Who’s going to be sacked for making-up global warming at Rutherglen? Part 2.

THE Bureau of Meteorology has spent several weeks looking for documentation to support its recent claim that, contrary to what is written in its published station catalogue, the weather station at Rutherglen was once moved. Station Catalogue

The Bureau hasn’t been able to find any actual documentation to verify this move. But this hasn’t stopped it nevertheless concluding that there was a station move and publishing a document, not in the peer-reviewed literature, but on the Internet to conclude that the weather station was once moved between paddocks.

Furthermore, but not even published on the Internet, this move between paddocks, that could have occurred in 1966 or even 1974 apparently justifies a drop down in minimum temperatures, with the largest change a 1.8 degree Celsius difference between recorded and homogenised temperatures for Rutherglen in 1913. The net effect on the temperature trend is the creation of a 1.73 degree Celsius per century warming, where previously there was a 0.35 degree C per century cooling in the minimum temperature series.

It makes no sense!

But it hasn’t stopped the alarmists on Twitter and at HotWhopper claiming the Bureau was right all along: the weather station had moved, there are maps and photographs and conclusions, just no actual documented evidence. Indeed it doesn’t seem to have occurred to Lotharsson and the rest of the HotWhopper cheer squad, that if the Bureau was able to find inspection reports, requests for replacement equipment and more, that there should also be documentation if the station had actually been moved!

Of course it may have been moved, and the documentation may have been lost, but that still doesn’t justify the change in magnitude and direction of the temperature trend for Rutherglen that the Bureau has made.

My colleague, Bill Johnston, has forensically examined the temperature data generated by the weather station at Rutherglen. Dr Johnston is of the opinion that it is possible that there was a site move in 1965. He can find a break in the data at this time. But even accounting for this, there is no overall warming trend in the data either side of this break. So, Dr Johnston has concluded after running three independent statistical tests over the data, that it is the Bureau that has changed the data, resulting in a bias in the data series where none previously existed.

HotWhopper haven’t actually examined the data, or thought about whether moving a weather station between paddocks in a relatively flat rural terrain could cause a change in the direction and magnitude of the temperature trend. They are just celebrating that the Bureau could publish on the Internet claiming a site move.

According to HotWhopper, quoting the Bureau, the need for an adjustment to the Rutherglen data was made through the application of, “an objective statistical test that showed an artificial jump in the data during this period” prior to 1966.

In fact statistical tests cannot show “artificial jumps”. What empty heads they are at HotWhopper and the Bureau!

Statistical tests can detect breakpoints in temperature series. These may be attributable to climatic or non-climatic factors.

The Bureau claims that, 1966 and 1974 dates match the two breakpoints in minimum temperatures through a “statistical comparison of its [Rutherglen] data with other site data in the region.”

So, the Bureau is suggesting the “jump” is non-climatic and due to the weather station being moved because, while there is “no firm documentation” for a site move, 12 documents have been located that together, they claim, provide some circumstantial evidence for the weather station having been moved between paddocks.

Importantly, the Bureau is not claiming any breakpoints in the Rutherglen data per se, but rather that the trend at Rutherglen is not consistent with its neighbours at about the times for which there is circumstantial evidence the station was moved from one paddock to another.

It is in fact disingenuous and illogical for the Bureau to suggest that what could only be considered an insignificant move, if indeed it did occur, that is a move between paddocks in a relatively flat rural terrain could create a change in the direction and magnitude of the temperature trend.

But there is more. While the BOM claim a discontinuity based on neighbouring stations the algorithms and/or test that might show such a result are not disclosed. We are expected to believe this is the case, but this is not science because no methodology has been provided. In fact, when one of my colleagues, Ken Stewart, tested this proposition. He found that the raw data for Rutherglen has a virtually identical trend to its neighbouring comparison sites, while the homogenised ACORN-SAT data for Rutherglen is strongly biased towards warming relative to those neighbouring sites.

So, I ask again, who is going to be sacked for making up global warming at Rutherglen?

94 Responses to Who’s going to be sacked for making-up global warming at Rutherglen? Part 2.

  1. jennifer September 19, 2014 at 7:51 pm #

    So after ‘fighting’ the sign-in security at HotWhopper I posted the following comment which is yet to appear…

    “You guys are extraordinary… the Bureau have not actually provided any documentation for a site move. But let me be generous and consider that one might have occurred… as the Bureau suggests… between paddocks… how could this account for a change in the magnitude and direction of the temperature trend. More here… http://jennifermarohasy.com/2014/09/whos-going-to-be-sacked-for-making-up-global-warming-at-rutherglen-part-2/

    The Hotwhopper thread is here… http://blog.hotwhopper.com/2014/09/remember-weather-at-rutherglen-bom-was.html

  2. Emily Frazer September 19, 2014 at 8:24 pm #

    Hotwhopper did not post the link back to this ‘Denier site’. Instead they posted the link to the Bureau’s obfuscations.

    That is actually censorship by HotWhopper.

  3. Debbie September 19, 2014 at 8:46 pm #

    I went to the hotwhopper link.
    After feeling like I needed a shower. . .I can only say that Sou has drawn an incredibly long bow.
    That Loatherson is a pompous ass (as well as loathsome 🙂 )
    As I commented at the last link. . .BoM’s myopic focus on neatly packaging temp data . . .including truncating and homogenising it. . .is not useful to the people of Rutherglen or Amberly or Bourke or Williamstown or Bathurst etc. . .let alone any of us who work in the REAL Australian environment/weather/
    climate.

  4. Glenn Tamblyn September 19, 2014 at 8:58 pm #

    So, having to abandon your argument that the homogenization was unjustified because there was no station move – a qualitative argument – the BoM have produced reasonable circumstantial evidence that a site move did occur, so your argument now is ‘its just another paddock’. With a sort of waving my arms in the air sort of well whats a paddock or two between friends sort of thingy.

    Questions to be answered Jennifer:

    – What is the height difference between the two ‘paddocks’? Every 10 meters of height difference is 0.065 Deg C difference in average temperature.

    – What are the micro-site differences between the sites – hollows, wind shadow, etc.

    I don’t know the answer to these questions and I suspect neither do you.

    If there were no site change then Occam’s razor suggests that some other explanation needs be considered more likely.

    However, with a site change, Occam’s Razor suggests that the impact of a site change should be considered the more likely explanation for the observed changes. If you wish to suggest otherwise then some appropriate surveying of the site is required combined with some detailed thermodynamic measurements and calculations.The ball is in your court on that.

    However, your blog and public comments have been premised on the notion that there was no site change. BoM have put up a reasonable circumstantial case for the fact that there was a site change.

    Basic civility suggests that you owe them at least some sort of apology.

    The ball is in your court on this Jennifer but this is one of those basic test-of-character moments. You said there was no move. They have made a case that this was wrong. So fess up.

  5. jennifer September 19, 2014 at 9:11 pm #

    Glenn

    My original comment was accusing the Bureau of inventing global warming at Rutherglen through inappropriate homogenisation of the data.

    The Bureau responded by suggesting there had been a site move and this justified everything.

    So, the Bureau went from ‘just’ making up global warming at Rutherglen, to also, I suggested making up a site move at Rutherglen.

    We are now, some weeks later, provided with some documentation that doesn’t prove a site move. It proves there is documentation for Rutherglen, but where is the evidence for a site movement? Given there is documentation for site inspections, and new equipment, don’t you think there would be a document explaining the site move… if such a site move existed?

    But, even if it did… this hardly justifies creating a warming trend where none existed!

    Have you examined the temperature data that is so contested? There is no warming trend to be found in it… not before or after the claimed time of the site move.

    If, as you suggest, the ball is in my court I repeat my call for someone at the Bureau to resign over all of this, in particular for making up global warming at Rutherglen.

    Cheers,

  6. spangled drongo September 19, 2014 at 9:14 pm #

    Yes, Hotwopper Sou, please explain how you can claim warming at Rutherglen based on this pearl of science:

    “I talked at some length with someone at Rutherglen Research Centre. (When I was driving home in the wee wee hours of this morning, I was looking at the outside temperature on the dash, and thinking how interesting it is that just a few meters difference can hike temperatures up to 3 degrees Celsius and then they’ll drop back down to zero or minus one degree, when the terrain or vegetation changes”.

  7. Robert September 19, 2014 at 10:17 pm #

    Glenn, in triumphal mode, informs Jen that she must abandon her argument and apologise.

    Jen doesn’t need nearly as long as the BoM to make a straightforward case and make it perfectly clear to Glenn she needs neither to abandon nor to apologise.

    Also, she seems to publish adverse comments much more quickly and willingly than HotWhopper. What does that reflect?

  8. Glenn Tamblyn September 19, 2014 at 10:26 pm #

    “So, the Bureau went from ‘just’ making up global warming at Rutherglen, to also, I suggested making up a site move at Rutherglen.”

    So, are you suggesting they falsified the documents they presented. Because the documents presented are certainly reasonable circumstantial evidence of a site move. Absolutely certain no. But on the balance of probabilities yes.

    “But, even if it did… this hardly justifies creating a warming trend where none existed!”

    Unless there was always a warming trend and the site move obscured it!!! To say no trend existed you are assuming either no site move, or that if a site move occurred then this didn’t impact anything.

    “…where none existed”

    YOU DON’T KNOW WHAT EXISTED, not unless you have a Tardis. To make that statement is a massive example of a confirmation bias on your part!!!

    You only know what a combination of observations and site movements combined actually ‘recorded’. Do the measurements tell you what was happening at a measurement site somewhere? Yep. Do those measurements give you a broad understanding of what was happening in that region? Nope. Not if the measurement site moved during that time?

    You seem to be operating with an assumption that the raw numbers tell you what happened and anything else is a distortion of reality. When an equally possible answer is the raw numbers don’t reveal reality because several factors are folded together and only one of those factors is what you are interested in. To find reality you need to disentangle those factors

    The BoM are confident that they have nailed this. You say not.

    OK, demonstrate this. And that isn’t about just writing blog posts or doing speeches at the Sydney Institute. It isn’t even doing mathturbation on old datasets.

    It is about evaluating whether,
    a, A site move occurred – which is an exploration of historical fact, however obscured by time and lost records
    b. The thermodynamics/climatology of the implications of any such move which can be evaluated independent of the actual history..

    – Do the number crunching.
    – Take measurements at the two sites to determine site differences.between them thermodynamically and climatologically.
    – Dig deeper into the records. Do historical research in the region.

    Because it is up to you! If you allege something inappropriate on the part of the BoM – which is certainly the tenor of most of what you write – then the onus is on you to back this up.

    He/She who makes the allegation is under the onus of proof. And simply doing some mathturbation and going – ‘gee, I don’t get this’ isn’t proof.

    You may not have noticed but our legal system is based on the presumption of innocence until proven guilty.

    You allege ‘guilt’ on the part of the BoM. They don;t have to prove their innocence, you have to prove their guilt. So far you have failed. You seem to be assuming that you simply alleging something moves the onus of proof onto them. Sorry sister, it don’t work that way.

    Build a real case if you can, not just innuendo in the echo chamber of the blogosphere.

    As an aside, you list yourself as an adjunct research fellow at a University – I hope that is the correct title. Do you really think that trial-by-blog is actually a professional way for any academic to act?

  9. Robert September 19, 2014 at 10:41 pm #

    So, insufficient evidence of a site move. Not as outrageous as the broad-splatter homogenisation – but what is?

    This really is going the long way round to just say “whoops”. And we’re funding that long way round.

    By the way, don’t you love it when a VERY high use blogger disapproves of Jen making her case on blogs? But our Green Betters live easily with such contradictions.

  10. jennifer September 19, 2014 at 11:04 pm #

    Glenn,

    I think we need to get back to basics. Homogenisation is a technique that is commonly applied in climate science to remove non-climatic variable from the data.

    It is based on certain principles. Going right back to James Hansen (Journal of Geophysical Research, Vol 106, Pages 23,947-23,963), but also more recent papers for example, Lei Zhang (Theoretical and Applied Climatology, vol 115, Pages 365-373), it is clearly explained that homogenisation should be applied very judiciously and only when two criteria are fulfilled:

    1. A statistically significant discontinuity is found in a temperature series, and

    2. Metadata supports the discontinuity establishing that it is non-climatic in origin.

    So, even if a statistically significant discontinuity is found there should be no homogenisation unless it is established that it is non-climatic. Indeed the breakpoint/discontinuity may represent something real and climatic in origin.

    So even though Zhang and Hansen support homogenisation, they explain, as does Blair Trewin from the Bureau (see International Journal of Climatology, Vol. 33, Pages 1510-1529), that it can’t be applied willy-nilly and shouldn’t dramatically change the magnitude or trend in the original data.

    This is really all common sense.

    Yet in the case of Rutherglen:

    1. The Bureau did NOT find any statistically significant discontinuity in the origin minimum temperature series for Rutherglen. This should have been reason enough to leave the temperature series alone.

    2. The Station catalogue says there was no site move at Rutherglen. Another good reason for leaving the temperature series alone.

    3. Even if you say there is circumstantial evidence for a site move: why homogenise the data? A site move will not necessarily create a discontinuity, especially if it is just to a nearby paddock.

    Now Glenn, can you tell us:

    1. The statistical test that the Bureau did to the original data that established a discontinuity?

    2. Which metadata that the Bureau used to confirm the need for homogenisation back when the original homogenising of the data was done over two years ago now?

    If you could answer these two last questions, we might have a basis for some rational discussion. So, far the Bureau has been unable to provide answers to these basic questions. Also the Bureau has not denied that it has very dramatically change the magnitude and direction of the minimum temperature series at Rutherglen.

  11. A C Osborn September 19, 2014 at 11:42 pm #

    Just who is this idiot Glenn Tamblyn, coming on this hosts post and “demanding” information?
    He states “Unless there was always a warming trend and the site move obscured it!!! To say no trend existed you are assuming either no site move, or that if a site move occurred then this didn’t impact anything.
    “…where none existed”
    YOU DON’T KNOW WHAT EXISTED, not unless you have a Tardis.”

    Let me explain to Glen, the BOM’s own RAW DATA has been analysed, there is NO TREND in any of the RAW DATA, including the local comparison sites.

    Perhaps you don’t believe in DATA, which you are quite happy to accept PROVES that there might have been a site move when to others it doesn’t?
    You have the outright nerve to accuse Dr Marohasy of “Confirmation Bias”.

    Glen the Warmist TROLL.

    Do you get it now?

  12. mtc September 20, 2014 at 12:38 am #

    What I find statistically fascinating is that almost without exception, every adjustment to raw temperature data has resulted in increased warming regardless of the homogenization reasoning.

  13. Pathway September 20, 2014 at 3:36 am #

    Jennifer:
    In response to your question,” How is going to be sacked” if the Australian government is anything like the one here in the good old US of A, no one will be sacked, they will only be moved laterally or promoted. This is because they support the agenda of which ever party is in power.
    I would be surprised if BOM ever changes the record to reflect the actual data. I guess I am just old and skeptical.

  14. handjive of climatefraud.inc September 20, 2014 at 7:02 am #

    HotFlopper quote:

    “While deniers were chastising BoM for not having someone on hand to hop in a time machine, go up to the research station and look at where the weather station was in the 1950s …”

    Couldn’t the BoM use the same time machine they use to see future climate that is here now?
    Or did they design a progressive time machine that only goes forward?

    How can the BoM tell us that climate in 100 years is here now if they didn’t come back?

    You know it makes sense.

  15. Debbie September 20, 2014 at 7:26 am #

    The sad part is that people like Glen don’t seem to care that BoM’s treatment of the temp data is not serving any useful purpose to the Austalian public.
    He also seems to believe that BoM has some type of monopoly over the Australian weather/climate & any research into our climate.
    BoM has become so enamoured with their fancy & expensive equipment and so keen to package up data so that it can be reported like some type of sporting event that they have quite simply lost their way.
    We are having a systemic cultural problem.

  16. Farmer Gez September 20, 2014 at 8:17 am #

    Apart from a need to smooth temperatures (which is in itself a ridiculous pursuit), let’s look at the predictive skill of BOM.
    Eastern South Australia and Western Victoria are currently very dry indeed and at a decile one for rainfall. BOM predicted a average to below average for the period which is a decile four to five. They cannot with any certainty predict a three month period of rainfall nor can they for temperature. If a five day forecast is the best they can do, why on earth are you all worried about past revisions when their base skill of prediction is so very poor?
    What are the performance benchmarks for the BOM? Does anyone get sacked for poor performance?

  17. Jennifer Marohay September 20, 2014 at 8:26 am #

    Indeed Farmer Gez, getting rainfall cycles right is more important than understanding a few degrees change in temperature. Except the institutional failure that has resulted in the corruption of the temperature record is also affecting the skill of the rainfall forecasts. I explain how in some detail in my most recent letter to Senator Birmingham.

    I would very much appreciate you reading it carefully and if you agree, sending it on to your representative farm lobby.

    http://jennifermarohasy.com/wp-content/uploads/2014/03/Birmingham_2014_08_12.pdf

  18. JohnS September 20, 2014 at 9:27 am #

    Shouldn’t clear evidence of a station move come BEFORE even thinking about any adjustments, not be frantically searched for AFTER some adjustments have been questioned?

    Also if a station move did happen shouldn’t there be some serious attempt to gauge, on site preferably, exactly how much difference it would make to the readings, if any?

    …assuming the people doing the adjusting actually gave a damn about accuracy rather than creating a hockey stick.

  19. Debbie September 20, 2014 at 9:32 am #

    I hope people also understand that BoM has been given greater authority to report on federal water resources via the Water Act 2007.
    Their behaviour in this space is remarkably similar to the systemic institutional problem that Jen, Bill, Ken & others are highlighting re the temp data.
    Instead of making it easier and more user friendly for people who would need to access this data, BoM is working at packaging it all up into homogenized national averages that leads them to conclude such things as ‘2012 was 30% wetter than average’.
    While the ability to create national graphs and determine percentages above/below a
    pretermined national average is an interesting
    statistical exercise. . .it is of very little practical use for those of us who are working in this space.
    Further. .that figure is virtually meaningless because precipitation & catchment information is so highly variable on a regional & seasonal scale. Smoothing this into national yearly averages and presenting it as a package is NOT offering a good public service!

  20. egg September 20, 2014 at 10:00 am #

    ‘You seem to be operating with an assumption that the raw numbers tell you what happened and anything else is a distortion of reality.’

    They have artificially created a warming bias at a number of stations and from a personal point of view I would like an audit to see how much of the system has been corrupted.

    This distortion of reality (assuming the practice is widespread) leaves us vulnerable in the face of global cooling, which as you know is gathering pace.

  21. egg September 20, 2014 at 10:51 am #

    ‘You allege ‘guilt’ on the part of the BoM. They don;t have to prove their innocence, you have to prove their guilt. So far you have failed.’

    These are early days and there is still a lot of groundwork to do, that’s why a comprehensive examination of homogenisation needs to be done. BoM is on trial to determine its guilt or innocence.

  22. egg September 20, 2014 at 11:16 am #

    ‘That Loatherson is a pompous ass (as well as loathsome)’

    He was a regular at the Deltoid stables and other places of disrepute.

  23. Johnathan Wilkes September 20, 2014 at 12:24 pm #

    @gt
    ” the raw numbers don’t reveal reality”

    Now I heard everything!
    Then why do we bother collecting them when we can just make up our own data to suit the preconceived theory

  24. Ken Stewart September 20, 2014 at 1:38 pm #

    mtc: I’m afraid you’re not correct in saying that “almost without exception, every adjustment to raw temperature data has resulted in increased warming”. In 58% of stations for minimum temperature that is true. But the point is: what effect does this have on national trends? I say lots, the experts say little.

    Glenn: You appear to be claiming that as there MAY have been a site move, this alone justifies making an adjustment, as it was supported by a breakpoint.

    What about the adjustment in 1928? Any site move for that?

    I do not care whether there were site moves or not. I do exactly what the bureau says it has done, and compare with comparison sites. Differencing between Rutherglen and each of the comparison sites clearly shows the discontiuities mentioned by the bureau, and a couple of others as well. So perhaps adjustment was warranted? However the adjustments applied have completely distorted the relationship between Rutherglen and its neighbours, which you would have clearly seen if you followed the link Jennifer provided above. Before adjustment, and in spite of admitted discontinuities, there is very little difference between Rutherglen and neighbours. After adjustment, the difference is enormous and glaringly obvious.

    It would really pay you to check the data for yourself before believing everything you are told. Go ahead, check my findings too.

  25. Larry Fields September 20, 2014 at 8:44 pm #

    mtc September 20, 2014 at 12:38 am #
    “What I find statistically fascinating is that almost without exception, every adjustment to raw temperature data has resulted in increased warming regardless of the homogenization reasoning.”

    mtc has just raised the $64 question. Caveat: “almost without exception” may be an exaggeration; well over 50% may be more accurate. This applies to homogenized data, and to TOBs adjustments. Each of these things changes a slight, century-long cooling trend for Australia, and a somewhat shorter one for the USA, into warming trends.

    But let’s be chivalrous, and say that the raw data show a zero trend for our two countries. I’m also assuming that the real raw data for each INDIVIDUAL temperature station still exist, and that they are not Top Secret, classified National Security secrets.

    Then we could apply the Sign Test to test the Null Hypothesis that the numbers of POSITIVE departures (heads) from the zero trend, compared with the number of NEGATIVE departures (tails), is due to chance. It’s elementary, my dear Watson.

    I nominate climate sleuth Steven Goddard for the job. Be sure to stock up on popcorn.

  26. Svend Ferdinandsen September 21, 2014 at 2:36 am #

    Have BOM made a circular homogenisation?
    Think of the following: They somehow find Rutherglen need some adjustment when compared to other stations. After the adjustment Rutherglen must be concidered a gold standard, so the other stations could be adjusted relative to the new Rutherglen gold standard.
    I hope at some time the homogenisations and the method is released for external review.
    From daily, weekly and other averages in Denmark, you can have large differences, and Denmark is a very small country surrounded by sea, so to use surrounding stations for any correction must be done with the greatest care.

  27. Oliver K. Manuel September 21, 2014 at 5:01 am #

    Thank you, Jennifer, for asking the BOM a question the US Congressional Space Science & Technology refuses to answer:

    “Why Did You Deceive Us?” Message to the Congressional Space Science & Technology Committee (Dec 2013):

    https://dl.dropboxusercontent.com/u/10640850/WHY.pdf

  28. Daryl McDonald September 21, 2014 at 7:14 am #

    WOW! Maybe the IPCC models have been right all along.
    If you feed dodgy inputs into the most perfect models, inevitably you will get dodgy outputs.
    In Engineering, we work often have to work with incredibly tight tolerances.
    The machines/devices we design often have life or death implications.
    Bridges, structures, automobile steering/braking systems, medical equipment.
    The analysis of loads/stresses/failure points is best broken down into respective
    key elements, where, in the main, the absolute laws of physics/thermodynamics/applied mechanics are applied. Precision is critical.
    Where possible, measurements are always taken from a known zero or datum point.
    Instruments are constantly calibrated, and repeatability of measurements is critical.
    Have you ever thought about the precision in Formula 1 racing. consistent lap times within a 1/10 sec in a 100+ sec lap, for 50+ laps. Outcomes PREDICTED with 0.1% accuracy.
    If an Engineer was in charge of Met station site move/change in conditions, he/she would simply take temp readings before and after and compare for discrepancies.
    Any adjustments would be made on empirical data, not some algorithmic assumption.
    Agency scientific bodies have become very much like the automatic control systems we design. A control system relies on both positive and negative feedback signals. POSITIVE SIGNALS MUST BE ATTENUATED, and balanced against a negative feedback signal,or otherwise the system will catastrophically fail.
    It seems we have created a funding/feedback system for agency science that only provides positive feedbacks.
    The cracks are appearing.

    Cheers, D Mac.

  29. Lawrence September 21, 2014 at 7:18 am #

    Jen you and Steve (Tony) are bloody fantastic don’t let it go.

    I really believe that AGW is the most politicised issues of our age, where the left (sorry but it has to be said) absolutely love the idea that modern capitalist society (the west and yes Australia is in that category even though geologically thousands of miles adrift) is ruining our planet and provides one of the most powerful reason to end capitalism.

    It is a religion that holds the rational luxury of worshiping a cause as opposed to a god. Therefore over the decades UKMO, BOM, NASA and so on has recruited not scientist but crusaders with a spun science degree earned in an age of AGW which provided a mission to prove the AGW is happening no matter what. So at every point of new data, new knowledge, new phenomena like the amazing breath taking record Antarctic se ice that we are seeing at the moment, these people either silence such info and if they can’t they are so convinced of AGW that they will make up a scientific reason which is due to AGW why the sea ice is at record levels.

    However I’m just a humble jumble of words but you with your scientific background and pedigree can nail these bastards down and prove that there is now such a bias haunting BOM, UKMO, NASA and all institution around the world that it is a spin that dare not speak its name. Get in there Jen because this goes further than BOM.

  30. Debbie September 21, 2014 at 7:29 am #

    Circular?
    Yep. . .that’s one good way to describe BoM’s behaviour.
    🙂
    Thanks for removing my earlier comments Ray. I did step over the line. The comments from that PA at hotwhopper & deltoid were incredibly disconnected and superficial BS.
    I should not give people like that any oxygen.

  31. Daryl McDonald September 21, 2014 at 11:20 am #

    Update to comment about Formula 1.
    In motor arcing, and many other fields of private enterprise, ideas and theories are tested in real world competitive situations. Duds quickly fall bf the wayside.
    How would junk science, supposition, and dysfunctional computer models fair in such circumstances???

    Cheers, D Mac.

  32. Another Ian September 21, 2014 at 11:32 am #

    Jen,

    Re sacking.

    Remember the phrase “Good enough for government work”, and so probably nobody.

  33. egg September 21, 2014 at 12:01 pm #

    ‘I should not give people like that any oxygen.’

    Loth and the Deltoidians are a formidable bunch, its unfortunate that they won’t come over here and have a discussion, because I can’t get on there or Hotwopper.

    Leaving me chafing at the bit.

    At the moment Loth is probably in sleep mode on the other side of the planet, but when he reads this we can expect him to man up and argue his case, instead of hiding in Tim’s old men shed.

  34. Ian George September 21, 2014 at 12:25 pm #

    How quickly do the electronic AWS thermometers react.
    Casino shows the temp as either 23.5C or 22.8C for 11am today.
    BoM site ‘Latest Weather Observations for Casino’.

    This site shows the highest point of 23.5C at 11:00am.
    http://www.bom.gov.au/nsw/observations/nswall.shtml?ref=hdr

    This shows the temp at 22.8C – also at 11:00am.
    http://www.bom.gov.au/products/IDN60801/IDN60801.94573.shtml

    Bet the old therms couldn’t do that.

  35. Richard C (NZ) September 21, 2014 at 1:11 pm #

    JohnS September 20, 2014 at 9:27 am #

    “Shouldn’t clear evidence of a station move come BEFORE even thinking about any adjustments, not be frantically searched for AFTER some adjustments have been questioned?”

    Yes that’s the sensible course of action that comes to mind first John. Except BOM (and BEST) have put their statistical breakpoint analyses first assuming all is well and reliable and only detail station vagueries in retrospect when pressured with contentions that all is not well i.e. they’ve been caught out. Pre-1975, the Rutherglen Research screen may have been on wheels for all we know from the documentation presented.

    The manual equipment was removed 12/12/2012 and the current site looks to be on a concrete plinthe from the photo (Figure 2b):

    http://www.bom.gov.au/climate/change/acorn-sat/rutherglen/rutherglen-station.shtml

    What was the setup for the manual equipment? What was the setup if it was somewhere else? I don’t find the 400m vs 700m argument convincing either.

    “A reference in the November 1958 inspection report (Figure 7b) to the site being ‘1/4 mile from office’ (400 metres). This differs considerably from the present site which is 700 metres from the office.”

    It doesn’t differ “considerably”. Distance estimation is subjective (I’ve worked with this in civil roading/structures/hydro etc, and electrical line construction both overhead and buried with land surveys and surveying equipment), maybe an experienced land surveyor might make a reasonable guess accounting for treelines obscuring line-of-sight but unless the estimator has actually measured out a similar distance a number of times then they will probably be wildly out, even by 300m.

    But for New Zealand, NIWA’s NZT7 approach is exactly as you ask John. Adjustments were/are only made at documented site changes. In the NZCSET v. NIWA court case there was never any contention over site changes, they were agreed from the documentation. The contention was, and still is, that NIWA has never been able to provide their adjustment methodology to be replicated. That’s because it is only very loosely based on Rhoades & Salinger (1993) but they don’t adhere to the methodology, A location series compiled in accordance with R&S93 returns adjustments and a trend different to NIWA’s for all 7 locations but not trend reversals like BOM’s.

    NZCSC/NZCSET compiled the NZT7 in accordance with R&S93, had it reviewed by 3 independent professional statisticians, and the Judge effectively dismissed it without consideration preferring instead the word of Dr Brett Mullen (climate scientist, not a statistician) that NIWA’s method was valid.

    This does highlight the completely different (in reverse) approaches on either side of the Tasman. Given the uncertainty (“While no firm evidence exists as to the exact location of the site in 1958 and earlier years” – BOM) I don’t see how BOM’s homogenization and adjustment can be valid for Rutherglen pre-1975.

  36. Pauly September 21, 2014 at 2:20 pm #

    Jen,
    This information is excellent, but I think that we may be looking for data that confirms our beliefs, good old fashioned confirmation bias.

    As an alternative we should present a falsifiable hypothesis. My suggestion is that the hypothesis that we should test is:
    “Homogenisation does not alter the overall trend for Australia”
    i.e. Homogenisation may move the trends of some sites upwards, but others will be moved downwards.

    The test shouldn’t be finding stations where the trend has been homegenised upwards, because if homogenisation has no net effect on the trend then logically there has to be some sites where this happens. The test should be finding sites where homogenisation has changed the trend from higher warming to lower warming.
    For every Rutherglen, Williamstown, Darwin or Amberly where homogenisation has caused a net warming trend logically there must be other sites whee homogenisation has caused net cooling effects (assuming homogenisation of the data does not affect the overall trend).

  37. Richard C (NZ) September 21, 2014 at 2:24 pm #

    BOM – “A 1953 document (Figure 5) which refers to a hill of over 700 feet (213 metres) approximately 300 yards (270 metres) to the south of the site. No hill exists for several kilometres south of the current site. The possibility that the ‘300 yards to south’ is a reference to the station buildings, not the observing site, can be ruled out as the hill on the Rutherglen property—whose height closely matches this description—is to the west, not the south, of the buildings. (The coordinates listed are not useful in accurately determining the site location as they are given only to the nearest minute, and would therefore only specify the location to about the nearest kilometre).”

    Does “the hill on the Rutherglen property” really match the description? What about the description? I know exactly what a volcanic cone 232m above sea level looks like (Mt Maunganui, Bay of Plenty, NZ). I see it from my kitchen window every day. I’ve walked up it, run up it, it is a landmark for miles around. It is more than a “hill”. I don’t think the buildings would be on a similar prominence albeit 19m shorter.

    The current site is already 175m above sea level so the “hill” is probably about 38m higher than the surrounding land. The description is here:

    http://www.bom.gov.au/climate/change/acorn-sat/rutherglen/rutherglen-fig5-tn.png

    BOM have read this incorrectly. I see:

    “Height of buildings approx 640 feet” – 195m. If that’s top of buildings it puts them only 20m above the current site in terms of sea level. Single story say 5m, hill 15m.

    “Hill runs to over 700 feet approx 300 yds to south” – South of what? The buildings? A 15m rise runs 213m long (not up) for 270m sth of the buildings?

    From previous >”Distance estimation is subjective”

    A range finder or vehicle odometer may have been used but I’m inclined to think the distances and elevations are very rough guesses.

    I’ll have to come back to this – got to rush off and work a night shift.

  38. BrianL September 21, 2014 at 3:40 pm #

    FWIW, google lets me view a few buildings just off Chiltern Valley Road, at 243 Research Station Lane – at -36.103980, 146.515758.
    BOM shows location as 36.10S, 146.51E – wonder if they are the same location?
    This site is apparently on only very slightly sloping ground, at almost the lowest point in the immediate vicinity – going by the small gully to the south and the surrounding cultivation.
    Only hill I can see is approximately 773 metres to the northwest.
    Question is – exactly where is the Stevenson screen located – is it within an 800 metre radius of the 243 address?
    BrianL.

  39. BrianL September 21, 2014 at 4:12 pm #

    But, if the hill I ‘think’ I see is approximately 4 mi SE of the town, then using Richard C(NZ)’s info, maybe it is located near the intersection of Stillards La or thereabouts, further to the north-ish.
    The BOM location, as stated previously, covers a fairly large area.
    Of course, I may be using the wrong type of aerial device to flit around and can’t see the hills for the cultivation – old age to blame.

  40. Graeme M September 21, 2014 at 8:02 pm #

    On reading the BOM page, I suspect they have made a few assumptions that may not be borne out, but who knows. Examining the location via Google etc is unlikely to show much. It would seem to make more sense to do some research into whatever state or local records can show. I think I’ve read that the station in 1953 was later replaced by a different building in a different location. Or find the people whose signatures appear on the inspection reports.

    One of the more telling pieces of evidence is the report that shows a woolshed 150 feet away with an evaporation tank about halfway between the two. A visit to the location might uncover evidence of those structures if the screen remains in the same location today. Google earth suggests the woolshed would have been across the road in the adjacent paddock, while the evaporation tank would have been where there is some object on the satellite image not far from the road. I think too that the coords for the site in 1953 match the GPS coords of the screen today.

    Still, the best evidence will come from either eye-witness accounts or documentary evidence of where the station building was in 1953.

  41. Kevin ONeill September 21, 2014 at 9:52 pm #

    Submitted this over at JM’s blog.

    In a perfect world homogenization would not be necessary. All stations would have been located several hundred years ago in ideal locations that, with crystal ball foresight, would remain essentially unchanged over the ensuing centuries so that future scientists would have pristine data to work with. Sadly, that is not the world we live in.

    Instead we are faced with numerous confounding factors – uneven distribution of stations, differences in Time of Observation, changes to local microclimate, different instrumentation, changes to instrumentation and station relocations come immediately to mind. As Menne (2009) write: “Unfortunately, changes to the circumstances behind a series of climate observations are practically inevitable at some point during the period of record. For this reason, testing for artificial discontinuities or ‘‘inhomogeneities’’ is an essential component of climate analysis.Often, the test results can then be used to adjust a series so that it more closely reflects only variations in weather and climate.”

    Now, one could just take the raw data, ignore all the known errors, and produce a graph and call it a day. That would be the easy way out. Of course comparing one year to the next, especially as the distance in time grew larger, would be comparing apples to oranges. The result would be of little practical use to anyone. Those who rail against homogenization are asking us to accept seriously flawed comparisons as the best we can do.

    Science doesn’t work that way. Sometimes one can take flawed data and make it better. Homogenization of scientific data has been around at least since the 1850s (Kreil). The various methods of homogenization used to produce temperature series have withstood peer review and numerous ad hoc studies. The peer reviewed literature is vast. Math works. Homogenization works.

    Does this mean the resulting homogenized data is perfect? Of course not. Expecting it to be perfect is unrealistic. And despite the fact that pseudoskeptic usually only cite stations where the homogenization lowers past temperature data, it doesn’t work that way. many stations see past data increased as Nick Stokes at Moyhu has ably pointed out.

    Those who have complaints about homogenization ought to devise a better method and submit it for peer review.

  42. Daryl McDonald September 22, 2014 at 6:35 am #

    For those out there that have seen the movie ‘THE DISH’.
    For those that haven’t, please go see.
    Remember the power failure just before the DISH was to be aligned to receive
    that incredible footage of Neil Armstrong stepping on to the Moon.
    The computers went down.
    ‘OH, how do we aim the DISH now?’
    Well they just stepped out of their office, and the Moon was staring them right in the face.
    Just aim the bloody thing directly at the Moon.
    No need for fancy computer algorithms.
    Direct, first principle measurement trump artificial intelligence every time.

    Cheers, Slowlurnr.

    Now Daryl, you know it can’t be like that. Your idea is so simple it must be wrong. What would happen to all the scientists?? Think of their children starving in a miserable garret, not to mention the kids of the army of bureaucrats living high on the CAGW religion. Moderator Ray

  43. Debbie September 22, 2014 at 7:36 am #

    Graeme M.
    The issue that is emerging is far more concerning than a speculative argument about a possible site move at Rutherglen.
    BoM has clearly focused on neatly packaging temperature data. . . and using those algorithms just because they can.
    The statistical exercise is no doubt very interesting but. . as far as providing a useful service to those of us who work in the Australian weather/climate/ environment year in and year out and who need BoM to focus on seasonal conditions and improve skill in this space. . .this homogenising and truncating and reporting national averages is an epic fail!
    As I commented earlier, BoM is behaving in the same manner reporting on water resources. Because they have this myopic focus on producing National averages they have made it increasingly difficult for us to access the information we actually need.
    A yearly national average is a virtually meaningless figure in a climate that is seasonally and regionally variable as Australia’s.
    As an example. . .BoM’s conclusion that 2012 was 30% wetter than average . . .does nothing to inform the Australian public about the why’s and wherfores of when and where those precipitation and inflows occurred or how they influenced Australia’s productive capacity.
    It was far far wetter than average in the Riverina between 1st Jan and 31st Dec 2012 . . .BUT!. . . it didn’t rain in Spring and the dry area crops were not good.
    BoM is ‘sweating the small stuff’ trying to speculate about a possible site move in Rutherglen. It does not justify changing a slight cooling trend in the raw data to a warming trend.
    As someone mentioned earlier their argument is depressingly circular. It is also full of bureaucratic double speak which sees them focusing on answering a question that wasn’t asked.

  44. Ian George September 22, 2014 at 7:54 am #

    Jennifer
    You may be interested in this from the BoM ACORN site. Just added.
    Re Aust temps prior to 1910.
    ‘This work has shown that temperatures in the southeast between 1860 and 1910 were similar to those experienced during the first half of the 20th century. This confirms that the warming trend observed in the Australian region since the mid-20th century has been historically significant.’

    See here at
    http://www.bom.gov.au/climate/change/acorn-sat/#tabs=Early-data

    Just a quick glance at the max temp map shows 1896, the Fed drought and the 1914-15 periods below the temps of the 1920s.
    Lots of work here for yourself and Ken et al.

  45. BrianL September 22, 2014 at 7:55 am #

    Aww, gee, I’m a drongo – my brain, fingers and eyes just weren’t all that well co-ordinated yesterday and my (un)trusty computer wasn’t letting me see all the comments correctly – so I opened my mouth, put both feet in and waggled like mad.

    If Ray the MOD was feeling like a kindly soul he would delete all of my previous comments on this thread for me and I could disappear back into the aether.

    I observe the mercury this morning to be 16.4*C – I wonder if that should be homogenised before I write it down in my daily diary, to allow for possible RHI due to site problems? (R for residence on a grazing property). Then there is the 10.5 mm of lovely soaking rain that fell overnight that may also need to be adjusted, perhaps? Grass, fruit trees, etc may benefit if we received more than we think – or vice versa.
    /sarc.

    As an old ex-farmer of no repute, I’ve always found it beneficial to keep records of the weather as it pertained to me. I suspect there are thousands of diaries in existence that have many years of weather observations contained within their pages – ‘If only’ they could be collected and collated…. but of course they would be of little use as they would be considered to be so totally inaccurate as to be useless.

    Funny how farmers need to know whether the weather is weather – or not.

    I had always been under the assumption that the BOM was really about producing weather reports for aerial & shipping services and civilian usage was a by-product.
    BL.

  46. Robert LePage September 22, 2014 at 9:05 am #

    Now if you move the 3rd row of deckchairs towards the back of the ship, when the Titanic sinks they will be safer.
    It is important to get these small details right even if they make no difference to the end result.

  47. Ken Stewart September 22, 2014 at 10:03 am #

    Pauly: But your null hypothesis is exactly what we are testing. And for every Rutherglen, Amberley, Williamtown etc with spurious warming homogenisation there is NOT an equal number that have been cooled. Yes, a number of sites have been cooled, but the majority have been warmed, and the overall national trend has been changed. Do you really think we haven’t already checked? Stay tuned, you may be in for a few surprises.

  48. Ian George September 22, 2014 at 10:32 am #

    Re BoM’s ‘Raw Data’ site:
    On closer examination, 1914 looks about right except it’s shown below some later years. According to BoM’s time series, the hottest year (Tmax) for SE Aust was 1914.
    http://www.bom.gov.au/cgi-bin/climate/change/timeseries.cgi?graph=tmax&area=seaus&season=0112&ave_yr=0
    Then I saw the magic words –
    ‘Reconstruction of climate variability for southeast Australia from homogenised pre-1910 records’.
    Aarh – ‘reconstruction’ and ‘homogenised’.
    That explains it.

  49. Robert September 22, 2014 at 10:34 am #

    RPL, it’s more like you’re arranging life-rafts on Al Gore’s (twin) giant houseboats

  50. Kevin ONeill September 22, 2014 at 12:13 pm #

    Ian George – should one correct for known errors/bias in the raw data?

    If yes, that’s homogenization.

    If no, then you’re not doing science.

    It’s that simple.

  51. Richard C (NZ) September 22, 2014 at 12:21 pm #

    OK, had some time to check BOM’s evidence. It does support the contention that there was at least one other site except that it doesn’t reveal what data was taken from where and when and whether there were overlaps along with Jennifer’s problem (see below).

    @ Brian. The current site is about 130 -150m south down the dirt road from the intersection with Research Station Ln. See the site map:

    http://www.bom.gov.au/climate/change/acorn-sat/rutherglen/rutherglen-fig2a.png

    The east-west line of trees to the north of the site run along the side of Research Station Ln. The dirt road is first on the left west along Research Station Ln from the Chiltern Valley Rd intersection. You can actually see the installation in satellite view by zooming right in.

    @ Graeme. >”Google earth suggests the woolshed would have been across the road in the adjacent paddock, while the evaporation tank would have been where there is some object on the satellite image not far from the road.”

    Yes, that’s this map (Figure 6):

    http://www.bom.gov.au/climate/change/acorn-sat/rutherglen/rutherglen-fig6-tn.png

    I can’t see those features at either the current site or BOM’s “former” site. No idea where this “other” site might be and yes a site visit might find where.

    @ Kevin >”Those who have complaints about homogenization ought to devise a better method and submit it for peer review.”

    You’re not up to speed with what is being investigated Kevin. Homogenization as a concept is not the problem (and Jennifer explicitly stated the understanding of the need for adjustments in a post at the outset of this latest furore). The problem is that every institution has a different methodology and they all produce different results for the same application. Then there are the quirks within each respective methodology that defy replication which is the essence of the argument I think. It should be possible to replicate a homogenized series independently from the raw data by established methods and the documentation. That proved problematic in the case of NZCSC being unable to replicate NIWA’s NZT7 by application of the established R&S93 methodology. NIWA are still unable/unwilling to provide their own method as I’ve described upthread. And it is proving problematic with ACORN-SAT.

    # # #

    A 1953 document (Figure 5) does support BOM’s “former” site now that I’ve had more time to look. That’s this:

    http://www.bom.gov.au/climate/change/acorn-sat/rutherglen/rutherglen-fig5-tn.png

    I thought it said “runs” but it does say “rises” wrt the “hill”. And it does place the “former” site north of the hill and therefore the main buildings rather than to the south of them. The data was “taken from 1:1 Military Survey Map” so those distances and elevations can be considered reliable.

    So now we have 2 sites, current south of, former north of the hill which have radically different microclimates in Min but not Max according to BOM. But adjustments to Min-only effectively adjusts Max by way of the Mean. And if ever there was (and still is for proof of method – see below) an opportunity for a data overlap in order to create a contiguous series, this is one.

    Jen’s problem:

    >”…this move between paddocks, that could have occurred in 1966 or even 1974 apparently justifies a drop down in minimum temperatures, with the largest change a 1.8 degree Celsius difference between recorded and homogenised temperatures for Rutherglen in 1913. The net effect on the temperature trend is the creation of a 1.73 degree Celsius per century warming, where previously there was a 0.35 degree C per century cooling in the minimum temperature series.”

    Just to put this net -1.8 C adjustment in perspective, here’s the “raw” data series from Jen’s link: ‘Annual mean minimum temperatures, before homogenisation, for Rutherglen’:

    http://jennifermarohasy.com/wp-content/uploads/2014/08/Mean-Min.png

    Net -1.8 C drops the 1913 datapoint off the bottom of the chart from 7 C down to 5.3 C. But look at the intermediate Min adjustments:

    http://jennifermarohasy.com/wp-content/uploads/2014/08/Screen-Shot-2014-08-27-at-4.22.55-AM.png

    1957 drops from 6.3 to 4.3 (-2). 1929 drops from 6 to 4.5 (-1.5). 1925 drops from 6.3 to 4.7 (-1.6). All by eye.

    1939 drops from 8.7 to 7.5 (-1.2). 1920 drops from 9.2 to 7.3 (-1.9). Again all by eye.

    Now this adjustment process is OK in concept (for terms of current site) but were Min temperatures really that much warmer on the north side of the “hill” (a mere 15m rise) compared to the south side when Max wasn’t? I don’t think so.

    The impasse could be broken simply by installing a SS at the “former” site and comparing an overlap for proof of method (or not). This would settle the argument for me. Sure there’s been some changes over time but there would still be somewhere clear of influences in the vicinity. This was suggested for Wellington when the location site moved across the city and up considerable altitude. Never done of course.

    Empirical proof of statistical method required I think, and it can be easily done for at least this one test case.

  52. Richard C (NZ) September 22, 2014 at 12:34 pm #

    Here’s BEST’s Rutherglen Research adjustment effort:

    http://berkeleyearth.lbl.gov/stations/151882

    No data prior to Jan 1965 (they pull in Rutherglen Post Office prior to that), and their major trend reversing adjustment occurs 1980 due to a “Record Gap”. Oddly, BOM don’t adjust for that.

  53. Ian George September 22, 2014 at 2:47 pm #

    Kevin
    You, therefore, should be able to explain the following adjustments to Bourke for Jan 1939. All temps over 30.0C have been adjusted down and all those under 30.0C have been adjusted up.
    AWAP ACORN
    1st 38.9 38.4
    2nd 40.0 39.1
    3rd 42.2 41.9
    4th 38.1 37.9
    5th 38.9 38.4
    6th 41.7 41.5
    7th 41.7 41.5
    8th 43.4 43.0
    9th 46.1 45.7
    10th 48.3 47.9
    11th 47.2 46.8
    12th 46.2 45.8
    13th 45.7 45.3
    14th 46.1 45.7
    15th 47.2 46.8
    16th 46.7 46.3
    17th 40.0 39.1
    18th 40.1 39.1
    19th 40.0 39.1
    20th 41.9 41.7
    21st 42.5 42.1
    22nd 44.2 43.8
    23rd 36.7 36.5
    24th 40.3 39.2
    25th 36.6 36.5
    26th 29.4 29.5
    27th 29.3 29.4
    28th 28.8 28.9
    29th 30.6 30.5
    30th 35.6 35.4
    31st 38.6 38.3
    Mean40.4 40.04
    Did they moved the site everyday?

  54. Jennifer Marohay September 22, 2014 at 2:58 pm #

    Pondering descriptions of hills rising, or not, in records from site inspections at Rutherglen… looking for something else I just noticed how the hills to the west of the weather station at Cairns are described in the current official ACORN-SAT Station Catalogue ( http://www.bom.gov.au/climate/change/acorn-sat/documents/ACORN-SAT-Station-Catalogue-2012-WEB.pdf )

    “The current site (031011) is located at the Meteorological Office, on the western edge of Cairns Airport, on level, open ground over mown grass. A mountain range rises steeply within a few hundred metres to the west.”

    Surely they could be a bit more specific with respect to when the mountain range rises steeply, and isn’t it more likely a few kilometres to the west?

    ALSO, surely someone who reads this blog lives not too far from Rutherglen, and could go and have a look for some footings that once held-up that woolshed?

  55. Pauly September 22, 2014 at 3:19 pm #

    Ken,

    Trust me when I say I will not be surprised, but what I have seen so far is that there is good evidence that homogenisation can cause some stations to appear to be warming faster than the raw data.

    Based on my experience with data analysis in the insurance industry I would say that if Homogenisation is being done properly one could expect things like:
    – Known Urban Heat Islands in major cities to be revised downwards
    – Examples of isolated sites being drastically reduced downwards. In a process like homogenisation there should always be some outliers at each extreme.
    – Good explanations for most outlier results
    – Whilst random variation will play its part in creating some crazy head scratching individual records, the overall data set is large enough that the random events will not significantly overall trends and averages.
    – No significant difference between the average of the raw data as a whole and the average of the homogenised data.

  56. Ken Stewart September 22, 2014 at 5:07 pm #

    Pauly:

    I would have to agree wholeheartedly with your comment that
    ” if Homogenisation is being done properly one could expect things like:

    – Good explanations for most outlier results

    – No significant difference between the average of the raw data as a whole and the average of the homogenised data.”

    But the discussion here is all about the ‘good explanations’ for adjusted data that are completely out of sync with the comparison stations used for homogenisation.

    And you won’t be surprised when you see national averages of ‘raw’ vs Acorn? It depends on your definition of “no significant difference” I suppose. What difference would you define as significant? +/- 5%? 10%? Please tell.

  57. Mikky September 22, 2014 at 7:35 pm #

    The BoM run algorithms to detect sudden changes in temperature data, fair enough, but they fail to deal with the issue of false alarms, i.e. changes due to the weather. Trewin admits this (indirectly) in his document CTR_049.pdf. They assume that ALL detections are genuine, all they check is metadata, and if there was any sort of change near the detection then it acquires a * in the documentation.

    It could well be the case that there are more false alarms than genuine detections, in which case the BoM procedure makes the data worse not better.

    The BoM procedure has a fundamental flaw in the methodology, the absence of any manual checking of detections. The algorithms being used may be wonderful, but ALL detectors have false alarms, i.e. ALL detections should be checked by experts on meteorology to see if they could just be weather changes.

    1974 was an exceptionally wet year in Australia, so surely the Rutherglen detection in that year may just be due to that.

  58. Graeme M September 22, 2014 at 7:47 pm #

    Jen, my point earlier re the woolshed etc is that if you take the sketch in fig 6 and adjust Google Earth to match orientation and scale and overlay them, the woolshed would be found to the ESE of the current site if there was no move since 1953. The evaporation tank aligns with some structure that can be seen on Google Earth. If you really want to establish whether the current site is the same as that in 1953, that’s where you’d start to look for old footings etc.

    BOM suggests that the woolshed north of the current buildings might match the old map, but the orientation of that shed seems not to match.

    Of course the likelihood that the site has or has not moved is less important than the homogenisation methodology and practice, but it would nonetheless be most illuminating if it could be established beyond doubt that the site has NOT moved.

  59. gavin September 22, 2014 at 9:03 pm #

    Ian George; Looking at Casino today, 24.2 C @ 1.59pm is the high and 23.3 is the temp at 2pm in the site time series issued at 8.35 tonight, right?

    Now I was wondering about your similar post a few weeks back and decided in ignorance, those temps then had to be from different instruments. They are not different instruments, just different days with the high comming from the previous 24 hours.

  60. Pauly September 22, 2014 at 9:32 pm #

    Ken,

    Firstly, my experience with data analysis is in insurance, and in insurance we never used homogenisation because that would screw up what was happening with individual cost centres. We had our own actual money on the line so it was important to know who was doing what.

    But every now and again you would have an outlier result that defied reasonable explanation, and we chalked those up to random variation, especially if they were not repeated. For example if there are 5 outlier sites I would expect that there should be easily understood and valid explanations for at least 4 of them.

    As for how for the homogenised trends should differ from raw data trends, the acceptable answer is 0.00% By definition homogenisation means to have everything redistributed equally.
    At a deeper level a weighted homogenised trend can differ from raw data trends, but there needs to be transparent and open explanation of why some sites should have a higher weighting than others, and the degree of weighting should be documented.

  61. Ian George September 22, 2014 at 10:12 pm #

    Gavin
    Same instrument, same day, same time period.
    I go to ‘Latest Weather Observations for New South Wales’ at:
    http://www.bom.gov.au/nsw/observations/nswall.shtml?ref=hdr
    Then click on the town for a 30 min interval update.
    My point is that the electronic thermometer at an AWS can pick up sudden surges in temp over extremely short periods (secs in some cases) which the older thermometers could not.

  62. Mikky September 22, 2014 at 10:38 pm #

    A question for Jennifer:

    You said in a comment above that BoM did NOT find statistically significant discontinuities for Rutherglen. Do you have a reference for that, I’ve not seen that anywhere else?

    I would have thought that the detection algorithms would only report events with statistical significance (but as you say that does not distinguish between moves and weather changes).

  63. jennifer September 22, 2014 at 11:58 pm #

    Hey Mikky

    They found discontinuities relative to neighbouring stations, not in the Rutherglen data series per se.

    But Ken Stewart has checked Rutherglen relative to neighbours and the ‘raw’ is actually very similar to neighbours, while the homogenised shows a warming bias relative to neighbours. You should be able to find the link across to Ken’s blog in the main blog post above.

    I’ve had a statistician look at the data, and he can’t find a statistically significant discontinuity in the minimum Rutherglen series, he can in the maximum.

    There is no reference at the moment. But I, and others, are working on technical papers. So, all will hopefully be revealed one day following publication.

  64. Daryl McDonald September 23, 2014 at 5:54 am #

    Round and round we go.
    For the homogenised crowd out there.
    Have you ever tasted chilled, fresh, full cream milk, straight from the cow?
    The difference is amazing.

    Cheers, Slowlurnr.

  65. gavin September 23, 2014 at 7:57 am #

    No Ian G, no. My point has always been that ambient measurement in free air won’t change dramaticly, even in the hour. More, ambient is quite stable over a wide area too.

    Also; a daily max or min can only be determined over a standard 24 hr period, eg an 11 am reading looking back from same day at 8pm could not have been the max (or min) in most circumstances.

    So what are those two recordings related to, if not the day before?

  66. Neville September 23, 2014 at 8:56 am #

    BTW there is now an admission that the temp along the west coast US since 1900 is actually NATURAL and does not involve human co2 emissions .
    You see it’s the PDO and more el ninos again, another big surprise NOT. So why not OZ east coast and other areas etc?
    IOW if they concede that the PDO rules in those areas then how much of the warming since 1900 is because of human co2 emissions ANYWHERE on the planet?

    http://wattsupwiththat.com/2014/09/22/surprising-pnas-paper-co2-emissions-not-the-cause-of-u-s-west-coast-warming/

  67. Richard C (NZ) September 23, 2014 at 11:52 am #

    Neville, I see in the Johnstone & Mantua Abstract:

    “An ensemble of climate model simulations run under the same historical radiative forcings fails to reproduce the observed regional circulation trends…………………regional mechanisms of interannual and multidecadal temperature variability can also extend to century time scales”

    This body of contra-CO2 MDV evidence is building slowly, belatedly, but surely and every one of these papers is a “keeper” in my bookmarks. But known by sceptics for yonks of course, see for example from Alan Cheetham:

    ‘The Sixty-Year Climate Cycle’

    http://appinsys.com/GlobalWarming/SixtyYearCycle.htm

    You ask:

    >”if they concede that the PDO rules in those areas then how much of the warming since 1900 is because of human co2 emissions ANYWHERE on the planet?”

    Yes, problematic. MDV was identified globally in another recent paper for which I have 4 bookmarks because of its importance:

    Diego Macias, Adolf Stips, Elisa Garcia-Gorriz. Application of the Singular Spectrum Analysis Technique to Study the Recent Hiatus on the Global Surface Temperature Record. PLoS ONE, 2014; 9 (9): e107222 DOI: 10.1371/journal.pone.0107222

    http://www.sciencedaily.com/releases/2014/09/140911092905.htm

    http://www.reportingclimatescience.com/news-stories/article/spectral-analysis-says-natural-variations-cause-pause.html

    http://www.sciencecodex.com/last_decades_slowdown_in_global_warming_enhanced_by_an_unusual_climate_anomaly-141430

    http://hockeyschtick.blogspot.co.nz/2014/09/new-excuse-for-pause-52-unusual-climate.html

    This is the graph:

    http://images.sciencedaily.com/2014/09/140911092905-large.jpg

    The green MDV curve wont bottom out until around 2032 but that’s the least of the CO2 proponent’s problems. It’s the red secular trend (ST) that is taking over now and it is turning down away from CO2 i.e. that hypothesis is failing.

    Which introduces an alternate hypothesis: that the secular trend in atmospheric temperature is bicentennial/millennial solar input to the ocean lagged for planetary thermal inertia of which the ocean is the largest heat sink.

    BTW, I found a similar turn down in the secular trend of HadSST2/3 (therefore HadCRUT3/4) some time ago (a couple of years or so) by the residual trend of an empirical mode decomposition (EMD) analysis when adding new data. The residual had previously been turning up, Scafetta used a quadratic to describe it – now invalid. The intermediate mode frequencies (IMFs) identify the oscillations e.g. MDV above. EMD has been used in the literature too, obtaining these same results. but hasn’t grabbed attention like the spectral analysis above.

  68. Richard C (NZ) September 23, 2014 at 12:28 pm #

    Jennifer,

    >”They [BOM] found discontinuities relative to neighbouring stations, not in the Rutherglen data series per se.”

    Statistically. I don’t think this will be resolved until thermometers are used to prove the method. See my call above, viz. “Empirical proof of statistical method”.

    >”I’ve had a statistician look at the data, and he can’t find a statistically significant discontinuity in the minimum Rutherglen series, he can in the maximum.”

    Well done, good course of action and it’s the local raw data that is paramount. This Max/Min disparity bugs me (discontinuity in one but not the other and one or the other depending on what statistical test is used). It hasn’t been apparent in the other compilations e.g. NIWA, BEST, because those are applied to the Mean.

    NZCSC deferred to 3 statisticians, didn’t help their court case but did uphold their methodology. Their initial statistical test is similar to above (as is NIWA’s), and only once a discontinuity is identified in the local (“candidate”) site are comparators (“neighbours”, sometimes remote in the case of NIWA and BOM) referred to.

    If in fact BOM found no discontinuity in the initial test there was no need for recourse to neighbours. To do so is a form of circular reasoning or confirmation bias (“regional expectation” in the case of BEST)

    >”There is no reference at the moment. But I, and others, are working on technical papers. So, all will hopefully be revealed one day following publication.”

    Good to know and good on you all.

  69. Neville September 23, 2014 at 1:24 pm #

    Jean S and Steve McIntyre show the timeline of how “hide the decline” was included in the HS graph.

    http://climateaudit.org/2014/09/22/black-tuesday-of-climate-science/ And it proves that Mann agreed with the work of Phil Jones and others when they added the instrumental temp record after 1961. Hopefully Steyn and his lawyers should have a field day with this new information.

  70. Ian George September 23, 2014 at 1:29 pm #

    Gavin
    I say again, same station, same day.
    When Sydney Obs had its top temp of 45.8C at 2:53pm on 18th Jan last year, the AWS was recording every 10 mins.
    At 2:49pm the temp was 44.9C and the next reading at 2:59pm was 44.8C.
    So the temp spiked 0.9C in 4 mins then dropped 1.0C in the next 6 mins.
    For instance, at 12:56pm today Casino was 24.6C. At 1:00pm the temp dropped to 23.7C.
    You need to be following the temp changes closely otherwise the Tmax will change if it goes above the previous high.
    High here
    http://www.bom.gov.au/nsw/observations/nswall.shtml?ref=hdr
    30 min timing here
    http://www.bom.gov.au/products/IDN60801/IDN60801.94573.shtml

    You could do this with any station using this BoM site.

  71. gavin September 23, 2014 at 6:28 pm #

    Quote from BoM – “High Temp” is the Daytime high temperature, measured in degrees Celsius, from 6 am to 9 pm on the current calendar day.

    These values are available for stations with high frequency reporting schedules. The report time for each highest value is included beneath the value.

    Note: Maximum temperatures in the Bureau’s climate records are measured from 9 am to 9 am and may differ from the values shown in the table.

    Apologies; High T and T max are not the same

  72. Ian George September 23, 2014 at 7:05 pm #

    The highest temp in Casino today is showing as 24.7C (1:58pm). The highest temp showing on the HFR site is 24.1C (2:00pm).
    I will check tomorrow on this site:
    http://www.bom.gov.au/climate/dwo/IDCJDW2026.latest.shtml
    or
    http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_display_type=dailyDataFile&p_nccObsCode=122&p_stn_num=058208&p_c=-677755011&p_startYear=2014

    to see if it is.
    However I do agree with you that the temp shouldn’t rise or fall that quickly (because of the reasons you point out above).
    But I ask, after looking at the data, why is this happening?

  73. Neil Crafter September 23, 2014 at 9:08 pm #

    Jennifer
    I’m wondering if historic aerial photos of the Rutherglen property might be a good way of looking for possible station moves?

  74. gavin September 23, 2014 at 10:44 pm #

    Ian George; when I noticed your first post on this subject I pondered on it for a few days then phoned in for advice. I took a while to find a number and I thank you persistant bloggers for that, however the answer to my question about the possibility of a different instrument being involved was definitly the previous day for T (max-my term) that shows in the BoM station summary.

    I don’t believe we can see a point 6 C chainge in several minutes at any station that is not subject to some human impact. The industrial equivalent would be like walking through a doorway to a furnace building or large cold room.

    Avoiding blogspots etc- link 3 (2010 paper) is a faster download than the later 2012 100 page update from our gov source

    http://www.bom.gov.au/climate/current/events/january-heatwave-methods.shtml

    http://theconversation.com/no-the-bureau-of-meteorology-is-not-fiddling-its-weather-data-31009

    http://www.stat.washington.edu/peter/593/Trewin.pdf

  75. Ian George September 24, 2014 at 4:40 am #

    Gavin
    We must be at cross purposes or talking about different things. I am just going on the ‘Latest Weather Observations for New South Wales’ site, tracking the temps and reporting the data. One shows an updated high temp and the other the prevailing temps at 30 min intervals (sometimes it shows 10 min intervals). I have seen rainfall totals being updated every few minutes.
    Surely you can do the same using the same site I am using and see for yourself.
    It shouldn’t be happening but it is and I have no explanation for it except to say ‘It’s the same station, same day’.

  76. Mikky September 24, 2014 at 7:45 am #

    Just noticed that the 1964 Tmin data is missing from the BoM website for Rutherglen.

    Spotted this by plotting the annual Tmins for Rutherglen and 6 nearby stations on the same graph, each set of data offset by a few C. Pretty much all the peaks and troughs matched across stations, except for Rutherglen which was shifted by 1 year.

    This suggests that the BoM dates are correct, with 1964 missing, my Rutherglen dates were out by 1 year in the above.

    Rutherglen Tmin variations (assuming the BoM dates are correct) match almost perfectly its near neighbours for 1971 to 1984, NO WAY that a sensible algorithm could detect an inhomogeneity in 1974. I wonder if the missing 1964 data caused the BoM to make the same mistake as I made originally.

  77. Ken Stewart September 24, 2014 at 7:57 am #

    Gavin

    The 5 minute or 30 minute data from many BOM stations will show quite large excursions in the course of a day, e.g. yesterday Rockhampton Aero
    9.00 19.7
    9.30 19.7
    10.00 20.7
    10.30 21.3
    10.57 20.9
    11.00 19.8
    11.08 18.7
    11.14 18.8
    11.28 19.7
    11.30 19.8
    12.00 22.4
    12.30 23.2
    1.00 22.7
    1.30 20.7
    2.00 20.2
    2.30 21.4
    2.42 19.8
    3.00 20.2
    3.30 20.2
    4.00 20.8
    4.30 20.7

    etc

    All to do with ambient weather conditions- clouds, wind, rain showers. It can vary by some degrees in a few minutes as you will see if you can get your own digital weather station, which may not be precisely accurate but will show the changes. The official maximum for yesterday will be determined at 9.00 am this morning.

  78. Daryl McDonald September 24, 2014 at 8:01 am #

    Can anyone provide a list of all sites where the homogenized data/temps have created a downward trend? Min 100yrs of data to have any credence.

    Cheers, Slowlurnr.

  79. Ken Stewart September 24, 2014 at 9:12 am #

    Daryl
    Here are 38 sites with Tmax cooled by -0.2C per 100 years (of varying length records, some quite short):
    Eucla
    Eddystone Pt
    Rabbit Flat
    Walgett
    Forrest
    Georgetown
    Thargomindah
    Cobar
    Nowra
    Carnarvon
    Robe
    Grove
    Williamtown
    Port Macquarie
    Bathurst
    Mildura
    Weipa
    Marree
    Mt Gambier
    Gabo Island
    Oodnadatta
    Yamba
    Perth
    Low Head
    C Bruny
    Camooweal
    Wagga Wagga
    Esperance
    Tarcoola
    Bundaberg
    Birdsville
    St George
    C Moreton
    Nuriootpa
    Geraldton
    Woomera
    Richmond (Qld)
    Sale
    and 27 sites with tmin cooled by -0.3:
    Tarcoola
    Forrest
    Grove
    Wyalong
    Marree
    Charters Towers
    Tibooburra
    Tennant Ck
    Butlers Gorge
    Normanton
    Birdsville
    Gayndah
    Coffs Harbour
    VRD
    Bundaberg
    Moree
    Wilsons Promontory
    Walgett
    Longreach
    Kalgoorlie
    Miles
    Marble Bar
    Meekatharra
    Low Head
    Dalwalinu
    Eddystone Pt
    Morawa

    Hope that helps.

    Ken

  80. Ian George September 24, 2014 at 9:56 am #

    Gavin
    As I reported above:-
    ‘The highest temp in Casino today is showing as 24.7C (1:58pm). The highest temp showing on the HFR site is 24.1C (2:00pm).’

    Temp for Tuesday just posted for Casino (Tuesday, 23 Sep) at 24.7C. See here.
    http://www.bom.gov.au/climate/dwo/IDCJDW2026.latest.shtml

    Highest temp on 30 min interval still showing as 24.1C at 2:00pm.
    http://www.bom.gov.au/products/IDN60801/IDN60801.94573.shtml
    (Scroll down to 23/09.)

    You can do your own checking but the fact remains there was a drop of 0.6C in two minutes.

  81. Ian George September 24, 2014 at 10:11 am #

    Daryl
    A neat way of checking temperature trends is to use the BoM’s DWO site here.
    http://www.bom.gov.au/climate/dwo/index.shtml

    Click on the state you want, then the town from the alpha index.
    Scroll down to ‘Climate statistics’ and click on the station you want.
    The statistics spreadsheet will show you all the years of data for that station plus monthly and yearly average temps.
    Then click on the ‘Use all years of data’ tab and it will show data for 30 year periods.
    Check the earliest 30yr period first and then the last 30 years.
    That will give you some idea of the trend for that station.

  82. gavin September 25, 2014 at 4:01 am #

    A big black cloud shadow Ian. They may come and go but do we know when?

    Ken; As with Casino, Rockhampton shows a little more instability from about noon but I say we are still only seeing a deviation of about 1 C max at any time during the warmer part of the 24 hr cycle. This leaves me to conclude without being pedantic, that ambient T was about 20C for quite a number of hours in the area.

    On the other hand Ian could conclude our modern super sensitive instruments arn’t giving us a beter picture than before. However, even the 30 min intervals are much better observations if we see them as only spot measurements on the curve, something we don’t have from older station data.

    All this forensic blog raking through station history is just side tracking from progress in the art of review developing in climate science. Your methods versus mine is not the real question in global warming senarios. Making use of what we have from historic weather data depends on filtering via the collective mind set. This hasen’t changed for any measurement process.

  83. Ian George September 25, 2014 at 4:41 am #

    ‘Our modern sensitive instruments’ are maybe giving us a better picture than before. It’s just that the older ones would not have picked a sudden change and, since we are right into homogenisation, that should be taken into account.
    Somewhere I read that because of the new equipment, there could be up to 0.2C diurnal difference compared to the old equipment.

  84. Ian George September 25, 2014 at 4:56 am #

    Gavin
    http://cawcr.gov.au/publications/technicalreports/CTR_049.pdf
    Quote on P7
    ‘There is some indication of a small (less than 0.2°C) increase in diurnal temperature range, most likely because of the faster response time of automatic probes relative to mercury-in-glass thermometers.’

    (BTW, having difficulty opening this site.)

  85. gavin September 25, 2014 at 9:25 am #

    Ian; it seems you have no practical experience with the concept of ambient temp.

    In any meaningfull measurement program it is necessary to get rid of “chatter” and with an inline process this could be just mechanical chatter from some mal aligned component such as a valve. But with old max min temp records, that chatter may be just sticky u tube markers being seen as true event recording, all very error prone at a glance.

    On the other hand mercury thermometers can be quick. Think what a nurse has gained with the digital probe.

    All my atmospheric and AC monitoring was done with a matched Hg pair in a sling and dare I say they too were and still are as quick as for my purpose in dismissing instrument sensitivity as the major hurdle in climate science.

  86. Ian George September 25, 2014 at 10:22 am #

    Be that as it may, Gavin, the Tmax and the nearest 30 min interval show some large discrepancies. The Tmax is used to register the temp for that day.
    In my checking of the AWAP and the ACORN temps, they always seem to match in recent years i.e. no homogenisation.
    If it is as you claim, the whole temp record is flawed and there would be no basis for the ‘warming’ or ‘cooling’ trends to be talked about.

  87. Moderator Ray September 25, 2014 at 12:38 pm #

    Ian, I totally agree that the whole temperature record, as presented to us by the official agencies, is flawed. The problem is, NONE of those agencies worldwide, are willing to admit that fact, ditch the homogenised data they have created, and start again with proper, scientific analyses of the original temperature data. Why do they refuse to do this? My guess is that they are fearful of losing their jobs & reputations.

  88. Ian George September 25, 2014 at 2:52 pm #

    What alerted me to the inconsistencies of the homogenisation process was Sydney Obs. In 2014, it had its highest May max mean on record at 23.2C (it also had the hottest day for all of Sydney).
    That’s OK, but it was warmer than every other station in Sydney, even though it has the lowest long-term average (19.6C) than any Sydney site.
    That should come under the term, ‘outlier’. In the past, ‘outliers’ are checked against other stations and homogenised, especially when they are regarded as records.
    Checking the ACORN data, this has not happened with Sydney Obs’ May mean – the temp data remains the same.
    This may mean that the instruments are above reproach and only the ‘older’ records need to be adjusted.
    Source
    http://www.bom.gov.au/climate/current/month/nsw/archive/201405.sydney.shtml

    I have just come across another point of interest.
    According to the summary above, the previous record for May was in 2007.
    ‘Sydney Observatory Hill had its warmest May on record in 2014, with mean maximum temperatures of 23.2 °C, 3.8 °C above the historical average and 0.5 °C above the previous record, set in 2007.’
    Yet according to the AWAP and ACORN data, Sydney Obs’ max mean for May 2007 was 22.4C. 1958 was 22.7C. The Bureau can’t even quote its own data correctly.
    See here at Sydney Obs’ raw temps.
    http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=36&p_display_type=dataFile&p_startYear=&p_c=&p_stn_num=066062

  89. Ian George September 25, 2014 at 4:09 pm #

    Subsequent to what I have discovered above re Sydney Obs’ May 2014 record beating 2007, now I find this from The Sydney monthly summary for May of that year.
    ‘Sydney has experienced its second warmest May on record. The average maximum (daytime) temperature at Observatory Hill was 22.4°C, or 2.1°C above the historic average1of 20.3°C. The warmest May on record was in 1958 with an average maximum temperature of 22.7°C.’ – that figure shows up in the CDO site so it’s legit.

    The ACORN data hasn’t changed, so what is going here? An increase of 0.3C with no reference to any actual data. Another hot recent May to erase 1958? Is there another set of books (sarc)?
    Source
    http://www.bom.gov.au/climate/current/month/nsw/archive/200705.sydney.shtml

    BTW, sorry Jen. This was supposed to be about Rutherglen and I just got carried away.
    The changes and misinformation are apparent everywhere. I should have waited for your Open Thread.

  90. Tor September 26, 2014 at 6:05 pm #

    Ian George September 22, 2014 at 2:47 pm #

    Kevin
    You, therefore, should be able to explain the following adjustments to Bourke for Jan 1939. All temps over 30.0C have been adjusted down and all those under 30.0C have been adjusted up.
    AWAP ACORN

    …snip individual measurements…

    Mean40.4 40.04
    Did they moved the site everyday?

    /quote

    Visual checking of the number didn’t support maintainence of the same mean, so I stuck the number into my spreadsheet. Here is a corrected mean line, reported to 3 decimal places (yes, I know there aren’t enough significant places in the source data):

    AWAP ACORN
    Mean 40.419 40.035

    Someone made an error.

  91. Ian George September 27, 2014 at 5:27 am #

    Tor,
    Unsure of you mean when you say, ‘Someone made an error’.

  92. Tor September 27, 2014 at 10:40 am #

    Oops… You’re right. Looks like I’m the one who made the error in this case. Apparently, I was too tired last and missed the extra zero in the ACORN mean you posted. My extended number does indeed round to what you posted, now that I read it right. Sorry.

  93. Ian George September 27, 2014 at 1:53 pm #

    That’s ok.

Trackbacks/Pingbacks

  1. Australian climate science | The k2p blog - September 24, 2014

    […] It could be the BOM. […]

Website by 46digital