data before January 2020, I don’t know how long the probe was recording on average 0.2 degrees cooler. At Mildura this was the case for some years. Then there was a step-up, and then it drifted down again.
Is an average discrepancy of 0.35 a problem. I would have thought so, if the objective of the measuring included an assessment of climate variability and change.
What I can see from the Brisbane parallel data, particularly when it is complemented with the parallel data from Mildura, is an inability to get a consistent reading. The probes drift, and thus need to be continually recalibrated. The Bureau claim to not recalibrate, but rather to replace.
The Bureau is perhaps not so worried about accurate observational data, because whatever they get is remodelled in accordance with theory. In fact, the Bureau can and do say that the readings/the observational data is so dodgy all the more reason to homogenise.
“I think I have provided much information, at a level suitable for a blog and newspaper article.”
If the data was important enough to create all this fuss, surely it is important enough to let people see it.
“The 0.15 refers to the average monthly exceedance after January 2019. “
I can’t see it on Chart 2. I’ve marked 0.15 with a red line:
Thanks Nick. I need to go back to my notes regarding the 0.15. I am guessing that the 0.15 refers to the average of all the daily values, that are shown in Chart 1. This second chart shows the average monthly exceedance. I may have the text wrong, in part.
I did much analysis, swapping between daily and monthly. I have attempted to pull something quickly together here, including because there was media interest.
The average of daily values has to be the same as the average of the monthly averages.
Did you base the statistical significance claim on the 0.15?
In Chart 1, from the ~1/1/2020 step-up to the mid-2022 drop: I see the most common difference being zero, with +.1 in close second place, and -.1 somewhat short of tying +.2 for third place. So, I see the average daily difference during the time between the ~1/1/2020 step-up and the mid-2022 drop as being between .0333 and .05 C in the direction of the probe reading warmer than the mercury thermometer. I estimate probably closer to .05 than to .0333 while being in between these, but this is a lot less than .15.
You are looking at a difference chart to me then. :-). I think you are confusing the monthly (chart 2) and daily (chart 1).
I was definitely looking at Chart 1, Daily difference, probe minus mercury, Brisbane Airport, August 2019 – July 2022. From the ~1/1/2020 rise to the mid 2022 drop, the average is nowhere near +.15 C unless one excludes the readings of zero or in negative territory.
How about you make an effort to assist in obtaining and digitising the BoM data? After all, if all is above board as you claim it will support and prove your claims of an honest BoM.
Oh, I thought not.
The BoM has a huge amount of data on its site. That is enough for me to digest. And far more than anyone here has interest to digest. The only interest is in harassing the BoM.
So a taxpayer asking a taxpayer-funded bureaucracy for some particular non-secret records it has in its troves is “harassing” them?
Woodward & Bernstein wouldn’t have gotten far if they’d subscribed to your philosophy of timid inquiry approach, would they Nick?
I can’t believe how arrogant this Stokes guy is — he is literally saying that if you aren’t part of the official anointed priesthood of climastrologers, you should not be allowed anywhere near the holy air temperature data.
This is not the air temperature data. That is all extensively published. This is about parallel measurements taken to check on the instruments. If they agree, as here, there is no new information. If not, they fix something.
What Jennifer is saying is “I want to check it too“. And, apparently, she has found nothing new to report. And she doesn’t think it is worth making the data public.
She is probably right.
Nitpick Nick strikes again.
The BoM should be harassed at every opportunity. They certainly do not provided anywhere near the quality of information they pretend to offer for the annual quarter of a million dollars of tax-payer dollars they get. Eg, when their ”prediction” for this autumn (on which many farmers rely) was ”warmer and drier than normal”. The EXACT OPPOSITE of which is actually true. A coin toss would be more accurate.
Sorry, yes billion.
If BOM is so confident in what it says then there should be no problem in publishing every line of (raw) data. It is a taxpayer funded organisation and should, as a matter of course, publish what the public is paying for.
When an organisation I, well Australians in this case, am paying for doesn’t publish then the conclusion I come to is that it’s our old friend economical with the actualité.
Judging by comments I read in the MSM, the ones that allow only lightly filtered comments at least, I’m far from being alone it that belief
It seems odd to me that people get on their high horse about their rights to the data, and when the BoM does release it, show no interest in looking at it (or being able to).
So, I received over a thousands reports/pages of data just before Easter. I spend Easter transcribing this data. The Bureau did not provide me with anything digitised. I transcribed the data over a period of two days. Then did some basic analysis. I’ve shared this first cut.
Now, can you explain the step-up that occurs between December 2019 and January 2020.
The bureau claim there was an error with the AWS. A single fault that was fixed. What do you think?
Oh I’m sure Nick wouldn’t ignore your question. He’s a straight shooter. Nobody can say that he only nickpicks at the minor discrepancies of skeptical arguments while remaining mute about the obvious dissembling by the BoM.
IIRC there was a Climategate email from a David Jones from the BoM boasting that when he got a request for data from anyone who was not part of “The Team”, he responded by swamping them with thousands of pages of unorganised records.
Seems that the BoM playbook from 20 years ago is still in use.
“the BoM playbook”
It is the FOI playbook. FOI says they have to produce the documents that exist. They don’t have to work to create the documents the applicant would like to exist.
Yes, we have lots of people working hard to distort the world’s temperature records.
I consider them criminals.
“What do you think?”
I have no special knowledge. But I have no reason to doubt that “there was an error with the AWS. A single fault that was fixed.“. And after they fixed it, it read very slightly higher than before.
Pretty obvious that you too have no interest in looking at it. So you just whine (a lot) when others take the time.
“when others take the time”
And who would they be? You? Do tell when you have actually something to say about it.
Duh! This would be Jennifer — not once have you failed to whine about a single one of her articles.
But did they ever release “the data”? I mean ALL of the data, raw and unrevised?
No they didn’t. And you still support their “hiding the incline”.
Anthony and this site covered the changes in temperatures in the US for the changes in enclosures, (paint ILO white wash, etc.) enclosure siting, temperature measuring devices, etc. in the US and the powers that be never addressed these issues, never noted sites that were deficient, etc.
It seems that this is a multi national issue. And every analysis seems to show that the errors are ALWAYS TO A HIGHER TEMPERATURE when averaged out over time.
Just a coinkydink? I think not.
For myself I have no ability or desire to examine this data.
However, like many other fields, I can make a decision by observation and reading where science is concerned. When any group decide that their data and methodology are top secret and beyond any other human’s comprehension then I have a suspicion that all is not well.
I am quite happy for others to examine the data, even you, and either confirm or disprove what is being claimed.
So far to me you seem to be trying to divert attention from the first issue identified in a first pass examination.
“When any group decide that their data and methodology are top secret”
There is no secret methodology nor data here. Nor data. The Bom just says there must be a limit to the extent they can be run around by Jennifer’s whims. Remember the 10000 pages they had to hand scan for Mildura. And all for people who “have no ability or desire to examine this data”.
”The Bom just says there must be a limit to the extent they can be run around by Jennifer’s whims.”
You are forgetting, BoM is actually a public service and should be open to scrutiny. They have an obligation to provide whatever necessary to the public.
“You are forgetting, BoM is actually a public service”
It isn’t Jennifer’s service. They need some time to look after their real duties to the public. They had to hand scan 10000 A8 forms for the Mildura caper. Looks like at least 1000 for this one.
Nick, all Jennifer and others have asked the BoM to do is what it states on its website it has openly committed to do –
Freedom of Information
The object of the Freedom of Information Act 1982 (the FOI Act) is to give the Australian community access to documents and other information held by the Government.
Your rightsUnder the Freedom of Information Act, you can:
. . .
You can ask to see any document that we hold. We can refuse access to some documents or parts of documents that are exempt. Exempt documents may include those relating to national security, material obtained in confidence, Cabinet documents, or other matters set out in the FOI Act.
You can ask to see documents, and you can walk in and the BoM will show you, if it has them. But they can’t usefully show you 1000 A8 forms. What Jenifer requires is that they scan those forms, each bound in books, so she can go through them at home.
And the BoM explained how they couldn’t manage this request?
Nick, we went through this a little while ago here –
a number of people experienced in volume scanning / OCR explained in detail how it would be a straightforward, <1-hour exercise for the work-experience kid at the BoM to do these or get them done at a service centre for a few cents a page.
Mathematics might be your forte Nick, but simple practical tasks seem to be beyond your comprehension.
This is not a simple practical task. For one thing, the A8 forms are bound in books. Each one has to be scanned, page turned, positioned, scanned again.
Hand held scanners Nick.
Gawd, even my Samsung phone can scan an A8.
And Jennifer volunteered to do the gathering of the data for them.
“There is no secret methodology nor data here.”
“however, the Forum concluded that it is likely to remain the case that several choices within the adjustment process
remain a matter of expert judgment and appropriate disciplinary knowledge.”
You are right about “data”, ACORN does not provide it. Expert judgment cannot be verified.
Diverting attention is his main mission in life.
I know you are lying, Nick, you are ‘full of it’, obfuscating and, moreso, writing it down.
“I know you are lying, Nick, ” but I just can’t see where…….
batterycarboi to the rescue!
Regarding “As for Mildura, the probe is often recording 0.4C warmer and up to 0.7.”: Chart 2 (monthly difference) shows for Mildura the probe recording overwhelmingly cooler than the mercury thermometer, and shows no points of the probe recording more than .11C warmer than the mercury thermometer.
Hi Don, Have a good look at chart 1 versus chart 2, and a think. :-).
Chart 1 does not say anything about Mildura. I think if you showed the daily differences for Mildura during the period covered by Chart 2 (May 2000 – May 2012), I expect to see a few points where the probe reads at least .4 C warmer and one maybe two where the probe reads .7 C warmer, and thousands where the probe reads cooler, and more with the probe reading at least .4 C cooler than at least .4 C warmer, and more with the probe reading at least .7 C cooler than at least .7 C warmer.
The point of the two charts was to provide daily and monthly data for Brisbane for the three years that I now have parallel data.
Mildura was provided as a point of comparison only.
The Mildura probe sometimes reads 0.4 warmer after June 2012. I am yet to write all this up, in part because some months were held-back from me when I was provided most of this data in December 2017. I had been hoping to get the missing months as part of the mediation.
There is some commentary and some charting of this data in a blog post from 2018, https://jennifermarohasy.com/2018/02/bom-blast-dubious-record-hot-day/
I have the problem that I am sitting on a lot of data, some written up in papers that I can’t get published, and other data that I am still waiting hoping for more data to complete.
I was much more successful at getting my rainfall work finished and published. There is a real reluctance of the mainstream climate science community and the Bureau to be upfront about temperatures.
Thanks for caring, and for continuing the conversation. I had perhaps misunderstood your initial query.
Isn’t it 0.7 degrees C that they say it has warmed
I also think that the change in the 70s from mercury to thermocouple has added something to the global temperature.
Mercury has a reaction with glass and tends to go out with time. Hence the reason for prescribed re calibration. But I think that also only started in the 70s
“Why are BOM making 350-500 million iterations to prior weather records at just one station? 24.05.21”
Via first comment at
And the cartoon that heads it”
My additional comment
“In recipe terms I’d sus that the pasta is vastly overcooked”
I am seriously disturbed by this data, and it shows that the design of the electronic probe system is either deliberately not equivalent to the glass thermometer, or that the design was never tested to check that they both measured the same temperatures. A mercury in glass thermometer has a time constant to arrive at a measurement, which is usually something like 5 minutes to 1% accuracy. This is because it takes heat flow through the glass and into the mercury, which is a fairly slow process. Because an electronic sensor (thermocouple or semiconductor) has very little mass, this process is much quicker. The response needs to be electronically modified to exactly match a glass thermometer, not really too difficult, but it does need to be very carefully tested. I assume that these thermometers have calibration to National and International standards quite regularly, and that part of the procedure is to check the response rate. If not the whole lot is basically useless and should be scrapped immediately. It is of course not possible to modify any of these records to correct the electronic ones, so the warming measurements are completely meaningless! There, immediately NO climate change!
Thanks The Real Engineer.
I had assumed that the step-ups that I see in both the Brisbane and Mildura data were a consequence of recalibration.
I have been told that the Bureau does not ever recalibrate. Rather it replaces.
“The Bureau verifies temperature probes to ensure that they are within specification. Verification is more rigorous and reliable than recalibration,” is the specific quote.
There are a few issues here that the engineer forgets. The mercury has a reaction with glass and depending on the quality of the glass the reading goes lower. Eventually the error goes too big and you buy a new one. But this was only realized in the 60s and by my knowledge only introduced in QC manuals in the 70s.
Thermo couples are much more reliable so the change was good. But the probes are connected to a recorder who measure every second and at the end of the day it calculates an average plus spits the max and min. So around the mid 70s most stations had changed over to thi automatic recording. I find in SOUTH AFRICA no change in T if I look at Means from 1975. My reasoning above always prevented me from even looking at earlier data because the new method is completely different than the old method where you looked at Tmax en Tmin for the day on the mercury thermometers and just took an average.
That is correct Henry, but I would expect that to be taken into account by normal calibration checking against standards at reasonable intervals, probably a year. It also gives a known constant offset, so normal calibration will eliminate this error. You will next find that the equipment has no calibration cycles, more errors!
Um, that comment from the Bureau is exactly what calibration does, the same as all out instruments. Someone doesn’t understand measurement, I presume it came from the PR department? I wonder what the accuracy specification is, correct to within a few thousandths of a degree would be quite possible? Hewlett-Packard used to make a thermometer system, to measure micro-degrees accurately, using a quartz oscillator as the sensor. The linearity of Platinum resistance thermometers certainly needs calibrating out, at least over any significant range, ie. -20-80C, if one wants high accuracy and resolution.
Great Comment “The real engineer” Just a few minor details. They use neither thermocouple or electronic (semiconductor) thermometers. They use platinum resistance thermometers. These too as you say have a shorter time constant. They claim to have thermally matched the time constant by adding a metal sleave or something similar. So not an RC time constant filter in the electronics but a mechanical change to the shape of the thermometer. I suspect they now have a curve that is not exponential but acts like a multi pole filter due to differing diameters and surface areas etc.There are however other issues with the sampling method that could also affect “diurnal range”. The time constant of any thermometer is altered by the medium. Wind speed, humidity etc. So an exact match is really hard but one thing makes it impossible. The time constants of mercury and alcohol thermometers are different to each other. So the single platinum cannot match both. They would need two resistance thermometers with two different time constants. Both individually matched across the ranges of humidity and wind speed.
It is why I have always recommended stopping the old record when a change is made and starting a new one. Even a change in LIG thermometers has no guarantee that the manufacturing of different instruments will always give 100% the same readings.
The mathematicians need for long records to prevent spurious trends when using a plethora of short records does not justify modifying past data “to make it compatible with current measurement technology”.
When I first heard about the modifications being made to data, my first thought was that people were attempting to make it fit their perception of reality. That is not scientific at all. It is similar to p-hacking to achieve the result you are looking for.
My understanding of scientific ethics (and legal also) is that changing data is admitting that data is not fit for purpose. It is as simple as that. Either use it as is, or throw it away. Black and white binary choice. If this were done, there would be no wrangling about “trends” as is going on right now.
Changing probes often is a perfect example of having technology that is not reliable over multiple decades. Modifying and homogenizing data based on this is not a scientifically justifiable task.
The same comments apply to platinum resistance measurements as to the others I mentioned. The best way to use such a sensor would be to immerse it inside some mercury inside a glass bulb (insulated of course). With a little care this could be made to work exactly as a mercury thermometer, and the same in alcohol to operate as an alcohol thermometer.
I do not see that a metal sleeve can be considered similar to a glass/mercury thermometer, and we have proof of this because of the speed of measurement response. 20 seconds or less is not at all similar.
It is this line that should be followed in questioning the higher readings, particularly questioning whether calibration readings have been taken against standard thermometers. I expect that the answer will be that the platinum resistance one has higher resolution (probably to a milli-degree or less) but this just pinpoints the problem, essentially the two types cannot be compared properly. Thus the two datasets cannot be compared or merged either, at least not in any scientifically supported way. In other words the increases reported cannot be substantiated.
That site info “metadata” you have for Brisbane site 040214 is not the most recent. In 2019 i emailed them with a few questions about the accuracy of that location.
Here is the reply.
Thank you for your emails which highlighted errors with the location of these two weather stations. The co-ordinates for these sites have been corrected internally to reflect the estimated location for each.
The correct Brisbane Botanical Gardens (040215) location is -27.4778, 153.0306. Note that this site was a rainfall site only, there was never any temperature measuring equipment at this site.
The Brisbane Regional Office (040214) location history is a little more complicated as there were some site moves, but the last location for this site before it closed is estimated to be -27.466, 153.027.
The metadata files available online are only updated once a year at the start of the Financial Year, so these updated locations will be reflected in those files from sometime after July 2019.
Climate Data Services
Bureau of Meteorology”
While they say “There was never any temperature measuring equipment at this site”, i looked into it a bit more and found that Clement Wragge had a Stevenson there for a couple of years after 1890 but he seems to have closed it well before the BoM were ever involved.
Thank you Lance. This is important information. You had already mentioned it to me. But I had forgotten.
They mention a few complicated site moves but the first Stevenson screens, three of them went in at the location of the photo on the 2019 blog of yours that you link too. I found evidence that they were there from 1/1/1887.
Your blog post was closed for comments. So i put some of it here. With links to source information.
They’re perfectly lying and anyone can demonstrate how.
Get yourselves one of these:
Because the sun heats air directly, that air has very low specific heat capacity, the probe will ‘see’ passing clouds. ##
i.e. What might be a cool & cloudy day but in the gaps between clouds the temperature seen and recorded by that probe will skyrocket.
It does. I haz the T-shirt. and the data.
THAT is the problem with the electronic probes.
Something dawned, esp when considering the speed response of them…..
We all rave about airports and jet-engine blasts – yes, fine OK you can see that.
What about glare from large and small ‘areas of glass’
e.g. Windows of buildings, parked cars
Do you see? = the exact same principle of the Ivanpah heat collecting solar farm. And others.
Just how immune are conventional Stevenson Screens from those sorts of events?
That as the sun moves and seasons change, any little weather station will find itself feeling like the salt tower at Ivanpah
Especially at airports, usually noted for having vast expanses of glass.
## My ‘Slew Rate’ test simply involved a cup of coffee.
I set it recording, at room temp and after a few minutes for it acclimatize, popped the probe into my coffee.
Downloading the record saw the temp go from a steady reading of 21°C to a new stable reading of 66°C inside just one 20 second time record.
That thing is fast.
One of the hotel towers in Las Vegas is curved in such a way that sunlight concentrated on the pool deck started a fire. The casino had to build a heat resistant structure at the location to protect their customers.
The old magnifier and ant thing.
man, such great work! I did get a chuckle out of this though.. “Bill Johnston was unsuccessful in his request for data from Canberra airport in 2014, and this was erroneously reported as all this data having been destroyed.”
Reminds me of NASA somehow just losing the telemetry data from Apollo 11. Only the most important even in man’s history and NASA, well, they don’t know what happened to that daggum data.
Nuttin’ weird there. Yet, everyone still loves NASA.
NASA has its problems, but, what happened to the Apollo 11 data is known:
SUMMARY OF REPORT (link below)
By the early 1980’s, NASA had a severe shortage of magnetic-tape. (“The Apollo 11 Telemetry Data Recordings: A Final Report” at 12)
A January 8, 1981 report said that NASA Goddard’s magnetic-tape recycling facility added a third shift. (Id. at 14)
The day Neil Armstrong and Buzz Aldrin walked on the moon in 1969, Goddard received 45 tapes documenting this achievement. (Id. at 13.)
It is almost certain that all 45 Apollo 11 tapes were re-used in the early 1980’s. The original data, gone forever. (Id.)
Observation of Investigators
NASA had no idea that it would be possible in 40 years to digitize that type of television data into high quality video. (Id. at 16)
NASA AND Goddard followed generally acceptable record preservation procedures. The gut-wrenchingly undesirable effect of this data destruction was not reasonably foreseeable. (Id.)
This was NOT the case with BOM, here, who did not follow generally acceptable data handling procedures. And, moreover, is actively trying to hide data. NASA fully cooperated with the investigation team that wrote the above report.
NASA ran out of magnetic tape?
Reminds me of Telstra telling me that they could not provide an extra phone line because they had run out of copper cable. The ombudsman was not amused.
My ultimate problem with this whole sorry story is the unaccountability of bureaucrats and administrators. That is unacceptable. There should be zero tolerance for non compliance. Meteorologists are dealing with the same stuff we ordinary citizens deal with everyday. It is not classified or a state secret whose release would endanger us. We need the name of the person who authorized not releasing this harmless public information. Give him a second chance and then see to it that he is fired and prohibited from holding government employment in any capacity. Only a few officials will have to be fired and the records will magically appear. It is time to start playing hard ball with these knuckleheads.
The BoM has “760 years of parallel data” ? Is this aggregated records?
Hi Streetcred, There are apparently 15 years for Brisbane, there were 20 years for Mildura (but much of it scrappy), there are at least 25 years for Giles. I don’t know exactly, but I’m guessing about 700 years of A8 reports that need to be transcribed and then analysed. It took me 2 days to transcribe 3 years of reports from Brisbane Airport, and begin this first cut/preliminary analysis.
Would it be more productive to request the already-digitised numbers that BOM plausibly used to arrive at their general statement that the Pt resistance probes performed adequately?
Hi Geoff. You can ask for that data through an FOI request. I got the Mildura component of it when I got the nearly 20 years of Mildura data. I got the same gross difference as Blair Trewin for the few years that he analysed (December 1997 to May 2000). He omitted to state that the difference is statistically significant. And it was only 2.5 years. I want the 700 odd years of parallel data for the 38 -odd sites. Call me greedy, but it is usually a lot less time to analysis data than it is to collect it. So many man hours were spent collecting this data, let’s transcribe and then analyse it all. :-).
As I am going on holidays, I’m reluctant to get involved in yet another BoM-harassment post by Jennifer Marohasy. Like previous posts on WUWT about Goulburn, Mildura, Cape Otway … etc, arguing that an AWS minus thermometer Tmax difference of 0.7oC on one or two days makes any difference, let alone that the difference is “significant”, is more of the same.
There is a step-change in the difference, probably from 1 January 2020 that could be equally due to the thermometer as the probe. Some people reading this, would know that such comparisons require data to be homogeneous through time. So, in comparing differences in means, either the earlier data should be ignored, or adjusted (or in a 1-way sense blocked as separate data). But then there is problem of autocorrelation of differences. Yes Jennifer, the problem that your particular brand of paired t-test analysis consistently ignores, which is that significance of the implicitly requires that data-pairs are independent.
Jennifer either knows what she is talking about or not. While, rather than dealing with or justifying her use of the paired t-test she hisses at the messenger, the autocorrelation problem has not gone away.
Noting that she demands no less of the Bureau, which is the whole point of this post, she should state the test she used and put the data she analysed in the public domain, so her claims can be independently verified. I routinely do that on http://www.bomwatch.com.au and in order to support her argument, she should do the same.
The instrument problem
Do people reading this post really expect two different instruments, even two maximum temperature thermometers observed by two independent sets of eyeballs, to agree precisely?
Oh, wait, JM has never undertaken regular meteorological observations so how could she know?
Does anybody commenting here have any idea how data ends-up in an A8 field book and then in the Bureau’s database? Did Marohasy check which data for each day ended-up in the online climate data database; or, if any of the apparent ‘significant differences’ were reported as ‘record-high’ daily temperatures?
While I have no problem with the view that the Bureau should have databased parallel data, I have a major issue with people being whipped into a frenzy because they didn’t.
With respect to parallel data for Canberra Airport that I requested in 2014 (https://joannenova.com.au/2017/08/another-bom-scandal-australian-climate-data-is-being-destroyed-as-routine-practice/), I dropped the issue because unlike Jennifer and husband John Abbot, who are generously funded through connections with the Institute of Public Affairs, my projects are unfunded.
A range of techniques are available to assess data; and it is the gross behavior of data (and data attributes) that determines if timeseries are affected by biases, and what that means in pragmatic sense. Thus, although a scandal at the time, I solved the Canberra problem using independent methods (https://joannenova.com.au/2017/10/canberras-hottest-ever-september-record-due-to-thermometer-changes-and-a-wind-profiler/).
Simply stated, trend in Canberra data was caused by site changes not the climate and it was from that project that I developed BomWatch protocols.
Jennifer says “I first asked for access to the parallel measurements for Wilsons Promontory lighthouse on 25th August 2015. This request was ignored”. However, had she read the metadata file, she would have discovered the data she wanted actually did not exist. Because Parks Victoria no longer wanted to measure the weather, the Bureau installed an AWS on the day thermometers were removed.
So, while ignoring that she did not study site-summary metadata, she still touts that the Bureau withheld parallel data they did not have. Non-existent data cannot be provided simply because Marohasy does a selfie and wills it to be so.
She says “BUREAU NOT CONSISTENT WITH OWN POLICIES”. As evidence she focuses on Ruthergen, which was a non-Bureau site run by Agriculture Victoria. While she has been told over and over that Agriculture Victoria, Parks Victoria and other state authorities are not beholden to “Bureau policies” she still runs the same fake-narrative.
Jennifer is either a knowledgeable scientist who considers all angles, or a bullshit artist, but she cannot be both.
Disclaimer: As a strong supporter and paid-up member of the IPA, I also contribute to the Marohasy-Abbott partnership. As they also represent me, I want many fewer selfies and much more rigor in what passes for Marohasy-science.
Happy holiday to me. While it won’t be published in The Australian, I’ll endeavor to send a me-selfie sitting on a rock!
All the best,
Dr Bill Johnston
Sounds like you’ve diversified into MarohasyWatch.com.au, Bill.
Yep. Triggered another long rant.
You continue to be a contradiction and to spread misinformation, including about me:
You established ‘BOMwatch.com.au’ because the Bureau often get the metadata wrong and/or more confused. It seems ok for you to report this, but not me.
You know that I ran field stations in Madagascar and then Kenya for a period of 6 years (1985 – 1991), and was subsequently responsible for logging temperature data in sophisticated glasshouses in Brisbane for a further 7 years (1991 – 1998), yet you continue to claim I have no practical experience recording temperatures.
And I really don’t appreciate the extent to which you have phoned and emailed various key opinion leaders over the last few months spreading further misinformation about me as part of the obsession you have with destroying my credibility. It is so mean and unjustified.
I think I am missing the point of this article. As I understand it, we have data from two locations where a conventional thermometer is compared to an electronic probe. In one case the probe measures a little warmer than the conventional thermometer, in the other case a little cooler. Why is this surprising or sinister? I wouldn’t be surprised if you replaced one conventional thermometer with another conventional thermometer and got a similar difference.
1.Difference is statistically significant
The bottomline is that there is a statistically significant difference between the measurements that is not random for either Mildura or Brisbane.
We need more data to understand how this varies with latitude and altitude and over time and with the new versus old probe design. That is if we care about the integrity of the measurements.
The managers at the Bureau will tell you that the raw data is ‘crappy’ so we best to just homogenise.
The remodelling of the data through the process of homogenisation is likely to have a bigger effect on overall trends.
2. Can theoretically recalibrate to however much warmer they want
The Bureau could decided to recalibrate according to, for example, a latest anomaly as regularly updated at https://climate.nasa.gov/vital-signs/global-temperature/ This is a hypothetical that I am adding for you Mark.
It is the case that there doesn’t seem to be much curiosity from you, and many others, about how they calibrate. It is apparently relative to a reference sensor, and what is this calibrated relative to?
What are the probes calibrated relative to in the US and UK? Is anyone monitoring any of this. How much warmer or cooler does the average probe in the US or UK read relative to a mercury?
3.Probes used by the Bureau are not electronic
You mischaracterise the probe, it is not electronic. It is a piece of platinum with the temperature resistance measured electronically by some wires hooked up to a data logger that is called an automatic weather station. There are potentially problems with calibration and electrical noise.
Given the available data for both Mildura and Brisbane, when the daily values are charted as monthly means it is apparent that the nature of the difference can change abruptly. What causes this? What caused the step-up after January 2020 at Brisbane?
Was this difference at Brisbane, then beginning to drift down, as we see happened at Mildura for a period of some ten years (green series in chart 2).
In the case of Mildura the current probe regularly measures 0.4 degrees warmer (data not shown), in the case of Brisbane the probe regularly measures 0.4 degrees warmer and occasionally up to 0.7 degrees warmer (chart 1).
The Bureau calls new record hot days when temperatures are some fractions of a degree warmer than previously recorded, and is claiming a catastrophe if temperatures exceed a 1.5 degree tipping point.
Part of the problem could be solved if the BoM numerically averaged the instantaneous readings made every second over one to five minutes, as purported occurs in the UK and US respectively.
But then there is still the problem of electrical noise and drift, and what exactly is the reference that they calibrate against?
Hi Jennifer – thanks for your detailed response. Probably worth one more round of responses.
1.Difference is statistically significant
I don’t understand why you place such stress on statistical significance. This only tells you there is less than 5% chance of such a big difference in the observed results if there was no difference in the thermometers. But I am not arguing that there is no difference in the thermometers. I am only asking if the difference matters.
As I understand it, the average difference was + 0.15 degrees in Brisbane and -0.34, -0.27 and -0.28 for different years in Mildura. So there doesn’t seem to be any evidence that the new thermometers give warmer results than the old thermometers – just different. How does the size of difference compare to when a thermometer is replaced with a thermometer of the same type? Or when there is a change in the observation routine? Or in the local environment?
2. Can theoretically recalibrate to however much warmer they want
I suppose so – but it looks like they didn’t or the difference would be consistently warmer (or consistently cooler). If the BoM is a dishonest organisation I am sure there are less obvious ways for them to cook the figures.
3.Probes used by the Bureau are not electronic
Fair enough. I don’t think the technology makes much difference to the argument.
What caused the step-up after January 2020 at Brisbane?
I can’t see why you reject their explanation – they fixed a fault in the AWS.
The Bureau calls new record hot days when temperatures are some fractions of a degree warmer than previously recorded, and is claiming a catastrophe if temperatures exceed a 1.5 degree tipping point.
But surely most record hot days are exceeding a record which was established quite recently (they certainly are in the UK) and therefore measured using the new probes? As a scientist I am sure you understand that a difference in the mean of thousands of measurements is quite different from a difference in one of the measurements. (The standard deviation of the mean is roughly proportional to the inverse of the square root of the sample size).
Statistical tests are routinely used in science to establish equivalence or otherwise. I can see you would prefer hand waving.
And I can see from the following comment that is wrong that you are here to make mischief:
“As I understand it, the average difference was + 0.15 degrees in Brisbane and -0.34, -0.27 and -0.28 for different years in Mildura. So there doesn’t seem to be any evidence that the new thermometers give warmer results than the old thermometers – just different.” end quote
The data can be analysed in so many different ways.
At Brisbane, on a daily basis the probe often records 0.4 degrees hotter, and sometimes 0.7 degrees hotter. These are daily differences.
If we consider the Brisbane data month by month, then for December 2019, as an example, the mean difference is minus 0.28.
There is a difference of 0.35 in the readings before and after January 2020.
Ideally I would be provided the entire 15 years of data so I could make even more meaningful comparisons.
“Statistical tests are routinely used in science to establish equivalence or otherwise. I can see you would prefer hand waving.”
Yours is the hand waving. Mark is making the correct point that statistical testing serves no purpose here. It may refute the null hypothesis that the instruments are the same. But no-one expected they would be.
Statistical tests are routinely used in science to establish equivalence or otherwise. I can see you would prefer hand waving.
I am sorry you find it necessary to make it personal.
Statistical significance is a useful tool but it is important to understand its meaning and limitations and not use it blindly as some kind of magic incantation. I assume you used a paired two-tailed T-test and assumed the same variance in both popuIations – right? And that the results were significant at the 95% level? Just to remind you (I am sure you are well aware of this) if something is significant at the 95% level then it means that if the null hypothesis is true (which in this case is that the means of the two populations are the same) then there is less than 5% chance of the result falling into the rejection zone. If the result does fall into the rejection zone, as yours did, conventionally (although not without controversy) we reject the null hypothesis and conclude there is a difference between the two means. But remember:
5% of the time the result will fall into the rejection zone even when the null hypothesis is true (i.e. the two means are the same). You have to be careful not to select the your sample data in a biased way – for example, you don’t say how you settled on the Brisbane data as opposed to any other.
Even if you can reject the null hypothesis validly, this says nothing about how big that difference is (i.e. the effect size). Which is why I was trying to explore what the difference in the means was. As you say there are many ways to cut the data – I could only work with the figures you supplied. Obviously just picking individual figures proves nothing. Why don’t you supply the 15 years worth of data then we can all analyse it?
I don’t know what you mean when say I am only here to make mischief. I am trying to understand the validity and scope of your work. Is that making mischief?
BoM made a presentation to the 2020 Disasters Royal Commission claiming fire seasons were getting longer looking at FFDI records from the 1950s. What they didn’t tell the Commission, was that they were comparing weather data collected manually at 9am and 3pm daily until the 1990s, with data recorded every second or minute with the new probes.
They haven’t published the work so we have no idea how they analysed the data. But I bet anything thew work is dodgy and and when independently scrutinised will be shown to be a fraud.