The Australian Bureau of Meteorology have been struck by the most incredible bad luck. The fickle thermometers of Australia have been ruining climate records for 150 years, and the BOM have done a masterful job of recreating our “correct” climate trends, despite the data. Bob Fernley-Jones decided to help show the world how clever the BOM are. (Call them the Bureau of Magic).
Firstly there were the Horoscope-thermometers — which need adjustments that are different for each calendar month of the year — up in December, down in January, up in February… These thermometers flip on Jan 1 each year from reading nearly 1°C too warm all of December, to being more than 1°C too cold for all of January . Then come February 1, they flip again. Somehow the BOM managed to unravel this bizarre pattern (cue X-files music) and figure out exactly what anti-horoscope-adjustments to use (and they were different in every city). Modestly the BOM did not explain to the public how clever their adjustments were; despite their $300m budget, it took volunteer Bob Fernley-Jones to reverse out the Special Horoscope Cure, and find the square wave algorithm that repaired our damaged climate records. Lucky for the BOM it’s all laid out below. 😉
Fernley-Jones had a theory that the best thermometers were in our biggest, richest capital cities. Instead he found fickle unreliable thermometers everywhere. Adjustments didn’t line up with the times the stations had moved. Sometimes the maxima step changed, but not the minima. It was as if the Bermuda Triangle rolled through the data. I asked Bob to tell me the best stations (the BOM would want to know). He laughed. Below he continues his exhaustive, diligent analysis of rural stations and finds the same old mess as he saw in the capital cities.
And on it goes. Hours of work. Bob Fernley-Jones does a special kind of graph which packs in 30,000 data points, looks a bit scary, but don’t be put off, the natural noise is easy to tell apart from those sharp man-made step changes. In a normal world there would be just as many step-downs as step-ups as adjustments went both ways, there would also be reasons for the steps (pah!). In a warming world where an ocean of concrete and bricks had been built near thermometers, the adjustments would step down to compensate, as those lines tracked from left to right across the page (but they don’t seem to).
We owe a big thank you to the tireless work of Bob and others who are holding our public “servants” to account. The message in here is that the rural stations are just as bad as the city ones, and that bizarre inexplicable adjustments are everywhere. See Bob’s table at the end for the meanings of the notes on the graphs. — Jo
Guest Post by Bob Fernley-Jones.
Corrupted Australian Surface Temperature Records.
Part 3; Temperature Anomalies at BoM Rural Weather Stations
This is Part 3 of a series of studies that investigates the credibility of the Australian Bureau of Meteorology (BoM) homogenizes surface temperature records. Before this, Part 1 covered six State or Territory capital cities in the expectation that those stations had the best resources and robustness in long-term record keeping. Unfortunately, even these iconics were not found to meet the institutional claim of “World’s Best Practice”.
That aside, it is now time for a wider look and comparison with some long-record rural sites where the Urban Heat Island (UHI) effect, as it is well recognised in big cities, is generally not a complication. To avoid any suggestions of cherry-picking, a greater number of twenty-four rural sites are herewith exposed. Also briefly considered within the total 112 ACORN sites are those with relatively recent start-up dates. (17% are short-record sites with an average of only about 59 years timespan through 2014, not the proper full 104-years that the Bureau claims to cover since 1910).
Importantly, this report summary, compiled from almost 80 megabytes of data, is NOT about the methodology that the BoM has used in its homogenization (or any “climate change” theories). Instead it is simply a test for reasonableness in the Bureau’s resultant “corrected” DATA over that of their currently published “raw data”, (regardless of HOW they got there). By raw is meant those recently retained records at the BoM website; ‘Climate Data Online’ (CDO). (Although in six of the examples to follow, whilst ACORN data is present online, the CDO daily information upon which it is believed to be based is not available online!).
The BoM homogenised data are officially known under the acronym; ACORN – SAT (abbreviated to ACORN here). In order to explore any anomalies in the quality of that data the methodology used was to electronically subtract the downloaded CDO daily data from ACORN and then chart their differences for visual comprehension of any strange patterns therein. The software employed was the widely accepted Microsoft EXCEL.
In firstly revisiting the Melbourne chart, it serves several purposes including that it is visually less complex than most rural station charts that follow, but is useful as an introduction to what are rather unusual graphics. Perhaps most unique is that daily data spanning >100 years, (or about 38,000 data points), are compressed across a single page width. There are consequential limitations in pixel definition and visual clarity, but nevertheless annual cycles are generally evident, and, in some cases, in the later charts some selected details are expanded in scale for improved definition.
Precursor chart for Melbourne (capital of Victoria) follows:
CLICK TO ENLARGE
|CLICK here for more information: §A Additional Notes for understanding the charts §B Some background on the BoM controversies §C Some comments on Melbourne and UHI effect|
Six charts follow exampling a plethora of problems at long-record sites. (CLICK to enlarge):
NOTE: In these charts, ‘anomalies’ are meant in the sense of incongruous outcomes in the differences obtained after subtraction of CDO (raw) temperatures from the ACORN (“homogenized”) temperatures.
Fig 1) Bridgetown P.O.
This 104 year-long single station record is one of the “World-Class-Quality” 112 ACORN sites. It was selected for having multiple concerns, although they are not individually the most extreme out of the examples to be revealed later.
- Area of interest ①in this chart shows in the ACORN adjustments over CDO, a sharply seasonal cycle of extremes in the maxima ranging some 4⁰C, whereas in the minima at less than 1⁰C, there is hardly any seasonal influence, (seasonal profile).
- Area of interest ②shows lesser adjustment but there is a starkly illogical difference in the seasonal profile each side of ~1982.
- These seasonal differences are in highly regimented cycles prolonging for decades without any apparent reason or logical thought to be discovered in the site history.
- There are 7 step-changes, of which only two might be justified by “a small move (10 m) in November 1935”. (10 m relative to what?)
- The large change in maxima temperature anomaly range in the 1920/30’s is not credible in proportion to the many other prevailing periods.
- CDO data are the same as ACORN in the maxima for almost 20 years but in the minima it’s almost 70 years. (BTW, Fig 10; Robe, has unexplained commonality going all the way back to 1910 or 104 years!)
These issues should not be found at even a single station in ACORN’s tax-payer funded “World’s Best Practice”.
Fig 2) Mildura:
- It was not emphasized in Fig 1, but a similar issue is found here and later that some site moves described in the ACORN catalogue are so vague, that any “corrections” seem to be rather “inventive”. (e.g. herewith see captioned history for; 1927, undated, 1933, 1943, between the two World Wars, and 1989),
- A new issue found herewith is a modest example at ①+① of data corrupted by CDO records being out of phase with ACORN by one day. (like Monday data being subtracted from Tuesday’s). A much more extreme example is found in Fig 8; Rutherglen; with its three prolonged blocks of corrupted data.
- The rather different looking annual profiles at <②> and prior to ~1946 are partly the consequence of much missing data, resulting in very many zero values.
- EXCEL found over 11,000 values missing in ACORN over the full 104 years, but they are predominantly before ~1946. (BTW it is possible, but impractical to plot that concern here, because it would visually over-clutter the image in this case).
“World’s best practice”?
Fig 3) Tibooburra:
This chart spans some 38,000 daily data points across a mere page width and consequently the limited pixel definition misleadingly results in apparent mass blocks of colour at ①&②. Thus a clarifying expanded detail view is provided to the right in this image.
- A stand-out issue is that highlighted <<<①>>> showing for almost a decade that the minima are devoid of data, whereas the maxima do have some data but still with many zero values.
- For some 22 years there are many missing data in both minima and maxima as elaborated in the expanded detail view.
- Missing data in ACORN over the full 104 years are found by EXCEL to embrace 5494 and 2630 days.
- Minima temperatures have been increased in ACORN, by an average of about 1⁰C (by eye) prior to 1940 together with unaccountable strong seasonal variations. This results in a cooling trend in the minima versus CDO.
- Maxima in the same period have been unaccountably reduced by about 0.8 ⁰C (by eye) which results in a warming trend versus CDO. (Remember that the popular global average warming over the past century is only of about that measure)
- Documented site changes shed very little light on these issues.
This long series single site record does not pass the ‘test for reasonableness’, particularly before 1960, and surely cannot conform to “World’s Best Practice”.
Fig 4) Cairns: Every station is unique in its data!
- Here, ACORN is the same as CDO in the minima back to a site move at the airport in 1992, but in the maxima they are unaccountably the same all the way back to about Dec/1934 (80 years). This arguably amounts to a loss of publically available information on how & why ACORN became the same as CDO in the maxima over such puzzlingly greatly varying periods! It gives rise to some potentially disturbing speculations!
- Prior to that mysterious 1934 event is revealed an astonishing underlying algorithm showing a monthly cycle that is devoid of the daily data that is foundational to its source. (!?) This is discussed in more detail in Part 2 of this series of studies (see link in References) and is seen in several other sites, including Rutherglen; Fig 8; following herewith.
- The eight step-changes here typically show little correlation with the documented site history. Notice for instance that in a documented move in “December 1929 from a relatively enclosed site to a slightly more open one” that the maxima were adjusted in ACORN upward, but the minima down.
This site record and the associated BoM practices do not conform to a credible claim of “World’s Best Practice”.
Fig 5 Kerang:
In this case, as an alternative graphical elaboration the missing data in ACORN are plotted as highlighted typically at ②, (this alternative graphic methodology becomes of greater significance in some later figures).
- In the two areas of interest ①&①, when attempting to plot anomalies, the ACORN results show actual temperatures in ⁰C (not differences) because there are no CDO data on record to be subtracted from ACORN. That is a tad odd because it is thought that ACORN is based on modification of that “raw” CDO data. (There are a whole bunch of these system failures, e.g. Figures 11/13/14/16/17 attached)
- In the expanded detail ④, it can be seen with careful inspection that there is a general clustering around monthly steps in each yearly cycle. (That’s a merging of daily data with a strange underlying monthly algorithm…..further to Part 2 in this series….see References for links)
Hardly “World’s Best Practice”!
Fig 6) Albany:
- The range of maxima adjustments in-town and the savage seasonal variations compared with the more exposed site at the airport after 1965 are seriously controversial and suggestive of corrupted algorithms or human error rather than anything daftly intentional. (BTW, Alice Springs; Fig 12 is an interesting comparator)
- Re-establishment of a site in-town to cross-calibrate the move to the airport some 37 years after the event is maybe; better late than never?
How can we be confident of our BoM in claiming its ACORN sites and practices are “World’s Best Practice”?
For figures 7 to 12 CLICK: (part3.1Figs7to12.docx)
For figures 13 to 21 CLICK: (part3.1Figs13to21.dox)
Summary comments on the long-record charts:
The focus of this study has been on rural stations having long records, mainly because the BoM homogenisation process has greatest relevance the older the data is. Preference has also been for single station sites for ease of process, but there are many stations of one name but which are actually multiple stations merged together, (commonly originally in-town location/s then moving to outlying airport location/s). There are also a significant number of ACORN stations with short records (E.G. 17% are with start-up dates between 1950 to 1975). Some of these are likely to have effectively no researchable data online because CDO is typically the same as ACORN over those shorter terms. Unfortunately to fully confirm this would be labour intensive and beyond my current intentions.
So….. back to our twenty-one long-record rural sites….. They have a great variety of non-credible data, sometimes of an astonishing nature, as is summarised in Table 1. It includes some subjective assessments, unlike the data itself in the charts, and there may be transcription errors despite repeated proofing. There are so many issues that it is too unwieldy to elaborate detailed conclusions:
Short-record sites are also important:
A controversial aspect that has been debated elsewhere is that the adding of new ACORN short-record stations* in hotter regions has resulted in exaggeration of recent warming trends. On the other hand, there are many abandoned older sites, (commonly at retired institutions such as Post Offices), which ought to be of importance in improving sample size for the declared objectives behind ACORN. However, prima facie this seems to have been seriously neglected by the Bureau.
*CLICK part3shorts.pdf For Table 2 (detail of nineteen post-1950 stations) together with three sample charts; Figures 22, 23 & 24
- The Part 1 study on six capital cities showed that the BoM’s ACORN DATA had no credibility, including illogical treatment for UHI effect. [Click above § C. …UHI effect]. However, rural site records are much worse in having a larger available sample displaying an even greater variety of problems. [Table 1 above]
- Almost half of the total 112 ACORN sites have been researched in Parts 1, 2 & 3 and yet more in continuing review, and ALL of them exhibit various data problems of similar severity to that exposed here for both long and short-term records.
- The ACORN DATA do not pass any test for reasonableness and thus cannot be properly used to assess any warming trends in Australia over the past century.
- These are forensic engineering assessments of DATA (not theory or speculative issues) using validated methodology and MS EXCEL software. [Click above § A for Additional notes… on methodology validation]
ACORN Station Catalogue. (Including history of sites involved…. Sometimes two or more entirely different locations)
Climate Data Online (CDO starting page with drop-down menus)
Part 1 of this series (subtitled; A tale of Six Cities)
Part 2 of this series (subtitled; Illogical Algorithms)
Acknowledgements and disclosures:
Thank you to Joanne for driving the format of this post to greater readability
I’m a retired mechanical engineer with no past or present funding for my subject research from anyone. (Or interests other than in wanting proper scientific integrity and effective use of government funding in science).
Compiled by Bob Fernley-Jones (Mechanical engineer retired) August 2015 Rev 14/Sep/2015