Australian Climate Commission bungles simple temperature anomaly chart

The Climate Commission has this downloadable pdf report “The Critical Decade – Queensland climate impacts and opportunities” – still available from their website.
Let me know if it goes offline and I can post my copy just downloaded today.
Canberra public servants will have flexed off for the weekend. It would take a miracle to fix this before Monday – but then Tuesday in Melbourne Cup day. Who knows when it will be fixed.
The very first graph on page 4 of 28.

As if the moving average can run OUTSIDE the data bars.
Does nobody proof-read these days. A proper chart could be made at this BoM page – the running average thing was not working for me just now. Thanks to readers who emailed this gem.

15 comments to Australian Climate Commission bungles simple temperature anomaly chart

  • […] Errors in IPCC climate science » Blog Archive » Australian Climate Commission bungles simple tempe…. Like this:LikeBe the first to like this. […]

  • Graeme Inkster

    Warwick,
    that rolling average doesn’t seem to have anything to do with the figures.

    Using a sight estimate of the anomaly figures and a rolling 6 year average from 1940 (obvious start 1935) gives a maximum of 0.79 and a minimum of -0.6. Accuracy doubtful of course, but hardly sufficient to give the (estimated) max. or 1.23 and min. -1.25.

    The 6 year figure seems likely from the start of the graph. Any longer average would reduce the span further.
    That dip back into negative (1983/4) doesn’t show in the rolling average.

    I would suggest their rolling average is a temperature chart for global temperature from HADCRUT or similar, and nothing to do with Queensland (or as you have demonstrated with reality).

  • David Brewer

    I don’t think the rolling average is HadCRUT, the amplitude is too big. I think it probably is related to the bars, and an 11-year rolling average, but just about twice the amplitude. In that case they should have had a second scale for the average, or – if all else failed – just got it right the first time.

    Also interesting that despite the recent decadal trend, the latest year is the coldest since c. 1985, and the last two years both colder than average for the first time since the 1970s.

  • Graeme Inkster

    David Brewer;

    more than twice, as the longer the averaging period the less the variance. And that dip around 1983/4 still wouldn’t show up.

    Any way, what are HADCRUT’s figures these day? The endless “adjustments” from the believers of doom is always to increase the recent figures, and decrease the temperatures before 1950. The effect is to increase the variance.

    As far as I know (relying on the Met Office UK) the last decadal increase was 0.03 ℃, or a simple minded projected “jump” in temperature of 0.3℃ by 2100. We are all DOOMED ( © J. Hansen, P. Ehrlich, D. Suzuki etc.)

  • Bob in Castlemaine

    Some new variant on Mike’s Nature trick perhaps. Or in this case maybe the go forth and multiply illusion?

  • Beachgirl

    What could have happened might be as simple as selecting the wrong column for the running average numbers on a spreadsheet. Easy to do if the spreadsheet carried many columns of various data with the same timespan.
    And nobody noticed which is the unbelievable thing. It pays to get a mate or two to proof your work, someone not working on the paper.

  • David Brewer

    Graeme and all,

    I think it really is a scaling problem. If you get the BoM’s programme to do it, you get the same curve, but with much lower amplitude: the correct line oscillates between about plus and minus 0.5 or 0.6, instead of from minus 1.2 to plus 1.5. I guess the scale is out by slightly more than 2, as well as not being exactly right in relation to the zero line. See here. If this doesn’t work, insert
    reg.bom.gov.au/cgi-bin/climate/change/timeseries.cgi?graph=tmean&area=qld&season=0112&ave_yr=11 in browser.

    Lord knows how they managed to bugger this up, when all they had to do was download the whole thing from an existing BoM webpage…

  • David Brewer

    BTW, the whole report is a doozy. I’ve just looked at the first four graphs and they are all complete stuffups like this one. Maybe you would like to put some of them up Warwick?

    Figure 2, which is on page 2 of the report just below Figure 1 that you show, is another shocker. First, they have put the 10-year average in bars, and the yearly figures as a line – the opposite of what they do in Figure 1, so extremely confusing when the two are sitting next to each other. Then – same problem as in Figure 1 – the 10-year averages run outside the yearly figures, so are obviously wrong, and on a different scale.

    Figure 3 on page 3 is sadly confused too. It consists of two rainfall trend maps of Australia, one for 1900-2011, the other for 1970-2011. The first is mostly green (presumably more rain) with a few spots of brown (less rain). The second one is about half brown and half green, but with much sharper colours. Well, so trends are larger over shorter periods, who would have guessed? The green-brown scale goes from minus 50 to plus 50 but it doesn’t say what the numbers mean. Is it percentages, or what? And is it the amount by which the years covered deviate from some longer-term average, or the difference between the start and end years, or the trend difference per year or per decade? They don’t tell you any of this, so you have no idea what the graphs actually show. All they say in the text is that it’s got drier on the Queensland coast since 1970, but wetter elsewhere.

    It does give a source for the maps, which if you look it up in the references gives you a url, which if try to go to, says You are forbidden from accessing this page. Never mind, if you trawl around the BoM site for a while you can find where they got their maps. There you can learn what they actually show. They aren’t in percentages, they are in millimetres, and it’s the trend in total annual rainfall in millimetres per decade. And talk about cherry-picking. If you take in just one more decade for the recent chart, making it 1960-2011 instead of 1970-2011, practically all the nasty dark brown on the Queensland coast disappears. So what the maps really show is that it rained a lot on the Queensland coast in the 1970s.

    Figure 4 – are you sick of this yet? – presents two maps of Australia showing dengue fever incidence in 2009 and dengue fever incidence in 2100. Apparently now you can get dengue fever as far south as Rockhampton, but in 2100 you’ll be able to get it in Grafton. A few caveats are mentioned – climate won’t be the only factor, technology and population will also play a role etc. If your read the text you find out that the map is based on a climate model that predicts warming, and if that warming occurred the vector mosquito for dengue could expand its range (in theory). But they don’t say that the vector can already be found as far south as southeast Queensland and that dengue is still as rare as hen’s teeth because very few people come back from overseas with it and infect the mosquitoes. They also modestly hide the fact that the chance of contracting dengue fever in Australia is extremely slight, and that if that ever changes it will have little or nothing to do with climate.

    The whole thing would be a disgrace if it were put out by some panic-merchant NGO. But get this, it has been signed off by four highly-paid “Climate Commissioners” after being vetted by a “Scientific Advisory Panel” of five. Your taxes being flushed down the great white telephone!

  • Philip Bradley

    The answer is pretty simple. The Climate Commission is staffed by incompetents, with little knowledge of science.

  • Maybe they’re looking for a Nobel Peace Prize.

    After all, isn’t that awarded to all those who save the planet?

  • Graeme Inkster

    David Brewer Says:
    I think it really is a scaling problem.

    It might be, but the graph above has the rolling average starting in 1916, so does it include figures from 1905-1909 not included on the graph? As the rolling average finishes in 2011, that should be the case.
    In the case of the graph you link to (above) the rolling average starts in 1916, but finishes in 2006. That implies figures from 1910-2011 as shown and the rolling average plotted off centre.

    When I run the figures on a spreadsheet with an 11 year rolling average I get a minimum of -0.56, yet a maximum of only 0.11. A range of 0.7 v theirs of 1.09. I can’t see this as anything other than a graph drawn for maximum impact rather than accuracy.

  • Siliggy

    In order to guess what that chart will look like next year and on I had a look at the near mirror image situation at 1960. Then add this. Click here.

  • David Brewer

    Graeme

    Good point. I have gone back to the correct BoM graph here. It rightly starts the 11-year running mean in the sixth year of data, and ends it in the sixth last year of data. This is because you need five years’ data either side of a year to calculate an 11-year mean centred on that year. For each of the first and last five years of data, you do not have enough years around it to calculate an 11-year mean.

    But the Climate Commissioners’ running mean goes right up to 2011. How did they calculate the 11-year running mean for each of 2007-11? I thought at first they must have padded the data, putting in values for 2012-16 – either the 2011 figure, or the average of recent years etc. You see various padding methods.

    But that’s not it either. The Climate Commissioners’ running mean is exactly the same shape as the correct BoM running mean, just more than double the amplitude, and shifted in time, so it ends in 2011 instead of in 2006. So then I thought, OK, they have just shifted the running mean five years to the right. I’ve seen that done before. It’s not great because the running mean isn’t really running in sync with the data, but it does show for each year what the mean was for that year and the ten before.

    But, wait again. The beginning of the 11-year running mean in the Commissioners’ graph is still in the sixth year of the series. They have not just shifted the mean five years to the right. They have actually STRETCHED the 91-year running mean (1916-2006) over 96 years (1916-2011). The 1916 mean is against the correct year, but then for every year after, the mean is gradually drifting right of the annual data. By 2011, the mean shown is actually the figure for 2006.

    Among other things, this (partly) explains why the running mean is wrong against the zero line. It is supposed to be set to zero on the 1961-90 mean, but seems to average a bit below zero for that period. The reason is that the mean line has drifted three to four years to the right by that time. If you look at it from about 1963 to 1994, it seems closer to the zero on average. Actually, it may be a bit high. The reason for that is that they have also moved the trend line up. In the original, it varies between plus and minus a bit over 0.5. In the Climate Commissioners’ version, the range is about minus 1.2 to plus 1.5, so it has moved up, as well as more than doubling.

    So, the running mean has not been calculated at all. The Commissioners have simply taken the curve from the BoM site, and plonked it on top of the bar chart, 5% wider and about 110% taller than the original, as well as shifted in relation to the zero line. The result is to show the line on the wrong vertical scale, the wrong horizontal scale, and incorrectly centred, both vertically and horizontally. Truly, a prize effort!

  • Graeme Inkster

    From Jo Nova; an interesting comment by John F. Hultquist (slightly deleted)

    “I think I’ve figured out the curve. The black line looks very much like the topographic profile of a subduction zone coastal cross-section. Note the ridge on the left between about 1930 and 1940 – that’s the active spreading center. Then there is the spike down at about 1954 – that’s the trench where younger material disappears under the continental shelf (1960 to 1985). The rest of the right-hand side is the mountainous arc on the edge of the continent. The curve would fit nicely to the Cascadia Subduction Zone of the coast of the USA State of Washington”.

    Just copying a profile and superimposing it might be the limit of their mathematical ability.

  • Alex

    It is a fabrication. A rolling average is (a) lagging; and (b) far smoother than the original data. This line has neither attribute.