Rutherglen is one of the seemingly best stations in Australia, apart from a break from 1955-1965. Bill Johnston looks closely at the raw data, finding that there is probably no trend — flat temperatures — rather than either cooling or warming. And that it’s difficult to fill in data from surrounding stations. He speculates that something fishy goes on in 1924. He also finds that rainfall probably drives a fifth of the temperature swings. He discusses his disappointment at the intellectual level of debate on The Conversation.
Because he knows the area, he also talks about the effect of wet years and dry years, and how that affects winter and summer temperatures. He has a dry wit, and lovely casual style.
I think that if we have to rely on statistical analysis to “know” whether data was shifted or moved when there is no documentation suggesting it was, all certainty is shot, and any definitive statement about temperature trends in Australia is a joke. — Jo
The Rutherglen stoush
Guest post by Bill Johnston
Before the big, fat, green, wrecking-ball totally trashes our economy and reduces our children and us to green-serfs, we need different national conversation.
- Are we really headed for the cooker; or is it just homogenised data that point that way?
- Has homogenisation produced a more believable ‘product’ or just a more-marketable brand?
- Is it really possible that temperature records are broken, seemingly every second day; when daily data have such obvious historical failings; and when modern data are not observed using thermometers and are possibly homogenised on-the-fly?
- Do we really need to irrigate the Southern Ocean with precious Murray-Darling Basin water; build expensively subsidised windmills and other green-trinkets; and even if you think so, will it really change our climate?
As shown in this essay, the biggest problem of all is that data were first collected in Australia to describe and understand our weather; not track the climate over time. People were no-doubt as vigilant as they could be. However, the odd missing-day was of little concern. Having measured weather for years, not far from Rutherglen, I understand the difficulties and how problems happen.
Observers also did not know that a generation or-so later; bunches of professors would pore over their data trying to polish-out some trend that could mean anything; or, like for these data, nothing at all. A close look at most of our long-term series unearths shiploads of problems. Starting with an arms-length independent audit of the Bureau’s data, let’s strike-up a new national conversation about climate science!
After contributing an analysis of Rutherglen Research’s raw annual minimum temperatures commentary erupted at The Conversation. I naïvely thought The Conversation was about having a conversation; in this case about data.
But it wasn’t. It’s more like a $6M opinion-mosh-pit run by universities that we constantly hear are too broke to teach our kids. It involved some who knew nothing about data; some who knew lots and wanted to share; and some defending their positions, come hell or high-water.
The data’s background.
To understand data, it helps to think about what causes it to behave the way it does.
The Rutherglen region straddles the temperate zones along the Great Dividing Range to the south of Wagga Wagga NSW; and the Mediterranean climates of Central Victoria and South Australia. It is one of Australia’s most productive mixed-grain, grazing, viticulture and orchard-growing regions. The institute located there has been a key facilitator of agricultural innovation for nearly 100-years.
Rutherglen is also down-river from the headwaters of the Murray River; its region is an integral part of the Murray-Darling Basin (MDB).
And don’t you know, the new Murray-Darling Basin Plan will eventually see huge amounts of precious fresh water go right past Rutherglen, on its way to irrigate the Southern Ocean. Climate-science said it should. That’s the new Australian green-way!
Rutherglen’s data are interesting because its climate is usually summer-dry, and therefore, summer-hot, which is why irrigation’s important. Winters are moist, cold and often frosty.
The climate changes from its summer to winter pattern when the high-pressure ridge abruptly migrates north in autumn; it migrates back again in spring. Anzac Day is a handy autumn benchmark for sowing crops. Frosts after the long-weekend in October can be devastating.
The region is sheltered from the east by Australia’s highest peaks. Coastal on-shore easterly-weather leaves it’s moisture on the eastern escarpment of the Great Divide, casting a summer rain-shadow across and beyond the Murray Valley.
In winter, polar lows of the roaring 40’s intrude, bringing rain to the western and south-western slopes; and a rain-shadow to the coast.
There are exceptional years of course, including occasional cyclonic intrusions from the Timor or Coral Seas; years also when the bi-annual climate progressions are early or late, and some when they don’t seem to happen at all. That was the case during the recent long-drought.
Data for Rutherglen Research (station 082010) has been collected, apparently carefully, since 1913.
Because it seems largely unbroken, its one of Australia’s important datasets – one of 116 designated ACORN-SAT sites.
The Rutherglen site is in the open, well away from urban heat-plumes. More importantly, according to the ACORN catalogue, it has always been in the same place. Its raw data should therefore be a useful and untainted record for calculating trends.
Background to the stoush
ACORN-SAT has been heavily marketed as a premier product by our Bureau of Meteorology. It is ‘world-class’, ‘best-practice’; it’s been peer-reviewed; turned upside-down, compared and polished; it glistens with assured, homogenised purity.
ACORN is also important because it forms the backbone of calculating Australia’s warming. Rutherglen’s data contributes to that, and as everyone knows, we’re headed for the cooker.
In Oz right now, global warming’s already a billion-dollar industry. For instance, millions of dollars of work have ‘greened’ their way off-shore because of soaring energy prices. Stressed communities, like Geelong (Vic.) and Elizabeth (S.A.), are on the block. As the Basin-plan bites over coming-years white-anting the food-bowl of its productive capacity, businesses will fail and MDB communities will suffer as well.
Our government’s broke and seemingly riddled by corruption. Debits for wind-farms, the MDB-plan, pipelines and desal-plants; and even our kid’s education, are piling-up to be paid-for generations into the future. We’re probably on some happy-race to the bottom.
Our kids may drown in global-warming-green-wash. So, it’s extremely important their parents and grandparents check what the data behind it all says.
The data story
Previously I checked Rutherglen’s raw annual average minimum temperatures and found irregularities, which is what caused the stoush. The stoush in-turn resulted in a public-berating, prompting me to review the data. That is what I want to talk about.
Rutherglen’s annual raw minimum temperatures are uncorrelated with time. Due to random excursions, they show a trend of 0.03oC/decade, which is no different to zero-trend. They are described fairly, as random numbers dancing along a time-line.
It was fair to believe the initial segment of the data (to 1924) was stitched from somewhere else. They were stepped-high by 0.77oC, suggesting a discontinuity; but no historical account was available.
A closer look, found the ‘hump’ in the middle of the record (1958-1964) contained two values (1963 and 1964) that I’d filled using near-by Wodonga’s data. Those values, which were high, made the detected step of +1.1oC spurious. Removing them removed the step but made no difference to the zero-trend.
Ignoring the missing years and using another two homogeneity tests (four in total), found only the 1924 discontinuity as influential. There was no real trend anyway, and with the 1924 inhomogeneity accounted for, trend was even smaller.
The stoush was about data; and I repeatedly asked the conversing-inquisitors to grab the data and do an analysis for themselves. They could have found out what I’ve now exposed; and we could usefully have conversed about that.
There are two other temperature datasets for Rutherglen’s annual minimum temperature; the HQ data, which I downloaded before the Bureau locked it away; and its replacement, the already-mentioned ACORN data.
Both were claimed to correct for inhomogeneties in the raw series.
If data are faulty, they should either not be used; or they should be transparently adjusted. However, the Bureau’s process of adjusting is a bit murky. There is also no indication from the raw data that they shouldn’t stand-alone.
Summarised as annual data, the ACORN series reveal problems in the numbers of observations in the 1920’s; the 1940’s; a break from 1955 to 1965; and the 1970’s. The 1924 shift may not have been a stitch at all; it could just reflect missing observations.
For the first half of ACORN, trend was zero. For the second half, it was steep (0.12oC/decade); possibly also because of missing data. A step-change in the ACORN series in 1965 is clear.
It’s revealing to graph the three series together. Presuming all of them are missing the same daily data as ACORN; their spurious overall trends are remarkably different.
The raw data trends at -0.03oC/decade; HQ, 0.09oC/decade; and ACORN twice that at 0.19oC/decade. Looks sus; sounds sus; probably is sus.
The main question is whether these data are fit-for-purpose.
Despite homogenisation; up to 1924, and after the break between 1955 and 1965, missing values greatly compromised all three series. Using Wodonga’s data, I illustrated that they can’t just be “filled-in” from somewhere else without creating considerable uncertainty.
For ACORN there was a remarkable up-shift from 1965, which is unexplainable. It’s also a curiosity that after 1974, which was a well-known and well-studied climate-shift year, datasets were in lock-step. It begs the question whether they are engineered on-the-fly to do that.
Finally, if we look at ACORN data from 1978 to 2013 (34 years), which had few missing values, we again find random numbers dancing along a time-line; having an accidental trend of 0.1oC/decade.
So which particular cooker do conversationalists think we’re headed for?
Data needs to be viewed in context.
There is one further aspect of Rutherglen’s data that’s worth a thought, which is that wet years are warmer than dry ones. It’s a conundrum, because we’d generally expect wet years to experience higher landscape-scale evaporation, which is cooling.
With rain falling mainly between April and October, cloudy, wet years, with their temperature inversions; fogs and misty rain, blanket the lowlands around Rutherglen; making it warmer in winter than it would be if skies were clear and frosts more frequent.
It’s the reverse of that case in the summer-rainfall north.
Rainfall explains around 17% of temperature variation. However, because it is random and non-trending, and the temperature data faulty, there is no easy way of obtaining a rainfall-adjusted series.
So we’re stuck with what we’ve got. Anyway, no matter way the data are adjusted, or which dataset is investigated; trends will likely be the same: a big fat zero!
In this case, the data don’t seem up to the job.
Dr. Bill Johnston, a former senior natural resources research scientist from Wagga Wagga NSW is familiar with Rutherglen. He has an enduring interest in the climate faux-debate. He passionately believes the scales need to rebalanced so our nation’s children and its rural industries can look forward to a bright future. “We can’t afford to fail at this”, he said, “No big F; our kids depend on it”. – Jo