Faux Record Hot Days, Including at Mildura

Alan Jones AO
Radio 2GB, Sydney

Dear Alan,

This spring has begun with the Bureau of Meteorology announcing a new ‘hottest September day on record’ almost every other day.

Last Saturday, for example, we heard that it was the hottest September day in Mildura since 1889. Really? Hotter than 1905 or even 1938 – those were hot years in Mildura.

Extreme heat in inland Australia is usually associated with prolonged drought. But this year there is water in the Murray River; further, modern Mildura is generally considered up to half a degree cooler on average than before the widespread development of irrigation – back in the 1930 and 1940s, for example, drought really was the norm, along with rabbit plagues and dust storms.

Before the advent of irrigation the Murray river would run dry during periods of extended drought. These were exceptionally hot years in the Murray Darling Basin, and include 1915 and 1941. Photograph taken by Jennifer Marohasy, upstream of Mildura some years ago – in about 2007, which was a drought year post the building of the Hume Dam.

I wrote to Minister Josh Frydenberg earlier in the week explaining that these new record hot days are unlikely to withstand scrutiny – for Mildura or anywhere else. My letter included a fair amount of technical detail, and a request for data to enable a direct comparison of temperature measurements from the old-style mercury thermometers with the new style electronic probes in automatic weather stations (AWS) – letter attached.

In the case of Mildura, the current AWS electronic probe was installed on 27th June 2012 – the last official measurement from a mercury thermometer at Mildura was actually on 31st October 1996. Assuming a time constant of 18 seconds for the current probe and given the Bureau’s current method of only taking one-second readings (i.e. taking spot readings – rather than averaging over at least one minute in accordance with World Meteorological guidelines), then the recent record hot day for Mildura is probably only comparable to measurements taken back to September 2012 (last 5 years), perhaps back to 1996 (last 21 years) – certainly not back to 1889 as claimed by the Bureau and reported in The Age.

Minister Frydenberg has been kind enough to acknowledge receipt of my letter, but he doesn’t seem to get how important sorting out the mess that is the Bureau of Meteorology actually is. He has suggested that we meet in a few weeks, when he is not so busy as he is at the moment. Of course, Ministers are always busy – what gets done depends on what they choose to make a priority.

I am writing to you now because you have been so effective in the past at getting some things done. For example, after you interviewed me on Wednesday 26 July about the Thredbo weather station and the artificial limits the Bureau had imposed on the measurement of cold temperatures: the very next day – after 10 long years of the imposition of these limits – they were lifted at Thredbo.

Perhaps, you could invite me back onto your program to discuss this issue of faux hot day records?

This is the first time I have actually requested to come on your program. As those close to me know, I usually don’t sleep the night before such a radio or television interview – because I find it all so frightening. I much prefer to analyse data alone, write technical papers with colleagues, and go canoeing with just nature.

Perhaps just writing this letter to you, and posting it at my blog with be enough?

You could perhaps interview Craig Kelly MP instead – because he is across this issue. I see that Steve Price has been standing-in for you recently: perhaps Steve Price could interview Craig Kelly MP? You/Steve Price would really only need me if someone from the Bureau was prepared to actually come-on your show – then I would be keen to come-on and we could have a proper discussion about the myriad of technical issues that need sorting. In the meantime, there really is a need for a few more people to start discussing these issues in general terms – then the Minister might realise, finally, that there is no need for the average Australian to be paying ever more for their electricity because we may not actually have run-away global warming.

For me the overriding issue is the integrity of the historical temperature dataset, specifically the Australian Data Archive for Meteorology (ADAM). Indeed, I am currently driving my work colleagues mad-with incessant complaints about no-longer being able to trust any of the data from the Bureau of Meteorology.

Previously, I’ve complained about the remodelling by the Bureau of the ACORN-SAT dataset, while working with data from ADAM for my rainfall forecasting. But since discovering the limits on the recording of cold temperatures, and then coincident to this that the Bureau are taking spot-readings rather than averaging – it all seems more desperate than ever.

Yours sincerely
Dr Jennifer Marohasy
Noosa, Qld

Craig Kelly MP
Steve Price (via Carla Horton)
John Roskam, IPA

Attachment/letter to Minister Frydenberg: http://jennifermarohasy.com/wp-content/uploads/2017/09/Marohasy-ABC-4-Mildura.pdf

24 Responses to Faux Record Hot Days, Including at Mildura

  1. Warren McLaughlin September 28, 2017 at 4:15 pm #

    Yet again B.O.M. policy has been directed towards reporting temperature readings higher than fact. In the case of the Thredbo weather station, after 10 years the artificial limits on measuring COLD were lifted. I believe the BOM should be asked to explain publicly why these limits were ever imposed! If BOM had to explain publicly the rationale for all of those steps you have challenged it would make interesting reading. I can think of at least 4 techniques BOM have used that (imo) have to be regarded as producing misinformation. I totally support your holding the BOM to account

  2. Fred September 28, 2017 at 8:07 pm #

    Don’t know if you have seen this site quite a few Australian archived climate records in here all downloadable and searchable.

  3. Siliggy September 28, 2017 at 9:37 pm #

    The 18 second time may be the figure for water immersion. So would be a lot longer for air but air speeds this time up a lot with wind speed. The Body surface area of the AWS probe may be far more affected by wind speed shortening the time constant. Back when Australia was very hot people who recorded far hotter temperatures than the official records now mentioned a “sudden hot wind” some of them called it the Australian Sirocco. Captain Charles Sturt, Sir Thomas Mitchell and Charles Darwin. All of them quoted amazing sudden high temperatures with it. The odd spikes in temperature that occur between the regular 1 minute last second of minute samples and the sporadic selective maximum sample indicate posible noise spikes. These may get in because the initial change rate is very fast. with no other samples being continuously averaged in, it is a recipe for upward offset. Here are two charts of thermometer time constants that show short times for glass thermometers. http://aapt.scitation.org/doi/abs/10.1119/1.2341219?journalCode=pte
    Lance Pidgeon

  4. Siliggy September 28, 2017 at 9:57 pm #

    The smaller recent Stevenson screen sise reduces volume of outside air compared to the hot box in the sun. So may warm the still air times. We very much need to see those side by side comparisons you have been chasing Jennifer. Some for Moree may exist.

  5. Dave Vought September 28, 2017 at 11:51 pm #

    Good luck Jennifer.
    Minister Frydenberg knows only to well what’s going on. He hasn’t got the ticker to take the BOM on.
    Both major parties state and federal have destroyed our energy security and jobs on the back of the BOM’s fraudulent claims.
    They have sold or given away our cheap reliable coal fired power stations on the advice of overpaid loonies who believe in expensive wind fairies.
    Our once great Australian industries will all be forced to the countries who haven’t fallen for the global warming scam and are building new coal fired plants in the hundreds to accommodate them.
    Now we have our prime minister begging AGL to keep open our sold off (cheap) power stations. How dire must it get before they act.

  6. hunter September 29, 2017 at 1:43 am #

    The implications for the worldwide edifice of temperature data is profound. There is no way the Australian experience us isolated.

  7. Siliggy September 29, 2017 at 5:32 am #

    1896 was a cool year according to Sir Charles Todd, yet for 23 days of it Australia was very hot over a far larger area than any recent heatwave. The rapid series of very extreme weather events for the rest of the year, like cyclone Sigma, kept the yearly average down. During that heatwave people cried out and prayed for a gust of wind but there was none. This would have the time constant of the thermometers at a very slow rate. Did not stop the thermometers going up though because the heat persisted day and night. Good thing they were not reliant on wind turbines to power aircon and meteorological equipment. Imagine the death toll. Sir Charles Todd watched it all “In the early 1880s Todd and his staff at the West Terrace Observatory in Adelaide were drawing inter-continental weather charts that had greater geographical reach than any other jurisdiction in the world.” Wiki. Bet Threbo got down past -10.4 that year too. http://trove.nla.gov.au/newspaper/article/14058062

  8. Siliggy September 29, 2017 at 6:26 am #

    “after 10 long years of the imposition of these limits – they were lifted at Thredbo.” Limit = -10.4. We are told this had never been a problem before. By a strange coincidence just a few days before the 16th (when Thredbo stopped at -10.4 six times and it was deleted) on the 12th the temperature got down to….wait for it ….-10.4. Then less than a week after the limit was lifted it got to -10.9 The year before in August it got down to….Can you guess? -10.4. In 2015 the coldest was, you know already, -10.4. In 2014 just by coincidence of course the coldest was -10.4. Missed by 0.1 in 2013. Is that because -10.4 was skipped on the way down? In 2010, what a surprise, -10.4. Then there was 2009 with -10.4. Do they believe their own spin, wash and rinse?

  9. Jennifer Marohasy September 29, 2017 at 9:14 pm #


    The time constant is the time required to achieve 63.2% of an instantaneous temperature change; the change could occur in water or air.

    I am concerned, however, exclusively with the measurement of surface air temperatures.

    Relevant WMO (World Meteorological Organisation) documents mention that this will be dependent on the air flow over the sensor, e.g. http://www.seedmech.com/documents_folder/wmo_no_8.pdf

    In this, and other documents, the WMO guidelines generally specify 20 seconds as the expected sensor time constant.

    Various Burea reports (e.g. A Preliminary Investigation of Temperature Screen Design and Their Impacts on Temperaure Measurement, Instrument Test Report No. 649), however, suggest the Bureau was experimenting with a time constant of 18 seconds.

  10. Mark M September 30, 2017 at 5:58 am #

    The rabbit hole of surface air temperatures.
    Leave a trail of breadcrumbs when venturing down there:

    There’s an over-emphasis on the surface air temperature. – Prof Matt England

    NASA: There is far too much focus on surface temperatures.

    Gavin Schmidt: … the surface temperature data analyzed and reported by NASA, NOAA and others is viewed as the gold standard.

  11. Ian George September 30, 2017 at 1:52 pm #

    Another case of one second spiking.
    In Casino today (30 Sep) a top reading was posted as 28.9C at 10:00am.
    However the temp posted at exactly 10:00am was only 28.0C.
    So somewhere during that minute, the temp dropped 0.9C in less than a minute.


    Bet the old therms couldn’t have picked that.

  12. Siliggy September 30, 2017 at 3:52 pm #

    Hi Jennifer
    Wouldn’t it be nice if the Bureau of Meteorology and the WMO specified what they have done properly.
    The thermometer can only report its own temperature. So the time constant changes with both medium and wind speed.
    In that WMO document. The next sentence after they specify 20 seconds is “The time constant depends on the air flow over the sensor.”. It is meaningless without an associated wind speed. A Japan Meteorological Agency publication claims that the WMO specify 30 to 60 seconds.
    ” WMO advises that the best representative value of air temperature is the average taken over a one-minute period, meaning that a number of readings should be made if a thermometer with a very small time constant is used. Rapid fluctuations are smoothed with a thermometer that has a large time constant. However, if the time constant is too large, lags in response to temperature variation will cause errors. The time constant of a thermometer varies inversely with the square root of wind speed, and WMO recommends that the time constant be 30 to 60 seconds for a wind speed of 5 ms-1. ”

  13. Siliggy September 30, 2017 at 8:56 pm #

    If the formula holds true over a wide range of wind speed, a way to see the effect of that Japanese wind speed to time constant formula (1/sqrt(x)) is to type the following words without quotes into Google “graph of one over square root of x”. Then use the x value on the zoomable movable graph chart that comes up to see the Y value multiplier. For example when X = 5M/S Y = 0.45. So the time constant at 5M/S is 0.45 x the 1M/S time. To convert a known time at one speed to another speed an example is for 5M/s to 0.5M/S. @5M/S Y=0.45 @0.5M/S Y=1.41 So the time constant of at 5M/S of lets say 45 seconds becomes 141 seconds at a wind speed of 0.5 Meters per second. If the formula keeps holding true at 0.05 meters per second which requires a real zoom in on that chart the time is 447 seconds or seven minutes and twenty seven seconds.

  14. Jennifer September 30, 2017 at 9:04 pm #


    In response to the first of your above comments (posted at 3.52)…

    The WMO guidelines are quite specific:

    1. Platinum resistant thermometers have a time constant that is smaller than that of liquid-in-glass thermometers. (pg 77, Ch2,

    2. A generally accepted rule of thumb is to sample at least once during the time constant of the sensor. (pg 539, Ch1.

    3. Sensor time constant = 20 seconds; output averaging time = 1 minute… with the caveat that “achievable uncertainty and effective time-constant my be affected by the design of the thermometer solar radiation screen. Time constant depends on the airflow over the sensor.” (Annex 1.E Operational Measurement Uncertainty Requirments and Instrument Performance).

    The Bureau is apparently in breach of WMO guidelines as I explained in a blog post (8th September) here: http://jennifermarohasy.com/2017/09/vindicated-bureau-not-following-wmo-guidelines/

    The Bureau responded to this blog post with a Fast Facts (11th September), which I responded to immediately: http://jennifermarohasy.com/2017/09/bureau-management-rewrites-rules/

    I wrote that blog post as soon as I was given the Fast Facts, as I was under pressure for an immediate response.

    But we clearly need an explanation that is more definitive, and also parallel data: temperature measurements that have been recorded from both an electronic probe (PRT) and a mercury thermometer at the same location for a period of time.

    I should be keen to hear from anyone who has been involved in any experimental work in this area.

  15. Siliggy September 30, 2017 at 9:27 pm #

    What 63.2 % means is that if the temperature fell by 10 degrees (EG: from 10 to zero due to hail) then it would take one time constant for the thermometer to follow it down to 3.68 Degrees C. It would take another 4 time constants to get down to 0.07 degrees C.
    1 = 63.2%, 2 = 86.5%, 3 = 95%, 4 = 98.2%, 5 =99.3%. A thermometer that takes too long to cool (time constant too long), in this situation could create a false warming of the minimums.

  16. Siliggy October 1, 2017 at 6:31 am #

    The worst way the BoM seem to defy the WMO recommendations is having irregular selective sampling of the two most extreme craziest outliers (that pass a few tests) as the daily minimum and maximum. Even if Jane Warne was not given the current PRT probe for her ITR649 paper, she found the Stevenson screens to only contribute 3.5 to 4.1 seconds. So the BoM publication “Fast Facts” in which they mix up thermistors and RTDs claims a longer time constant than WMO recommendations. They are doing their own wildly different thing either way. I suggest people get a copy of this incase it mysteriously goes. http://www.bom.gov.au/inside/AWS_Review_Fast_Facts.pdf

  17. Siliggy October 1, 2017 at 6:34 am #

    It seems the BoM state that the probes are faster ” 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. This finding is consistent with international experience (Trewin, 2010).”

    They also state they are slower.
    “The response time of the sensor used in the Bureau AWSs is as long or longer than the changes in the temperature of the air it is measuring.” “Fast facts.” Must have incoporated some TARDIS technology.

  18. Peter Meadows October 2, 2017 at 9:38 am #

    Well, done, Jennifer. Your letters to both Alan and the Minister are excellent. I am not sure if the Minister will understand the full import of what the BoM is actually doing, but let’s hope he passes your request on with a firm instruction for the BoM to provide the data you have reasonably requested. I trust that the parallel readings from mercury thermometers are still available and have not been destroyed, as they seem want to do with data they do not like.

  19. Alfred (Melbourne) October 4, 2017 at 4:14 pm #

    Someone needs to carry out a study where the old way of measuring temperature is done right next to the newer way.

    There seems to be plenty of money to spend on useless things like wind-turbines and batteries while the underlying science is totally neglected.

    Frankly, it should not cost that much to carry out such an experiment. A CCTV camera can track the old mercury thermometer and can be logged remotely – like for the moder equipement. No need for anyone to get up at all hours to write it down.

  20. hunter October 4, 2017 at 9:25 pm #

    Alfred, I proposed just that earluer this year.
    A side by side comparison test of the historical techniques vs. the current technology would most revealing.
    The cost would probably be quite reasonable.
    There is clearly something not right with the temperature record. A meaningful comparison test woukd be one way to identify the issues.
    This might even be useful to Jennifer ad a way to nail down just how bad the new data system is, as well as to clearly demonstrate what the climate obsessed have done to the science.

  21. Kneel October 5, 2017 at 11:05 am #

    “The Bureau is apparently in breach of WMO guidelines …”

    As I posted at JoNova’s, the Bom will hide behind “average” – unless it specifically says “running average of multiple measurements” they will claim they have in fact been doing averages – averages of the min and max over the last minute. That this distorts the “answer” is of no matter to them – they can say they were not contrained by the method of averaging, so they picked what they thought, in their expert opinion, was best.

    If so, they need to defend their choice. Without constraints from WMO, you need to show how much variation can be introduced by changing the “free” part of the method.
    Otherwise, they will dismiss you with a “we are the weather/climate experts, not you”.
    Sad, but true.

  22. Siliggy October 5, 2017 at 12:59 pm #

    Hey wait up. Reading that Fast Facts a bit more carefully it only says “…that each one second temperature value is not an instantaneous measurement of the air temperature but an average of the previous 40 to 80 seconds.” As can be seen from my comment on September 30, 2017 at 9:27 pm, this could be five time constants making 40 to 80 equate to a single time constant of 8 to 16 seconds. So no TARDIS technology required for both documents to be saying that the response time is faster for the PRTD. Fast facts also seems to say this another way. “The response time of the sensor used in the Bureau AWSs is as long or longer than the
    changes in the temperature of the air it is measuring.”. So how long is that? Answer from the same document “The air temperature fluctuates frequently on the scale of seconds.”.

  23. Siliggy October 5, 2017 at 3:02 pm #

    Ways not to do it are clearly specified by the WMO. For example
    “ (a) That atmospheric pressure, air temperature, air humidity, sea-surface temperature, visibility, among others, be reported as 1 to 10 min averages, which are obtained after linearization of the sensor output;”

    I read that as not 18, 20 or 40 to 59 seconds, not before the linearization and not before the sensor output.
    I also read it as not being the thermometer time constant. Which is specified with a different time in a different section so can’t be the same thing.

Website by 46digital