The Nature of Inclusive Climate Science

Not so long ago, the United Nation’s Intergovernmental Panel on Climate Change, IPCC, released a report about how it is now hotter, and that there is a 95% chance this has been caused by human activity.

This key finding was from just part 1 of a 3-part report. Part 2, is probably more relevant to those on the land as it will discuss “impacts” including how changes in temperature are likely to affect rainfall. Part 2 is not due for release until March 2014. Red gums flooding

But it is possible to get a sneak peek by reading some of the peer-reviewed papers that will contribute to the second report. According to one of these by a team from the Centre for Australian Weather and Climate Research, CAWCR, which is a partnership between the Bureau of Meteorology and CSIRO, there may be an overall 1.8% decline in rainfall in the Murray Darling Basin [1].

Arriving at this -1.8% value involved considering the output from 27 different General Circulation Models, GCMs, run by 27 different teams of scientists from around the world. The final IPCC report due in March will include output from about 50 GMCs run by about 50 different teams, all providing a rainfall simulation for the Murray Darling.

If we consider just the output from the first 27, well it is significant that we don’t know how accurate any single one of the 27 GCMs was at forecasting actual observed rainfall in the Murray Darling. This information is not provided in the paper by the scientists from CAWCR and will not be provided in the final IPCC report.

But we do know that individually the GCMs gave widely divergent results. That is, some of the models predicted a large increase in rainfall in the Murray Darling in the future, while others, under the same temperature scenario, predicted a large drop in annual rainfall.

If I was running the program at the IPCC and at the CAWCR I would insist that the climate scientists first grade the models from best to worst in terms of their capacity to actually forecast rainfall in the Murray Darling. After that, I would insist that only the best model, or combination of models, be used to provide a final figure.

But this is not how modern climate scientists think. They like to be inclusive and so the -1.8% value is an average from the 27 different GCMs, with the final value to be published in March likely to be an average from about 50 models. That is, the final value will be a sort of consensus including output from models that forecast a huge reduction in rainfall and also models that forecast a significant increase with no ranking according to actual ability to forecast rainfall.

If the IPCC had to use such forecasts for anything practical in the real world like growing a crop of food, it would soon be ruin. But it doesn’t, it only has to make pronouncements.


[1] ‘Climate projections for Australia: a first glance at CMIP5’ by Damien Irving, Penny Whetton and Aurel Moise, published in Australian Meteorological and Oceanographic Journal, Volume 62, Pages 211-225.

A version of this article was first published in The Land newspaper on October 3.

The photograph is of a red gum forest not far from Koondrook flooding not so many years ago.


45 Responses to The Nature of Inclusive Climate Science

  1. Luke October 7, 2013 at 10:25 pm #

    Damn – I find myself in agreement with Jen. ARGH !

  2. cohenite October 8, 2013 at 8:57 am #

    Stockwell looked at the faults of the DECR report in 2010:

    Nothing has changed.

  3. bazza October 8, 2013 at 9:40 am #

    There are two ways to cope with increasing uncertainty. You can assume there is none or you can adapt. Note cherry picking a winner is not one of them. Multi-model ensembles always win! There was once a Dilbert expose of the classic scam. Dogbert ( who is to Dilbert a bit as Jen is to some of her readership) explains “I’m creating software that will help small investors pick stocks. It combines past trends that are not indicative of the future with user’s hubris and ignorance. Now all I need are testimonials from people whose results are not typical”. Dilbert observes “So it works”.

  4. spangled drongo October 8, 2013 at 11:37 am #

    The only way the IPCC can prosper is similar to the MSM: by alarming the population.

    They will always be confident there is a disaster just over the horizon.

    Science also profits enormously from this same philosophy. And the consensus is there simply to reinforce the BS.

    For example, today, a story in which the MSM had been reassured was now all good news and nothing to worry about, still proceeded to a full report on how they had thought, 5 minutes earlier, it was nothing short of catastrophic.

    You could see the tears of regret in their eyes.

    It is hard to believe that the mug punters who are always saddled with the bill for all this potential catastrophe would ever consider the modelled 95% let alone a 1.8%.

    Jen, from your lovely photo you can see how this “catastrophe” has unfolded:

  5. jennifer October 8, 2013 at 1:21 pm #

    Spangled, This thread is about climate models and rainfall projections, please post other stuff at the Open Thread. I’ve deleted your second comment and will delete other off topic discussion and comment.

    Thanks Cohenite for the link. And Bazza for the comment.

  6. Robert October 8, 2013 at 2:19 pm #

    A man who makes an excellent Dinky Toy model of a red Ferrari does not know much about Ferraris. He knows a lot about making Dinky Toys.

    Similarly, a man who makes a model of global climate…

  7. Luke October 8, 2013 at 3:05 pm #

    The thing you like about this blog is pig ignorant philistine comments like Roberts. What an incredible contribution of intelligence.

  8. Alan October 8, 2013 at 3:21 pm #

    For those interested, two associated papers from the recent Aust. Jl Earth Sciences

    K. Mills , P. Gell , P. P. Hesse , R. Jones , P. Kershaw , R. Drysdale & J. McDonald (2013) Paleoclimate studies and natural-resource management in the Murray-Darling Basin I: past, present and future climates, Australian Journal of Earth Sciences:60:5, 547-560,

    M. Brookhouse , R. Drysdale , J. McDonald , S. Haberle , M. Reid , M. Thoms & J. Tibby (2013) Paleoclimate studies and natural-resource management in the Murray-Darling Basin II: unravelling human impacts and climate variability, Australian Journal of Earth Sciences: 60:5, 561-571,

    Haven’t had time to read in detail but look interesting. Oh subscription required for the full paper, abstract via the link


  9. Debbie October 8, 2013 at 5:08 pm #

    Love your concluding statement Jen.
    Pronouncements are just that. . . Pronouncements. The widely divergent results are not particularly useful in a practical sense.

  10. Luke October 8, 2013 at 8:11 pm #

    Well hands up who a widdle go at reading the paper – nobody it seems. Well if you did I think Jen has been far too harsh in the intent. I’d be interested to see what Robbie and Debs thinks about figures 9 and 10.

  11. jennifer October 8, 2013 at 8:51 pm #

    If anyone wants a copy of the paper to read…

    [1] ‘Climate projections for Australia: a first glance at CMIP5’ by Damien Irving, Penny Whetton and Aurel Moise, published in Australian Meteorological and Oceanographic Journal, Volume 62, Pages 211-225.

    Send me an email to jennifermarohasy at

  12. Davefromweewaa October 8, 2013 at 8:55 pm #

    Intent Luke? Good intentions are no excuse for crap models, are they Luke?
    BOM’s models led them to predict above average rainfall for winter, it didn’t turn up. Then they predicted above average spring rain in August and backflipped on that in September. Might as well ask Elle McPherson as take any notice of BOM’s models.

  13. Luke October 8, 2013 at 10:31 pm #

    We’re not talking about BoM’s seasonal forecasting models Dave. But thanks for playing. BTW sample size of one doesn’t mean anything. And they don’t “backflip” – the forecast is the forecast. If you don’t read the accompanying material with such information don’t bother using the information. It said ” Such odds mean that for every ten years with similar climate patterns to those currently observed, about six or seven years would be expected to be wetter than average over these areas, while about three or four years would be drier.”

    The forecast is NOT categorical.

  14. Luke October 8, 2013 at 10:32 pm #

    Jen the paper is at

  15. Luke October 8, 2013 at 10:39 pm #

    Well given the blog has descended into random news service sound bits (thanks Neville) we might actually examine the paper Jen has editorialised. How novel. Some people call this reading.

    It says

    “The pronounced lack
    of model agreement over the Murray Darling Basin (MDB)
    during summer, autumn and winter and Eastern New South
    Wales (ENSW) during autumn and winter is also noteworthy,
    given the importance of the regions as agricultural and
    population centres respectively (Table 2, grey text).”

    ” In considering the level of confidence associated with
    regional climate projections derived from any ensemble of
    model simulations, one vital piece of information relates
    to the ability of the models to accurately reproduce key
    features of the present day climate (e.g. Irving et al. 2011).
    It is evident from the analysis presented here that many of the
    deficiencies that have been identified in CMIP3 simulations
    of the present day Australian climate still persist in the
    CMIP5 ensemble. Of particular note are the large biases in
    the intensity of both the monsoon in northern Australia and
    summer rainfall over central Australia, and the pronounced
    errors in reproducing the climatological mean annual
    rainfall cycle over the Tasman Sea and southern Australia.”

    So Jen – you have totally over claimed what the paper asserts IMO. The paper isn’t overly hyped at all and one might be less than enthusiastic about many aspects of model performance. The authors say as much….

  16. jennifer October 8, 2013 at 11:09 pm #


    Until this last Winter 2013 the BOM was using statistical models for its seasonal forecasting. So this discussion might not have been relevant.

    HOWEVER, the BOM is now using General Circulation Models for its seasonal forecasting, in particular POAMA.

    Yet there is no evidence to suggest that POAMA has any skill at all in generating a seasonal rainfall forecast. In fact, I would suggest that Elle McPherson would be able to give you a better forecast than POAMA by simply calculating the long term average for the season/month of interest.

    The nation, in particular the industry bodies representing you, should be up in arms that the BOM could have changed from using statistical models to GCMs for your seasonal forecast. But silence.


    The paper, as per your quote, acknowledges “the pronounced lack of model agreement”. And I see in Table 2, that the estimated decline of 1.8% is plus or minus 20.8%! For Autumn its +0.7 degrees C, plus or minus 30.4%!

    Its embarrassingly bad.

  17. Luke October 9, 2013 at 12:08 am #

    Jen – it’s not “embarrassingly bad” – it’s what it is ! You are simply selectively hyping an objective report.
    If it was “embarrassingly bad” why publish it? Come on …. Clearly directions for improvement are indicated and in specific mechanisms – this is what science does and Australia is a fraction of the global use. The authors have not suggested any in-field use of the analysis!?

    Then off-topic on POAMA

    “no evidence to suggest that POAMA has any skill at all in generating a seasonal rainfall forecast”

    patently untrue and you have spent no time looking have you?

    Otherwise you’d be quoting

    Industry bodies should be up in arms that you are slandering BoM’s latest product with no evidence? How can you make outrageous claims with no substantiating evidence and easy to find evidence to the contrary. (Google!)

    Bazza rang Elle and she said – show us your skill scores !

  18. Debbie October 9, 2013 at 6:49 am #

    That. . . . ‘ pronounced lack of model agreement’ has much to do with the idea that ‘one might be less than enthusiastic about many aspects of model performance’.
    In the case of the MDB. . . it is embarrassingly bad!
    We would likely be almost as accurate by flipping a coin out here. . . esp re rainfall predictions.

  19. Luke October 9, 2013 at 7:21 am #

    Nope – where’s your evidence? Your selective qualitative, “I’m only interested in the broad social aspects” memory? So how many years have you been following this model now Debs?

  20. cohenite October 9, 2013 at 7:22 am #

    Ho Hum, BOM has it is head up its bum; as usual. Classic luke:

    “Jen – it’s not “embarrassingly bad” – it’s what it is !”

    Figures 9 and 10 are meaningless because of the large uncertainty coded by the paper as “Models agree on a substantial change in rainfall, but do not agree on the direction of that change”. Statistically this means no change at all

    Most of all the paper does not consider Australia’s past rainfall history:

    What is an increase or decrease compared to that?

  21. Debbie October 9, 2013 at 7:54 am #

    Don’t you mean THESE (as in plural) MODEL(S) Luke? 🙂
    Just check whom/what I was quoting.
    I’m gonna take a wild guess here and suggest that I would be more likely to be interested in the practical application of GCMs than you would likely be.
    Got just a teensy bit to do with the industry I am involved in.

  22. Luke October 9, 2013 at 8:35 am #

    “Most of all the paper does not consider Australia’s past rainfall history:”

    Try reading the paper Cohers – Figure 8.

    Debs – You haven’t been using any of these models at all for rainfall prediction. Not really the point.

  23. jennifer October 9, 2013 at 9:06 am #

    It is extraordinary the junk dressed up as science published by modern climate ‘scientists’.

    It is extraordinary that the BOM is using a GCM to produce its seasonal forecasts when GCMs can’t forecast rainfall out more than a week with any level of skill.

    And it is extraordinary that this rubbish (run on $30million dollar computers) is defended by Luke.
    OK. Not surprising that Luke defends the indefensible. And Luke could you please give others a go at this thread. We’ve read your first comment, then your about face. Now back off a bit please.

  24. Luke October 9, 2013 at 9:29 am #

    So Jen can’t defend her position except to hand wave ! Data and analysis Jen – data not handwaving. You are simply making wild accusations on seasonal forecasting. No evidence. No argument.

    To conflate weather forecasting and seasonal forecasting indicates a pretty clueless position Is that what you think happens? Really Jen? This is tiresome old sceptic shell game stuff.

    As for change about face – well my comment was that “all-in” GCM “averaging” isn’t that useful. Paper concludes as much. Suggests way forward for progress.

    Jen nobody else was having a go – you’ve got zero serious input except random grizzles ! Cheer squad stuff. What input “on topic” have you had except mine? The only reason you want me to back off is that you guys can make sledging comments with no justification. So much for an evidence based blog?

    But last comment from me. Goooonnneeee !

  25. cohenite October 9, 2013 at 9:37 am #

    I did read the silly thing luke; figure 9 gives trend; my link shows the variability from year to year which the paper’s figure 9 masks. The point is that variability is huge so the paper’s inability to tell whether future changes in rainfall will be up or down is meaningless but ironically the most valid point it makes.

  26. bazza October 9, 2013 at 10:09 am #

    Jen, I thought you knew a bit more than your vocal yokels but really even the ever-loyal acolyte Cohenite should tell you that you have lost the plot claiming “It is extraordinary that the BOM is using a GCM to produce its seasonal forecasts when GCMs can’t forecast rainfall out more than a week with any level of skill”. There was one worse. Jen launched her neural net seasonal forecasting paper on this blog with the claim that it was “One way to help shift an accepted scientific paradigm “ (AGW). I wonder how it performed these last few months forecasting the Qld drought and why has it not been commercialised yet.?
    I was surprised you were not aware that multi-model ensembles were superior – no wonder you had to find another attack. You might have hilited how the article in question confirmed scary temperature trends and continued drying in SW WA. I am not sure even Debbies risk management style could handle +4 C.

  27. jennifer October 9, 2013 at 10:27 am #

    Hi Bazza,

    There are about 130 people working on POAMA at the BOM in Melbourne?

    The GCM has been under development for about 15 years?

    Now a couple of basic questions:

    1. Can POAMA reliably produce a seasonal or monthly forecast better than climatology/the long term average?

    2. Can POAMA reliably produce a more skilful forecast than the old statistical method?

  28. Robert October 9, 2013 at 10:55 am #

    Multi-model ensembles sound superb. “Scary” temp trends, continued drying, temp increases Deb couldn’t handle. Yikes! Sounds like Australia in the early 20th century! Or early 21st, for that matter. Or the “scary” 1790s in Sydney, with that “scary” mother-of-all-monsoon-failures in India.

    The Jehovahs Witnesses came by yesterday. Wish I could have shown them the real millennarian stuff. You might want to send some on to them, Bazza. They’re very big on “scary” climate change. They’ve had to post-cancel a couple of Armageddons…but you know how to do all that.

  29. Debbie October 9, 2013 at 11:52 am #

    Good questions Jen,
    Bazza and Luke. . . People like me want this type of work to get better and more useful. At the moment it isn’t adding much value in terms of risk management. . . It doesn’t matter how many different ways you try to spin it.

  30. cohenite October 9, 2013 at 12:58 pm #

    No need to get personal bazza; you say:

    “I was surprised you were not aware that multi-model ensembles were superior”

    That’s because they’re not, certainly not with precipitation:

  31. davefromweewaa October 9, 2013 at 8:48 pm #

    Hahaha, multi model ensembles Bazza.
    Let’s ask Elle McPherson, Naomi Campbell and Claudia Schiffer!
    +4c Bazza? What should we do about that? Pay more tax perhaps? If we pay enough tax can we get ice back into the Bingara glacier?

  32. Luke October 10, 2013 at 12:31 am #

    Hey Dave maybe ice caused Waa Gorge

  33. bazza October 10, 2013 at 9:50 am #

    Cohenites equally desperate attempt to support Jens hasty and unresearched critique of multi-model ensembles has backfired badly. He has in fact supported my evidence of the superiority of multi-model ensembles based on the Dilbert analogy. Given 19 models and multiple criteria it is not going to be hard to find an individual model that wins on one criteria. As the article Cohenite linked to states:
    “Also, identifying a best model depends on the purpose of the analysis or the question being asked. For example, CCSM4.0 captures the spatial pattern of temperature trends better than all other models; however, it underestimates long-term persistence and overestimates the global-average temperature trends (50% or more)”.
    But I thank Cohenite for generously providing a link to an article which confirmed CO2 as the main driver of global warming.

  34. Johnathan Wilkes October 10, 2013 at 1:56 pm #

    you need tinting, using bazza as your alter ego is far too transparent!

  35. Luke October 10, 2013 at 3:15 pm #

    Nice try JW – but if you had been around for a while I don’t think you would come to that conclusion. Besides Bazza is even better looking and richer than myself. Well hung too (I’m told).

  36. cohenite October 10, 2013 at 5:15 pm #

    “But I thank Cohenite for generously providing a link to an article which confirmed CO2 as the main driver of global warming.”

    You’re delusional bazza.

    The paper has 3 interesting aspects; firstly it applies Hurst criteria to both temperature and precipitation and finds that temperature exhibits LTP which means the climate system has negative feedback; secondly, it finds there is far less LTP in precipitation which means that temperature does not define water although the opposite can and obviously be the case; and since CO2 is temperature dependent that would mean CO2 is also water dependent.

    Thirdly even though their models were ok with temperature, they were not ok at a regional level, which is what Koutsoyiannis has also found. Given that temperature is a [bad] proxy for energy, which is the true and crucial factor of AGW, and that GMT anomalies can go up while total energy in the system can go down based on regional differences, their findings about temperature have nothing to do with AGW.

    Impress me and explain how GMT can go up but global energy balance goes down.

  37. cohenite October 10, 2013 at 9:18 pm #

    Hey SD, did you ever go back to Deltoid about rising sea levels; did you argue the issue of rainfall unpredictability?

  38. ghl October 15, 2013 at 9:38 pm #

    It does not matter how many people or dollars are committed to models, both rainfall and temperature. They will always give a wide spread. They must include average/normal/realistic results, if they do not include reality, they prove themselves useless. If they do not also produce catastrophic extremes, they do not scare anyone.

  39. Leo G October 19, 2013 at 9:17 pm #

    “I would insist that only the best model, or combination of models, be used to provide a final figure.”
    But how much more convenient, if you don’t have any genuine confidence in the predictive capacity of your models, to use the two opposing extreme outliers to delimited your prediction. How embarrassing when reality can’t properly constrain itself. Clearly the outliers just haven’t been sufficiently extreme.

  40. Alan Herath October 21, 2013 at 4:24 pm #

    Having not read any of the papers and models etc it is worth considering (particularly in relation to streamflow in the MDB) that the average hardly ever occurs. For example,in the extreme, if there are only droughts and floods with nothing in the middle then “the average” is irrelevant. Perhaps the median or the distribution of all the projections is more important. “Averaging” the future rain “makes” it more susceptible to evaporation and thus being of less use to wheat crops, and after runoff to irrigated agricultural/horticultural crops.
    Better to start looking at the usefulness of future rain intensity.

  41. Mack October 21, 2013 at 8:00 pm #

    Cohenite, Spanglers, I don’t think, in the end could be bothered with the Deltoids. Had a bit of a laugh when Luke went over there for a spell, lost it, got all ratty and foulmouthed trying to argue a bit from our side. Trouble is Luke is not a “reactionary old codger” and therefore unable to cope with idiots of his kind. “Reactionary”…mmm brings back memories of SJT.

  42. Ian Thomson October 24, 2013 at 5:22 pm #

    The real climate problem to be solved. MONEY . How can it be made ?

    “There are several ways in which a meaningful global price on carbon can be achieved, ranging from taxes to market mechanisms, and each country must decide what makes most sense given its national circumstances.’
    The executive secretary of the UN Framework Convention on Climate Change, Christiana Figueres,
    BBC, via ABC.
    Tax works better, because poor people get to give bankers money, in the time honoured way

  43. Brian Hatch December 12, 2013 at 12:51 pm #

    The BOM rainfall records for the Murray-Darling basin show that rainfall has increased over the last century plus. Look at Dubbo, Cooma and Queanbeyan. Would any of the models picked what has actually happened?


  1. Jennifer Marohasy » The Nature of Inclusive Climate Science | Cranky Old Crow - October 8, 2013

    […] Jennifer Marohasy » The Nature of Inclusive Climate Science. […]

  2. Jennifer Marohasy » Same Information: Different Opinion. Part 2, The Tragic versus Utopian Vision of Climate Science - November 21, 2013

    […] 1.The Nature of Inclusive Climate Science […]

Website by 46digital