Sections

Services

Information

UK United Kingdom

Predicting El Nino: a tale of two authorities

Over the last two summers, Eastern Australia has experienced two of the hardest hitting La Niña events since 1974. Widespread floods resulted across great swaths of the country. As expected, the La Niña…

Whither the weather: the Bureau of Meteorology’s dynamic climate modelling is not the only forecasting method. AAP/Paul Miller

Over the last two summers, Eastern Australia has experienced two of the hardest hitting La Niña events since 1974. Widespread floods resulted across great swaths of the country. As expected, the La Niña conditions faded to result in the current neutral conditions as measured by eastern equatorial Pacific Ocean sea surface temperatures and the Southern Oscillation Index (SOI), a measure of regional atmospheric behaviour in the western Pacific.

The question many will be asking is “What comes next”? Will the next summer season be a repeat of the previous flooding? Or will the El Niño/La Niña Southern Oscillation (ENSO) climate pattern settle down into a neutral state? Or perhaps even a return to El Niño conditions with the prospect of deficient rainfalls?

The leading provider of seasonal forecasts in Australia is the Bureau of Meteorology, who issue a bi-weekly ENSO update. The most recent BoM ENSO update came on 5th June 2012, and states:

Tropical Pacific climate indicators remain at neutral values for this time of the year. This includes the Southern Oscillation Index (SOI), trade winds, cloudiness, and sea surface temperatures. Ocean temperatures below the surface are currently warmer than average in the central and western Pacific on a monthly scale, with the eastern subsurface Pacific closer to normal, but slowly warming.

Climate models surveyed by the Bureau of Meteorology show that the tropical Pacific Ocean is likely to warm further over the coming months. All seven models surveyed indicate conditions are likely to approach, or possibly exceed, El Niño thresholds during the late winter to early spring period. Large parts of eastern Australia are typically drier and warmer than normal in winter/spring as El Niño events develop. No climate models favour a return to La Niña.

This most recent update first acknowledges the current neutral conditions but then goes on to state that further warming is expected. Of note, all seven models surveyed indicate El Niño or near-El Niño conditions to develop. Perhaps of even more relevance is that simple observation that no climate models indicate a return to La Niña. So on the basis of this synopsis of climate models, El Niño and reduced rainfalls seems highly likely, whereas La Niña and potentially another series of summer floods events seems highly unlikely.

Indeed, the BoM’s manager of climate monitoring, Dr Karl Braganza, seemed very certain of a coming El Niño this week when he spoke to ABC’s Radio National. While the strength of the coming El Niño was as yet unknown, Dr Braganza said, the modelling left little doubt it is on its way:

Look, it’s looking reasonably straight forward now in terms of the way the models are going. When you get a whole lot of models going in one direction that increases your confidence that they’re doing something real.

Rain or shine? Flickr/PNNL Pacific Northwest National Laboratory

Dr Braganza’s confidence is misplaced as it is based on only one approach to ENSO prediction - dynamical climate models.

In contrast, whilst the US National Oceanic and Atmospheric Administration (NOAA) is similarly engaged in forecasting ENSO, it also surveys statistical models of ENSO behaviour.

The difference between the two approaches is that BOM’s attempts to simulate everything in the ocean-atmosphere system, whilst the other attempts to compare current ENSO behaviour to the statistical historical occurrences of ENSO. Dynamical climate models are based on the simulation of the physics of ocean-atmosphere interactions requiring substantial data inputs and computational resources.

In contrast, statistical models are much simpler as they are based on the recent observations of the leading indicators of ENSO initiation and growth. The likely ENSO behaviour is then based on the previously observed ENSO behaviour.

The US authority NOAA therefore provides a more comprehensive survey of the alternative approaches to predicting ENSO when compared to the climate models used by the BoM. NOAA have recently released (7th June 2012) their ENSO diagnostic discussion.

On the basis of a greater diversity of models and most importantly, modelling approaches, NOAA have declared their ENSO Alert System Status to be “El Niño Watch”. That is, they say that El Niño conditions may be developing and that we should be on guard for that possibility. Importantly, NOAA is not prepared to definitively call an El Niño at this early stage of the ENSO cycle. Unlike the Bureau, their surveyed models display a range of possible outcomes. They state:

… most of the dynamical models predict El Niño to develop during JAS (July, August, September), while the statistical models tend to favor the continuation of ENSO-neutral. Thus, there remains uncertainty as to whether ENSO-neutral or El Niño will prevail during the second half of the year. The evolving conditions, combined with model forecasts, suggest that ENSO-neutral and El Niño are roughly equally likely during the late northern summer and fall. The CPC/IRI forecast calls for ENSO-neutral conditions through JAS, followed by an approximately 50% likelihood for El Niño during the remainder of the year.

NOAA offer a note of caution that the Bureau should take into account – it is still very early in the typical annual ENSO cycle to make strong predictions, and the different approaches to ENSO prediction disagree. Undeterred by this reality, the Bureau have put all of its faith in the complex dynamical climate models, ignoring the other approaches to prediction.

The dynamical climate models predicting early El Niño development require large increases in sea-surface temperatures over the next two months. Indeed, closer inspection of the individual model predictions reveals that they expect a 0.1-0.2 degree Celsius warming for the month of June, followed by a whopping 0.6-0.7 degree rise over the month of July alone.

In contrast, NOAA’s survey of statistical models of ENSO are largely based on behaviour of the observations alone. These appear to be far more conservative in that they are indicating neutral conditions to persist.

I would argue that it is certainly too early to be definitive about the expectations of ENSO for the next summer period. The Bureau’s apparent confidence is almost certainly premature.

El Niño does appear a possibility at present, but what we do know for certain is that it is only over the next few months that we will be able to gauge where ENSO goes and whether the climate models or statistical models were right. It is only then that we might prepare for what ENSO might throw at us next.

If an El Niño event does emerge over the next few months, then it would be reasonable to expect a drier, warmer summer than we’ve had over the last two years. One can also argue that if an El Niño does emerge, we enter it in the best position possible as eastern Australia is currently primed with water.

Perhaps the most important insight gained from the Australian and US authorities is not so much the chances of El Niño forming - rather that La Niña conditions look highly unlikely.

This itself should serve to provide some form of comfort to flood-ravaged Australians: south-east Queensland and northern New South Wales are particularly hit by ENSO as we have witnessed with the two summer seasons of La Niña-driven floods. The prediction that La Niña is highly unlikely should mean a much reduced risk of seeing the catastrophic flooding of recent summers.

Above all, whether El Niño or neutral conditions persist, in either case it should give low-lying floodplain dwellers some welcome relief.

Comments welcome below.

Articles also by This Author

Sign in to Favourite

Join the conversation

57 Comments sorted by

  1. Karl Braganza

    Manager, Climate Monitoring Section at Australian Bureau of Meteorology

    Stewart thanks for clarifying the inherent uncertainty in predicting El Nino or La Lina events many months in advance. This is really important.

    I agree that it is still too early to make a definitive call on whether we will have an El Nino event later this year. This is captured in the Bureau's official forecast.

    As Stewart suggests, perhaps the most important inference from the models is that, out of three possible outcomes — El Nino, La Nina and neutral conditions — a La Nina event seems…

    Read more
    1. Bob Beale

      Journalist

      In reply to Karl Braganza

      Speaking as one who lives on a low-lying floodplain, at least Franks has reassured me by making it very plain who I should trust. He offers a two-step argument as to why we should trust NOAA’s forecast more than the Bureau of Meteorology’s. i.e.:
      1: The Bureau’s forecast is based only on dynamical models and ignores other approaches to prediction;
      2: NOAA provides a more comprehensive survey by including statistical models as well.
      Yet the the Bureau’s climate prediction team say it does include statistical models and that NOAA and BoM liaise closely and don't have different approaches. The head of NOAA’s climate prediction team confirms this.
      In short, it seems reasonable to conclude that Franks is ignorant of how the Bureau’s seasonal forecasts are prepared and does not realise that NOAA and BOM agree not only in their methods but in their conclusions.

      report
    2. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Karl Braganza

      Hi Karl - thanks for your response and your candour

      As you note - the coupled models are pointing more to El Nino, than NOAA's statistical models. It's good to survey them also, but the BoM ENSO outlook refers to the coupled models only. I do think that this is not representative of the true uncertainty, especially this early in the ENSO cycle.

      I sympathise with the difficulties in getting a simple message across through the media - perhaps 'early signs of a possible El Nino' is less vulnerable to misinterpretation.

      In any case, I think that we do agree on the substantive insights from current indicators/models - that there is a diminished likelihood of La Nina - an important and hopefully more robust insight than any current El Nino prediction. Also that the next few months can differentiate between the climate and statistical models as the cycle progresses.

      Best wishes, Stewart

      report
    3. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Bob Beale

      Hi Bob

      NOAA and BoM both produce periodic updates on ENSO. BoM provides an outlook - the most recent demonstrates 7 of 8 climate models predicting El Nino (87% at face value), but does not cite statistical models at all. Consequently, the recent outlook provides a false confidence in models used to predict ENSO.

      NOAA explicitly discusses statistical models which do not predict a rapid shift to El Nino conditions. Consequently, it's ENSO discussion is inherently more balanced.

      I think we do all agree that NOAA's probability of El Nino being about 50% is probably appropriate. You would not reach this conclusion reading the Bureau's outlook and survey of 8 models (7 El Nino)

      It is currently too early in the typical ENSO cycle to be as definitive as 7/8

      report
    4. Bob Beale

      Journalist

      In reply to Stewart Franks

      So now you say that your real beef is the "substantial difference between the BoM Outlook and the NOAA ENSO discussion - the lack of any discussion of the statistical models". You've changed your story. It's not the way the forecast was made that you have a problem with, but the way it was discussed after the fact?
      You unequivocally accused the BoM of not surveying models that it did in fact survey. You cast doubt on the credibility and reliability of the Bureau, with no apparent basis in reality…

      Read more
    5. In reply to Bob Beale

      Comment removed by moderator.

    6. Bob Beale

      Journalist

      In reply to Ian Ashman

      Ian Ashman's response is a spoof, folks (lest anyone take it seriously).

      report
    7. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Bob Beale

      Bob - I've not changed my story at all

      If you read it, you'll see that it quotes and links to NOAA where they survey statistical as well as dynamical model. It also quotes and links to the BoM ENSO Outlook, where only the dynamical is presented.

      By focussing the outlook solely on the dynamical models, the uncertainties associated with the outcomes of the statistical models is not presented.

      No agenda other than to point out the false confidence communicated to the public through the Outlook's over-emphasis on the dynamical models.

      report
    8. Bob Beale

      Journalist

      In reply to Stewart Franks

      You're wriggling. You said above: "Dr Braganza’s confidence is misplaced as it is based on only one approach to ENSO prediction – dynamical climate models.
      In contrast, whilst the US National Oceanic and Atmospheric Administration (NOAA) is similarly engaged in forecasting ENSO, it also surveys statistical models of ENSO behaviour."
      Both the Bureau and NOAA say that's incorrect - they each survey both models. And their conclusions agree. As I read it, the issue here is false doubt, not false confidence.

      report
    9. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Bob Beale

      Bob

      Karl is quoted as saying 'When you get a whole lot of models going in one direction...'

      He clearly is not talking about the statistical models because they are not going in one direction along with the dynamical models

      Look at the BoM Outlook - no mention of statistical models, no mention of anything other than dynamical models. Surely this cannot be disputed???

      report
    10. Karl Braganza

      Manager, Climate Monitoring Section at Australian Bureau of Meteorology

      In reply to Stewart Franks

      Hi All,

      Last post from me.

      The Bureau does look at statistical models and we liaise with NOAA and other met services. We take in what those groups are looking at — and take advice from them as well. The Bureau's outlook is very similar to NOAA's. Hopefully we have clarified this for the readers here.

      If the issue is my statements on the ABC. I have clarified those in my first post. Stewart and I are not in disagreement about the ENSO outlook. I think we can move on.

      Specifically to…

      Read more
    11. In reply to Bob Beale

      Comment removed by moderator.

  2. Andrew Watkins

    Manager Climate Prediction, National Climate Centre at Australian Bureau of Meteorology

    Its good to see Stewart highlighting the uncertainty in the models, just as we do in our statements at the Bureau. On the model summary page ( http://www.bom.gov.au/climate/ahead/ENSO-summary.shtml ) we offer a higher level of assessment of the models. This highlights that "However, results from individual models continue to show a moderate to high level of spread in their forecasts, meaning there remains a reasonable level of uncertainty in this outlook. Climatologists will continue to monitor conditions and outlooks closely for any further developments over the coming months."

    report
    1. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Andrew Watkins

      Hi Andrew

      Thanks for your contribution. Uncertainty is indeed the key here. Irrespective of any prediction, it is good to develop the dialogue as ENSO and its indicators progress.

      I do believe that by focusing on the coupled dynamical models in the ENSO Outlook that a false impression of certainty is provided. If statistical models are surveyed it might also be useful to cite them as NOAA did in their recent discussion document.

      Certaintly, the next few months will provide greater clarity as to ENSO developments.

      Best wishes, S

      report
    2. John McLean

      logged in via email @connexus.net.au

      In reply to Andrew Watkins

      But Andrew, how good are those models?

      I look back to September 2011 (see http://www.bom.gov.au/climate/enso/archive/ensowrap_20110914.pdf) and most of the models were predicting Nino 3.4 temperatures to rise.

      Two months later (see http://www.bom.gov.au/climate/enso/archive/ensowrap_20111012.pdf), when temperatures had remained flat for 2 months, most of the models were predicting the next 4 months to be cooler than it was in early November.

      That's a huge swing in the models in just two months, and these are models that can supposedly predict Nino 3.4 temperatures up to nine months ahead.

      report
    3. Alvin Stone

      logged in via Facebook

      In reply to John McLean

      From a lay point of view, two things stand out in the pages you have referenced.

      1. The La Nina did indeed form, just deeper than expected.

      2. BOM specifically says: "Please note the Bureau’s POAMA model has recently been upgraded. The NINO3.4 forecast shown is for the new
      version of the model, POAMA2.4."

      Obviously the upgraded model performed better than the earlier one, as would be expected. The swing seems primarily a result of improvements, unlike the arguments of climate change deniers…

      Read more
  3. Alvin Stone

    logged in via Facebook

    Just one question, historically which of the two approaches - BOM's or NOAA's - is more accurate in forecasting ENSO conditions? It's all very well to say the two use different methods but if the outcome is the same, it shouldn't be an issue.

    report
    1. Andrew Watkins

      Manager Climate Prediction, National Climate Centre at Australian Bureau of Meteorology

      In reply to Alvin Stone

      G'day Alvin,

      Actually, NOAA and BoM don't have different approaches - I know, I speak to Michelle L'Heureux (their head of climate prediction) a couple of times a month at peak periods! We both assess a range of models and a range of observations and come to a conclusion about future ENSO possibilities. They definitely use a range of international dynamical models just as we do.

      On the dynamical and statistical model front - Tony (Anthony) Barnston from the IRI has done a lot of work on this…

      Read more
    2. Alvin Stone

      logged in via Facebook

      In reply to Andrew Watkins

      Thanks Andrew, it's great to have people at the coal-face answering questions. Cheers.

      report
    3. Mike Hansen

      Mr.

      In reply to Andrew Watkins

      So Franks' article is much like his previous contribution to The Conversation - a beat up. Another strawman bites the dust.

      report
    4. Derek Bolton

      Retired s/w engineer

      In reply to Andrew Watkins

      Andrew, would I be right in thinking that both approaches produce not only a prediction but a std dev? That the std dev is higher for the stats method on those occasions when models do significantly better? So there is a way to weight the two predictions to get the best overall?

      report
    5. Karl Braganza

      Manager, Climate Monitoring Section at Australian Bureau of Meteorology

      In reply to Derek Bolton

      Hi Derek,

      That is a great question- and a whole field of research!

      Unfortunately, there is no simple way of combining such forecasts. While you can do a range of calibrations - such as model weighting, and variance adjustments - to give you good verification results over a historical period, that does not necessarily reflect the actual forecast accuracy of the multi-month prediction scheme.

      Additionally, there is a difference between the confidence of the forecast, and the accuracy of the…

      Read more
    6. Michelle L'Heureux

      Meteorologist

      In reply to Andrew Watkins

      I'd like to confirm what Andrew has written here. There are times when the Bureau and NOAA emphasize slightly different aspects of ENSO and the outlook, but this is not one of them. While NOAA has officially declared chances of El Nino at 50%, it's important to remember that this is not a simple coin flip between two outcomes. There are actually three possible outcomes (El Nino- Neutral- La Nina) and so a 50% chance of an El Nino in the coming months is actually implying that we are favoring those…

      Read more
    7. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Alvin Stone

      Hi Alvin,

      There are two different classes of models relevant here - statistical and the dynamical coupled models. The point is that their outcomes are not always the same. In the current case, the statistical models are more conservative and do not result in El Nino. The coupled models tend to El Nino.

      My article is to highlight the fact that if one cites one class and not the other, then the true uncertainty of the ENSO forecasts is not conveyed. For this reason, NOAA's discussion is more balanced and appropriate than the Bureau's with its sole focus on the dynamical coupled models.

      Best wishes, S.

      report
    8. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Andrew Watkins

      Dear Andrew

      I fully agree that the different classes of models are complementary not competing.

      It is for this reason that my article highlights one key substantial difference between the BoM Outlook and the NOAA ENSO discussion - the lack of any discussion of the statistical models. When 7/8 coupled models predict El Nino and when the statistical models don't, then to present just one of the classes of models provides a false confidence in the likely ENSO outcome.

      report
    9. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Michelle L'Heureux

      Hi Michelle

      Many thanks for your contribution to this discussion.

      I think that we all agree an El Nino is favoured, and that 50% chance of El Nino is probably appropriate.

      The article itself highlighted the key difference between the BoM and NOAA in their ENSO discussions. the BoM presented only 8 dyanmical models, 7 of which indicate El Nino (87.5%). In contrast, the NOAA discussion was explicit in its consideration of statistical models for ENSO prediction.

      This is an important distinction, especially if the two classes diverge in their projections as this then does affect the perceived confidence of the predictability of ENSO.

      Best wishes, S.

      report
    10. Dan Collins

      Meteorologist

      In reply to Karl Braganza

      Karl,

      As a member of the NOAA prediction group that provides information for ENSO forecasts, Michelle L'Heureux suggested that I may want to clarify the work on climate prediction related to ENSO at NOAA. There certainly are a number of ways to put the dynamical models and the statistical (historical) information together, and this is an area of ongoing research and development at NOAA.

      Currently the systematic biases (in both mean and spread) in dynamical models are corrected using historical…

      Read more
    11. John McLean

      logged in via email @connexus.net.au

      In reply to Mike Hansen

      Sticking to the subject would be appreciated.

      report
  4. Hugh Sturgess

    Student

    So, going from the comments by Andrew Watkins and Karl Braganza, BoM highlighted uncertainty in its predictions quite frankly. Moreover, NOAA and BoM both use the same methods - both coupled models and statistical analyses - of which the models have performed best for the last decade. Stewart, none of this came across in your article. Indeed, you state that the BoM put "all its faith in models", which Andrew has made clear is not the case.

    Perhaps, as the author of the article about how Tim Flannery wasn't an expert in anything climate related, you should think about how well placed an engineer is to make definitive statements about meteorology?

    report
    1. Paul Gregory

      Honorary Fellow at University of Melbourne

      In reply to Hugh Sturgess

      I thought the idea of The Conversation was to give experts in their fields the opportunity to present information to the academic community and general public at large.

      This piece seems to be nothing more than an op-ed. There are no validation statistics or data to support any of the author's contentions. Rather, he seems to have selectively cut'n'pasted several media quotes and statements from the Bureau in order to push his own agenda (I am unsure what it is at this point).

      Surely these sorts of articles are better suited to a blog.

      report
    2. Matthew Thompson

      Editor at The Conversation

      In reply to Hugh Sturgess

      Dear Hugh,

      Thanks for taking part in the discussion.

      Stewart may work in Newcastle University's School of Engineering, but that doesn't make him an engineer. His profile should give you an idea of his academic qualifications, which are indeed climate-related:

      https://theconversation.edu.au/profiles/stewart-franks-7543

      Yours,

      Matt.

      report
    3. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Hugh Sturgess

      Hi Hugh

      There is a substantial difference between the way in which the Bureau reports its ENSO outlook and NOAA's ENSo discussion. Most obviously in the citing of statistical models which do indeed indicate different outcomes to the dynamical coupled models. If the Bureau did cites these in their outlook, there would be a more appropriate balance and represention of the predictive uncertainty

      Flannery has nothing to do with this. Please check my profile re: expertise

      report
    4. John McLean

      logged in via email @connexus.net.au

      In reply to Paul Gregory

      Hmm, an academic with a background in ENSO and hydrology is accused of not being capable of comparing the outputs and predictions from ENSO models used by the BOM and NOAA ?

      I don't think your argument holds water at all.

      report
    5. Alvin Stone

      logged in via Facebook

      In reply to John McLean

      No, an academic is highlighting the lack of validating statistics or data to support the position of this article.

      Fair enough too if the author is an authority on ENSO and hydrology.

      report
  5. Michael J. I. Brown

    ARC Future Fellow and Senior Lecturer at Monash University

    The quote fragment used by Stewart Franks in the above article does not seem representative of the interview as a whole (http://www.abc.net.au/pm/content/2012/s3523615.htm). Quite cautious language is used by Karl Braganza and Gramae Anderson (and we don't know if other cautions and caveats were edited out).

    For example...

    KARL BRAGANZA: As you get into June and July you start to get a bit more certainty, the system starts to get going in a particular direction and once it gets going it tends to be easier to predict.

    And most of those models are saying probably neutral to El Niño conditions over the next few months but then out into spring a lot of the models are going for an El Niño event.

    report
    1. John McLean

      logged in via email @connexus.net.au

      In reply to Michael J. I. Brown

      I'll point out the obvious - we are in June.

      Another obvious - as per the 5 June statement, all the BoM models predict a sharp rise towards warming, many crossing the threshold by July and even those that don't cross it tracking very near to it, as if that made much practical difference.

      As Stewart Franks says, the fact that all these models go in one direction whereas NOAA models predict a wider variety of possibilities suggests some fundamental differences between the models used by the two organizations.

      report
    2. Alvin Stone

      logged in via Facebook

      In reply to John McLean

      I'll point out the obvious. As noted in the comments above, NOAA and BOM are in constant conversation and are in rough agreement. They are simply reporting on what they are seeing in their versions.

      NOAA say historically the odds are in favour of El Nino.

      BOM says the models strongly suggest an El Nino.

      There is of course a margin for error, as both NOAA and BOM clearly spell out, although this seems to be ignored by the writers.

      Everything after that is simply playing around the margins to cast doubt on the authority of BOM. After all, if you cut down the leading forecasting authority in this country, in exactly the same way you attack the CSIRO, you can create more doubt and undermine climate science.

      Of course, you could come up with some robust science yourself that proves all those climate scientists have got it wrong. The Nobel prize awaits.

      report
  6. Graham Readfearn

    logged in via Twitter

    It seems this article had two points to it.

    One was to criticise Karl Braganza for being "very certain" of a coming El Nino and for having "misplaced confidence".

    Another was to demonstrate that there's more than one way to skin a cat, the cat being predicting El Nino/La Nina and the skinning method being two different types of models.

    Yet it now appears that for his first point, Stewart Franks has quote-mined the ABC's report to construct his argument, leaving out Braganza's more cautious…

    Read more
    1. In reply to Ian Ashman

      Comment removed by moderator.

    2. John McLean

      logged in via email @connexus.net.au

      In reply to Graham Readfearn

      I'll ignore your closing irrelevant paragraph and comment only on what preceeds it.

      Franks is hardly responsible for BoM comments that express caution and elsewhere a BoM employee, Braganza, speaking with possibly misplaced confidence and without the caveats in the BoM statement.

      report
    3. In reply to John McLean

      Comment removed by moderator.

  7. Michael Silverton

    logged in via Facebook

    The picture of the computer room takes me back to my youth. I would guess it would have to be at least 30 years old.

    report
    1. Stewart Franks

      Professor School of Engineering at University of Newcastle

      In reply to Michael Silverton

      Dear Michael

      I liked the photo too - did you use such facilities? I'd expect that the programmers would be very careful with their code given the set up and runtimes.

      These days my students laugh when they find a floppy disk under a pile of books in my office... what do you do with that? they ask.

      Best wishes, S.

      report
    2. John McLean

      logged in via email @connexus.net.au

      In reply to Stewart Franks

      I hope that your students don't think that because a computer provided an answer then the answer must be correct.

      We see it much too often. Complex situations with some well-understood components are bundled with poorly understood components and the output from the resultant model treated as gospel.

      I recall back in the late 1970s my university lecturer laughed at a CSIRO model that included one factor to a precision of four decimal digits (i.e. ten-thousandths) but only approximated another to a multiple of ten (whole numbers) because the first was well understood but the second poorly.

      I wonder if this is the case with these ENSO models.

      report
    3. Alvin Stone

      logged in via Facebook

      In reply to John McLean

      John, do you work with climate models? You seem to be such an authority on how they work.

      report
  8. Gil Hardwick

    anthropologist, historian, novelist, editor and publisher at eBooks West

    Thinking through this, I take the view that there are diminishing marginal returns in trying to make the science too exact, the predictions too precise, when the vagaries of nature will anyway frustrate the effort.

    I have written on a number of occasions of the need to get the business models in order. Given a reasonable allowance for error in climate and weather forecasting it is not too difficult to factor that error into the business model. Farming is frugal, conservative, anyway. There are…

    Read more
    1. John McLean

      logged in via email @connexus.net.au

      In reply to Gil Hardwick

      Unfortunately some people set great store by the crossing of the artificial El Nino and La Nina thresholds. In reality the states are a continuum, so falling just short of the threshold is hardly discernible from just crossing the threshold.

      From a farmer's point of view there's a huge difference between getting rain that will save parched crops and not getting it until a couple of weeks later when it's too late. And crop insurance has become prohibitively expensive in recent years, probably due to the sustained drought from 1997 to 2007/8 in the southeast of Australia.

      report
    2. Alvin Stone

      logged in via Facebook

      In reply to John McLean

      And yet farmers set enormous store in the accuracy of BOM's forecasts, short and long-term. So, I guess they do have some practical outcome.

      report
  9. Comment removed by moderator.

  10. Comment removed by moderator.

  11. Comment removed by moderator.

  12. Richard Mackey

    Director

    What Prof Franks has written is essentially correct and uncontroversial. It is simply a statement of facts.

    There are two families of mathematical models used to predict ENSO activity. One family is a statistical one based on prior happenings. The other family is the family of GCMs. The NOAA uses the first and foolishly the BoM uses the second. The statistical families invariably give better predictions but give no insight into the dynamics of ENSO. The GCMs will always give relatively…

    Read more
  13. Iced Volvo

    Phsycian

    Franks raises some interesting issues albeit in a non technical manner. There are actually three basic climate models in use: integro/differential models aka GCMs, statistical/observational models and the "others" (which include obscure arts such as chaos theory etc.)

    I don't know much about the statistical type and only the Russians seem to be interested in "alternative" approaches but I do know a LOT about GCM models: enough to know they are basically useless as predictors of climate: The problems…

    Read more
  14. Comment removed by moderator.

  15. Michael Anderson

    Engineer

    Thanks Prof. Franks for this article. I appreciate your input both here and on Q&A recently; both were refreshingly reflective, and cool headed in an increasingly partisan and hotheaded discussion. Hope to see some more articles soon.

    report