THE Bureau has a network of 695 automatic weather stations (AWS) across Australia. In a report released late yesterday it acknowledged issues with the performance of just two of these: Goulburn Airport (Goulburn) and Thredbo Top Station (Thredbo). These are the same two weather stations that I reported at my blog were not recording temperatures measured below minus 10 degrees on the 5th and 18th July, respectively.
While the Bureau strenuously denied it was setting limits, the Minister Josh Frydenberg nevertheless insisted on a review of the entire AWS network.
The Minister phoned me late yesterday to let me know that the report had just been published, and that the Bureau’s investigations confirmed that Goulburn and Thredbo were the only sites where temperature records had been affected by the inability of some Bureau AWS to read low temperatures.
What are the chances? Of the nearly 700 weather stations, I stumbled across the only two with problems.
Goulburn was discovered because my friend Lance Pidgeon lives nearby and was up early on the morning of 2 July concerned his pipes were going to freeze and burst – while watching the live AWS temperature readings tick-over on that weather station, then letting me know when the record for July of minus 10.4 was reached: only to see it rounded up to minus 10.0.
Thredbo was discovered because, after making a fuss about Goulburn, I wanted to check that the Bureau had actually lifted the limits on readings below minus 10. So, two weeks later I decided to get up early and watch the one-second reading at one of the stations in the snow fields on the Sunday morning of 16th July thinking it might be a cold morning. Why did I choose Thredbo – of all the weather stations in the Australian Alps? Simply because my school friend Diane Ainsworth died in the landslide there twenty years ago.
Never mind – I’m vindicated!
The Bureau has now acknowledged that it had inadvertently set limits on how cold temperatures could be recorded at Goulburn and Thredbo.
To be clear the equipment has a general operating range to minus 60 degrees Celsius, but smart card readers – with a nominal range to only minus 10 degrees Celsius and that stop reading all together at minus 10.4 – were inserted placing limits on the actual recordings, not the measurements.
According to the report published late yesterday, the cards were inserted into the Goulburn weather station in September 2002, and into the Thredbo weather station in May 2007. So, for a period of nearly 15 years there has been a limit on how cold temperatures can be recorded at Goulburn, and for nearly 10 years at Thredbo.
This Goulburn weather station was first opened in 1990, and had previously recorded temperatures below minus 10 degrees Celsius in 1994,1999 and 2000 – with a record cold minus 10.9 recorded on 17 August 1994.
The Thredbo weather station opened in 1966, and recorded an average of 2.5 days per year below minus 10 degrees until 1996 when an automatic weather station was installed – replacing the previous liquid-in-glass manually-read thermometers.
Since the AWS was first installed, back in April 1997 there has been a reduction in the average number of days per year when temperatures have fallen below minus 10 degrees Celsius, as shown in the chart.
Further, since May 2007 when the MSI2 sensor interface card was replaced with the MSI1 card (see page 50 of the new report from the Bureau) there has been no potential to record below minus 10.4. Yet not far from this location, at Charlotte Pass, an all-time record low temperature of minus 23 degree Celsius was recorded on 29 June 1994; this was with an old style liquid-in-glass thermometer – not with an AWS.
How can this review possibly conclude that there are no problems with the other 693 automatic weather stations – and there has been no impact on official temperature records from the limits it now acknowledges were placed on recordings from Thredbo and Goulburn?
Surely, there is now evidence enough for a proper external review to be initiated, this should be a Parliamentary Enquiry, through the House Energy and Environment Committee.
The Bureau’s report can be downloaded here: www.bom.gov.au/inside/Review_of_Bureau_of_Meteorology_Automatic_Weather_Stations.pdf
Jennifer Marohasy says
More in today The Australian from Sam Buckingham-Jones: http://www.theaustralian.com.au/news/nation/weather-bureaus-temperature-errors-caused-by-old-hardware/news-story/7a0306d34f31703146de7a37324d1514
“Notwithstanding the soundness of the bureau’s data quality control, there were clearly failures in some of the bureau’s internal processes dating back to the mid-1990s that allowed equipment that was not fit for purpose to be installed at a small number of locations,” the review, by three independent experts, found.
The bureau purchased the MSI1 data acquisition cards with their Almos-brand equipment in 1993. The following year, however, the bureau and Almos discovered the card would not work below -10.4C. The bureau immediately purchased the updated “MSI2” card…
The rest is behind a paywall… consider buying a subscription.
One is tempted to think something has been swept under the carpet here, to put it mildly.
Mark M says
Gratz, Ms M, and thanks for your dedication and hard work.
It’s incredibly disappointing to see that depts like BoM are more interested in patch protection than doing their job.
Instead of being transparent and accountable they make up excuses for themselves.
The probability that those 2 are the ONLY (!) stations that have that issue and you have just amazingly and coincidentally stumbled across the only 2?????
There seems to be a fundamental problem with the way supposedly independant ‘authorities’ are operated.
They are granted a type of ‘legislative monopoly’.
Sadly, way too often, the management ends up managing the system for the benefit of the management.
Excellent work Jennifer.
So for a start, 20 years of minimum temperature data at Thredbo are clipped, and not fit for purpose.
Secondly, 25 years of minimum temperature data at Goulburn are clipped, and not fit four purpose.
And minister Frydenburg has been asked to believe only these 2 stations (the two that were examined by the JM group) are the only 2 out of 695 AWS stations where data has been clipped?
Come on Josh, you are better than that.
A full independent investigation of the AWS measurement and recording process is urgently needed now. Throw in the 1 sec reportingproblem and you are on the edge of a major scandal.
The chances of the internal report by BOM being thorough or accurate, much less truthful, are not.
Congratulations on being proven right.
Your steadfastness us highly admirable.
“Vindicated” We have been accused of all sorts of things. On Facebook I was accused of faking the screenshot that showed -10 had replaced -10.4. even though you (Jennifer) and several other people we know had also seen it and the BoM tweets about -10.4 by then. Also hundreds of others visit the busy BoM site. That accusation now paints the accuser as an idiot. On the trendy fashion “Academic rigour, journalistic flair” site “The conversation”, the biased moderators have deleted gentle questions from people with doubts about the so called “attacked again” line while things like this got left. “Got any of them fancy ‘screen shots’ revealing the big Chinese / UN / Rothschild conspiracy? Or just this one thrilling moment when there was a glitch at BOM?”
The simple reality is that we witnessed real problems and did what it took to get attention to them. There are many more problems like the one second spike peaks being measured instead of an averaging time like inglass thermometers that still need to be public knowledge.
“Since the AWS was first installed, back in April 1997 there has been a reduction in the average number of days when temperatures have fallen below minus 10 degrees Celsius, as shown in the chart.” Again the data is at odds with the claims of “more extreme weather” So either the data is not correct or this cold downturn is part of a return to colder climate. My bet is bits of both.
Jennifer Marohasy says
Great blog post by Jo Nova: http://joannenova.com.au/2017/09/bom-review-finds-skeptics-were-right-but-say-trust-us-it-doesnt-matter/
Steven Fraser says
If the station raw data are still available, a graphed time series should show the clipping with a mark1 eyeball. Values of exactly -10.0 would not occur often naturally over any regions, but a cluster of readings within an area all at the clipping level would be an indicator of the effect occurring.
Steven Fraser says
And, I have to ask… were not formal requirements published for the operational range of thr devices? It seems very strange that someone would think that a device that could not measure down to previous recorded values would be acceptable, nay, preferred in network use.