The attribution question

It seems as though the issue of trying to attribute an anthropogenic influence to an extreme weather event is controversial on a number of levels. It sometimes seems to divide even those who largely agree, and almost always produces a push-back from those who appear to want to dismiss the risks associated with climate change. Some of this push-back is justified, but some is no better than those who make overly strong claims about an association between climate change and an extreme weather event. Not being able to make some kind of formal attribution does not mean that there is no link between climate change and these extreme events.

My own view is that asking whether or not climate change caused a particular extreme event is silly; we can’t answer that question. What we can do, though, is consider how anthropogenically-driven climate change has influenced the likelihood of certain events. Naively one might argue that in a warmer world we would expect, for example, an increase in the frequency and intensity of heatwaves and extreme precipitation events. This may be reasonable globally, but as this paper by Friederike Otto and colleagues points out, regionally it depends on both thermodynamic changes (more energy) and changes in atmospheric circulation.

To try and understand how climate change is likely to influence extreme events – regionally at least – one therefore needs to consider both warming (thermodynamics) and how it influences the movement of atmospheric patterns (circulation). Circulation changes could reduce the likelihood of some events in some regions, while enhancing it in others. The key point, though, is that this is not an attempt to determine if climate change is causing these extreme events; it is simply trying to establish how climate change influences the overall risk of these extreme event.

I may not have explained this properly, but fortunately the author is on hand to do a better job than I can.

This entry was posted in Climate change, Climate sensitivity, ClimateBall, Research, Science and tagged , , , , , , . Bookmark the permalink.

112 Responses to The attribution question

  1. John Hartz says:

    Here’s s the latest in what are known as “single event attribution” studies. This one is notable for being the first collaboration between scientists at the World Weather Attribution (WWA) project and the US National Oceanic and Atmospheric Administration (NOAA).

    Torrential rains unleashed on south Louisiana in August were made almost twice as likely by human-caused climate change, according to a quick-fire analysis released just weeks after the flood waters subsided.

    The team of scientists concluded that such an event is expected to occur a minimum of 40% more often now than in 1900, but their best estimate is that the odds are now twice as high.

    Climate change doubled odds of Louisiana heavy rains, scientists warn by Roz Pidcock, Carbon Brief, Sep 7, 2016

  2. Thanks much for pointing out the Otto, et al paper. It also has Stott and Allen on its list of co-authors, which makes it to be a must-read, in my book.

    I’ve also come across a 2014 paper by Deng and Ebert-Uphoff which, using a non-standard analysis, suggests that teleconnections are weakening with warming. climate. If so, that could be a mechanism for realizing a suspicion that, as climate warms, the rules of meteorological forecasting change, making past qualitative experiences less relevant and, even perhaps affecting the skill of meteorological models, at least in the long run.

  3. Interesting, I thought one of the justifications for statistical downscaling was that the models did a better job of (i.e. were reasonably good at) predicting changes in atmospheric circulation than (sub-) regional climate. The “how well do models reproduce/predict circulation” was something I was hoping to return to at some point (and perhaps the downscaling as well).

  4. Imback says:

    Using the word ’cause’ without a baseline context is problematic. Suppose we ask what caused my toe to be stubbed. I could say the Big Bang caused it, because if the Big Bang didn’t happen, my toe wouldn’t even exist now, but the Big Bang did happen, and my toe is clearly stubbed, so therefore the Big Bang is putatively the single cause of my toe being stubbed. But probably the question implicitly has the baseline context that the universe already exists and that I existed as I did the moment before my toe was stubbed. Then I could say that that rock there caused it. That could be true if that rock is a variable and not part of the baseline context. But that rock has been there for years, so it probably belongs in the baseline context. Then I have to admit my wandering outside at night in my bare feet must have caused my toe to be stubbed. Indeed, in the active voice, I stubbed my toe. But even then, I have wandered often like that before and have not stubbed my toe, so the full attribution cannot be to my wandering. There could be multiple causes that are outside the baseline context of the question. When I hit my toe on the rock, I was both wandering and distracted by thinking of climate change. Actually, given that I wander often, consider that my wandering is part of the baseline context too, and that the only variable remaining is what I was thinking about. Then in this case, my toe getting stubbed was caused by climate change. In the active voice, climate change stubbed my toe.

    Kidding aside, the question of whether or not climate change caused a particular extreme event is not necessarily silly. We can indeed answer that question if we understand what is the constant baseline context and what is variable. As Friederike Otto says, the tools to help answer these attribution questions are climate models. Making the changing circulation variable in the question (and not part of the baseline context) does add difficulty, but does not make it impossible or silly.

    A particular question came up this summer about how much the fires in Alberta were attributable to climate change. There had been an El Niño leading up to it. There was a massive ridge lingering over western Canada that brought hot dry air. Then winds helped spread the fires. We know El Niños happen, massive ridges happen, and winds happen. Should these basic facts be part of the baseline context of the question, and only changes to these circulation patterns as well as the actual secular warming be considered when asking that question? That is my question.

  5. Certainly, within constraints, some aspects of the general circulation of the atmosphere are predictable. Namely, the latitudinal gradient of temperature imposed by the shape of the orbiting, revolving earth, predicts the presence of westerly jet streams in each hemisphere.

    However, there is always an infinite array of circulation orientations and ultimate developments for any given gradient.

    One cannot accurately predict individual cyclones ( beyond a few days ), why would one believe predictions about cyclone frequency?
    One cannot accurately predict El Ninos, why would one believe predictions about El Nino frequency?

  6. TE,
    It seems like you’re setting impossible standards. A key point about climate modelling is that there is no sense that we are able to make predictions about the weather beyond a few days, but there is a sense that we can use models to estimate how the average of the weather (or the climate) will change under increasing anthropogenic forcings. That we may not be able to accurately predict the precise occurence of an event does not mean that we can’t determine how the frequency of such events will change as we warm. Maybe we can’t, but using our inability to accurately predict individual events as an argument against the ability to predict how their frequency will change, seems weak.

  7. John Hartz says:

    ATTP:

    The folk at http://www.climateprediction.net/ have posted an article that plows much of the same ground as your OP and includes the same video interview of Dr Friederike Otto.

    To access the climate prediction.net article, The Attribution Question, click here.

  8. John Hartz says:

    ATTP: In the OP, you state:

    My own view is that asking whether or not climate change caused a particular extreme event is silly; we can’t answer that question.

    How does this statement square with the work being done by the World Weather Attribution (WWA) Project?

  9. Andrew Dodds says:

    TE:

    I can’t accurately predict a single fair coin toss, but I’m pretty sure about the distribution of 100 such tosses.

    And, of course, given sufficient data I’d be able to detect when the coin was swapped for an unfair one.

  10. One cannot accurately predict individual cyclones ( beyond a few days ), why would one believe predictions about cyclone frequency?

    Setting apart intensity, the paths cyclones take seem more alike than not, even if these are byproducts of the weather about them, to do the degree that reasonable trajectories can be simulated. And these are physics-free projections.

  11. lerpo says:

    Hi TE,

    One cannot accurately predict individual cyclones ( beyond a few days ), why would one believe predictions about cyclone frequency?

    Gavin has a very accessible talk on chaos in the climate here: https://www.youtube.com/watch?v=oqWfN3Zl8AM

    He uses lorenz attractors to illustrate that changing the boundary conditions gives a predictable outcome even though the individual trajectories for any set of initial conditions are unpredictable.:

    So predicting individual cyclones is very difficult because we don’t have accurate measurement of the initial conditions. Changing the boundary conditions however can result in predictable outcomes.

  12. > One cannot accurately predict individual cyclones ( beyond a few days ), why would one believe predictions about cyclone frequency?

    Because it’s easier to estimate frequencies than to predict individual cyclones, perhaps?

    Smells like the meteorological fallacy.

  13. Willard,

    Smells like the meteorological fallacy.

    I think it’s worse than we thought. I’m smelling Uncertainty Monsters: Teh Modulz are stoopid, and can’t be relied upon for policy decisions.

    We need not defend hurricane frequency paradigms to hack that one to bits.

  14. Magma says:

    Statistically speaking it will take time for a detectable change in the number and/or intensity of rare meteorological extreme events to emerge from the background noise. I suspect the smarter (as always, a relative term) deniers know this, and that they can play that hand out for a decade or two.

    In the spirit of the Schneider ethical dilemma, I have no problem at all with a climate scientist replying to a question from the media with “Yes, greenhouse gas-driven climate change almost certainly made X event worse” if X was a type of event that basic atmospheric physics, climate models, or studies of past climate have predicted will increase or worsen with warming.

  15. lerpo,

    So predicting individual cyclones is very difficult because we don’t have accurate measurement of the initial conditions. Changing the boundary conditions however can result in predictable outcomes.

    Only if there is one and only one set of long-term statistics which result from randomly chosen initial conditions, i.e., the system is transitive. When multiple persistent states can develop from randomly chosen initial conditions, the system is intransitive, and predictions can be very difficult.

    In 1968, Lorenz proposed that it is possible the climate system is what he coined “almost intransitive”, meaning essentially that it could persist in one regime for long periods of time and step change to a different regime for no immediately apparent reason and then persist there for a long period of time.

    The appeal to the “climate is unpredictable” crowd is obvious. I see it differently; the last thing I’d want to do to a system I believed to be almost intransitive is go nudging it toward some unpredictable and/or radically different attractor.

    In 1989, Lorenz suggested the climate system is unlikely to be intransitive. That’s the prevailing assumption today, though I know of some challenges to it here and there which I think are worth considering.

    I scanned Dr. Schmidt’s excellent talk (thank you) for any discussion of the transitive/intransitive distinction, not expecting to find it, and I didn’t. It’s possible I missed it, however.

  16. lerpo says:

    Interesting! Thanks Brandon.

  17. lerpo, you’re welcome!

    Magma,

    For Atlantic hurricanes

    … the data are pretty noisy, but there is a discernable trend. I don’t know if the following calcs make sense, but here they are. For major hurricanes, it’s 0.0131 storms/year, standard deviation from the trend 1.52, that’s 116 years. Same calcs for tropical storms and hurricanes are 88 and 194 years respectively.

    I’ve never worked any of this out, difficult for me to know what to make of it.

  18. Angech says:

    The attribution question.
    The holy grail or smoking gun.
    Tamino has put up one case re extreme climate events
    Australian coral has put up another.
    ATTP a third.
    Not being able to make a formal attribution to a link between climate change and extreme events when one is trying extremely hard to do so might suggest that there is either a problem with climate change AGW occurring as expected or in extreme events deciding to respond.
    The third problem is in the probabilities and timescale but that can be developed in the discussion.
    A number of people have already claimed attributions.
    Others prefer to make predictions about future events.
    A list of 10 recent ones with the degree of confidence in each attribution would be good. Brandon Gates would you have some suitable candidates.
    As Mosher would say, tread carefully, it is a minefield mainly as the different protagonists do not seem to communicate with each other and sing from different hymn books.

  19. dikranmarsupial says:

    TE’s comment suggests that he doesn’t understand that climate models don’t work by predicting weather but by simulating weather and seeing how the statistical properties of the weather (i.e. climate) change as a result of the forcings applied in the scenario. This is a very common misunderstanding.

  20. Marco says:

    “Not being able to make a formal attribution to a link between climate change and extreme events when one is trying extremely hard to do so might suggest that there is either a problem with climate change AGW occurring as expected or in extreme events deciding to respond.”

    Nope. It is much more likely to be an issue with getting sufficient data – which is the problem with *any* type of extreme event. They’re rare to start out with…

  21. dikranmarsupial says:

    We can’t a make a formal attribution of particular incidences of cancer to smoking either, no matter how hard anyone tries, but most would agree that smoking is a cause of cancer. Causality is not that straightforward a concept to define, although we generally know it when we see it, which unfortunately gives plenty of scope to avoid admitting it when it doesn’t suit us.

  22. Actually, I did a calc a couple of years back, and there’s more statistical power focusing on trends for out-of-primary-season hurricanes in the Atlantic. I also did not restrict to land-falling, as some people. do.

  23. Angech says:

    Just 10 attributable events?
    Sigh
    Let’s start with
    “The hot water temperature that drove the devastating bleaching on the Great Barrier Reef this year was made 175 times more likely by human-caused climate change, and could be normal in just 18 years, according to preliminary findings by leading climate and coral reef scientists.” The Guardian
    Yes?
    No?
    Or 175 times more likely?
    That is precision attribution with a capital A.
    “Causality is not that straightforward a concept to define,”
    A little disagreement in the ranks.
    A little scepticism needed perhaps.
    “it seems as though the issue of trying to attribute an anthropogenic influence to an extreme weather event is controversial of a number of levels”
    I agree.

  24. angech,
    I don’t know if it’s 175 times more likely, but it doesn’t seem unreasonable. Given that coral bleaching appears to be associated with warmer waters, and given that AGW has produced warming, it would certainly appear to have made it more likely.

    I think Dikran’s point wasn’t all the controversial. All I think he was pointing out is that we can suggest that A causes B, even if we can’t formally attribute A as a cause for individual events.

    “it seems as though the issue of trying to attribute an anthropogenic influence to an extreme weather event is controversial of a number of levels”
    I agree.

    This, however, does not mean that AGW is somehow not influencing extreme weather events.

  25. Dodds
    TE:
    I can’t accurately predict a single fair coin toss, but I’m pretty sure about the distribution of 100 such tosses.

    One difference is this:
    coin flips are independent of one another
    the future states of the atmosphere are dependent on the past states.

    El Nino/La Nina fluctuations are thought to be circulation changes ( of various orientations and intensities ). There is a seasonal component but the circulation changes of ENSOs are composed of different wave patterns and frequencies of wave patterns. It does not appear to be the case that ENSO occurrence is predictable. It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.

  26. It does not appear to be the case that ENSO occurrence is predictable.

    Seems reasonable.

    It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.

    Citation needed!

  27. John Hartz says:

    The issues addressed in the following article also bear on this discussion.

    Saturday is considered by some to be the “peak” of the Atlantic hurricane season. In an average year, half of our hurricane activity will have already occurred. This means that for coastal residents, the risk of landfall feels like it’s subsiding and the tropics are becoming much less interesting. But how we define “peak” is actually pretty complicated.

    In an average season, there would be six named storms, two hurricanes, and one major hurricane — Category 3 or higher — by Sept. 9. This year, there have been eight named storms, three of which became hurricanes. Of the three hurricanes, Gaston was the only one to become Category 3 or higher on the Saffir-Simpson scale. So, by this definition, the Atlantic is already running ahead of average in 2016.

    If we consider another way to measure a season, accumulated cyclone energy, we can paint a different picture. In terms of energy, this season is just 85 percent of average.

    There is a lot of subtle meaning in the word average. Which period of record is being used, and which metric of activity is best? There are no right or wrong answers to these questions, just informed preferences.

    When is the ‘peak’ of hurricane season? It’s more complicated than you think. by Brian McNoldy, Capital Weather Gang, Washington Post, Sep 9, 2016

  28. These authors seem to think that it might be possible to project a change in the frequency of ENSO events.

    The increased frequency arises from a projected surface warming over the eastern equatorial Pacific that occurs faster than in the surrounding ocean waters13, 14, facilitating more occurrences of atmospheric convection in the eastern equatorial region.

  29. Angech says:

    175 times more likely is a very big figure ATTP.
    When you are trying to determine attribution and point out that it can be very tricky to see, due to the time scales we have to operate in, a doubling would be a very big change.
    When we talk of a 2 to the 7th power multiplication occurring in 18 years we are talking about impossibilities, of wild eyed imagining.
    the equivalent scientifically of that America being frozen overnight film.
    Sorry.
    It does seem unreasonable in the light of day.
    Even more so the specific number at such a high figure.
    The reasoning behind such calculations would be amazing to see.

  30. Also “I can’t accurately predict a single fair coin toss, but I’m pretty sure about the distribution of 100 such tosses.

    One difference is this:
    coin flips are independent of one another
    the future states of the atmosphere are dependent on the past states.”

    Another difference – we can arrange to tightly control the coin flip environment to remove all variables.

    The climate system barely so. There may be unknown interactions, insufficient data points and length of time recording.

    Trying to predicting random weather/climate events when we do not yet understand the nature of the randomness involved is premature to say the least.

  31. angech,
    It sounds like you’re presenting an argument from incredulity. As I understand it, the first recorded global bleaching event was in 1998. There have been two others since then.

    When you are trying to determine attribution and point out that it can be very tricky to see, due to the time scales we have to operate in, a doubling would be a very big change.

    If we’re talking about events that are on the tail of a distribution, I don’t think it is quite this simple. Changing the distribution slightly, can have a big influence on the probability of an extreme event, even if it only changes the likelihood of common events by a small amount.

    When we talk of a 2 to the 7th power multiplication occurring in 18 years we are talking about impossibilities, of wild eyed imagining.

    Again, this seems to simply be an argument from incredulity. If there were no global bleaching events prior ro 1998, and 3 since then, it certainly sounds like the change in likelihood has been quite substantial.

  32. Steve,

    Trying to predicting random weather/climate events when we do not yet understand the nature of the randomness involved is premature to say the least.

    Noone is claiming that we can predict climate/weather events. The suggestion is that we have a sufficient understanding of the underlying processes to make estimates of how the likelihood of events will change as the climate warms. We can’t know if these estimates will be correct, but suggesting that it’s too complex to do so is not a wonderfully compelling argument.

  33. Angech says:

    From the article ” in the usually cold and dry equatorial eastern Pacific”
    I doubt equatorial sea level Atlantic or Pacific could ever be described as usually cold and dry.

    More to the point
    “Here we present climate modelling evidence for a doubling in the occurrences in the future in response to greenhouse warming.”
    Note that they are trying to predict a doubling in severe ENSO frequency mentioning only two events that qualified in 1982 and 1997 .
    Presumably the article gives a time frame, I would guess 60-100 years to see a doubling presumably before the end of this century
    Purchase article full text and PD. £22 too rich for me.
    Compare and contrast this reasonable guesstimate and time frame to the Guardian article which implies that El Ninos, the presumptive cause of the coral bleaching, will become an annual occurrence in 18 years time.
    Which is quite ok if you believe in a 2 to the seventh or 175 times rate of occurrence of El Niño events.
    Happy to call this attribution number 2.
    But remember what the girl on the video said about attribution.

  34. TE,
    you seem to be confusing weather and climate. What we can test is whether a model is able to reproduce the observed cyclone frequency. The same is true for El Nino/La Nina (in fact it’s well documented in the literature how each CMIP5 model performs wrt ENSO behaviour). As an aside, the WWA project uses an atmosphere-only model setup, i.e. the strength and phasing of ENSO is prescribed and hence “correct” by definition. Back to the cyclone example. Either the distribution of observed and modelled frequency is okay – then we can (confidently) use the model for attribution – or it is biased one way or another, in which case it’s getting a bit tricky. While bias correction for temperature and precipitation is fairly straight forward, for cyclones we have to be more creative. For example, we can use analogue methods to extract patterns that resemble the extreme event we are interested in. If we find more/less cyclones in the ensemble of the actual world as opposed to the counterfactual world, we have at least some indication as to what might be the attributable component. However, we haven’t done cyclone attribution yet, as we are limited to the regional domain (the coarse global model can’t produce proper cyclones) which doesn’t span an entire ocean basin at the moment.

    Either way, even if atmospheric circulation isn’t always reproduced perfectly, we can say something about thermodynamic changes. We do analogues now more regularly for temperature and precip extremes as well, enabling us to also say something about the dynamic/circulation contribution to a given extreme event. If you are interested, there are two papers on the subject of model performance/model validation available: Massey et al 2014 and Mitchell et al 2016. Two more paper in preparation (I’m gonna write one of them).

    John,
    we (WWA) always only ask: “How has the risk of occurrence of a given extreme event changed as a result of climate change?”. We can’t answer the question whether or not climate changed caused a particular extreme event. The question would be silly indeed.

  35. Willard says:

    > I’m smelling Uncertainty Monsters: Teh Modulz are stoopid, and can’t be relied upon for policy decisions.

    Your smell anticipates, BG. Mine was used to detect. I did not anticipate this:

    One difference is this: coin flips are independent of one another the future states of the atmosphere are dependent on the past states.

    I cannot say when I’ll die, but I can foretell I will. From the end to now all my days depend upon one another. TE’s argument looks like special pleading to me: TE’s difference makes no obvious difference.

    The following move was bound to happen:

    Just 10 attributable events? Sigh Let’s start with […]

    I always anticipate peddling when angech comes to play. I can’t predict when he’ll peddle, but I am quite confident that he will. Sighing would be futile.

    Senior’s position would be more palatable to me if he did not indulge so often into the meteorological fallacy. Maybe it’s a background thing.

    Have I just made a saucer pass so that Senior shoots in the gap?

  36. angech,

    Which is quite ok if you believe in a 2 to the seventh or 175 times rate of occurrence of El Niño events.

    It was bleaching was 175 times more likely, not ENSO events occuring 175 times more often. Also, I didn’t say I believed it; I simply see no reason to doubt it just because it sounds like a big number.

    Compare and contrast this reasonable guesstimate and time frame to the Guardian article which implies that El Ninos, the presumptive cause of the coral bleaching, will become an annual occurrence in 18 years time.

    Did it?

    But remember what the girl on the video said about attribution.

    Girl?

  37. Angech says:

    The first global bleaching declared by NOAA was in 1988.
    That is different to recorded.
    Coral bleaching has occurred for thousands of years in corals in shallow hot waters, not all coral fits this bill, some live in quite deep waters and cannot get bleached. Studies of coral reefs show these past bleaching events globally but strangely no signs of NOAA exist before 1900.
    A massive new reef off South America that no one knew of has just been discovered and a smaller one off Tasmania which will never get bleached in the next 500 years.

  38. This site also seems to suggest that we’ve just had our third recorded global coral bleaching event. Maybe you need to provide some citations? Also – in case it wasn’t obvious – if the first recorded global coral bleaching event was in 1998, and we’ve had two more since then, that would seem to suggest quite a large change in the likelihood of global bleaching events.

  39. Karsten,
    Thanks. I had forgotten that you’re in the same Department as Friederike Otto.

  40. Willard says:

    > The first global bleaching declared by NOAA was in 1988. That is different to recorded.

    How do we record bleaching?

  41. John Hartz says:

    Lest we loose sight of the fact tht manmade climate change impacts more than just the atmosphere…

    Global warming is making the oceans sicker than ever before, spreading disease among animals and humans and threatening food security across the planet, a major scientific report said.

    The findings, based on peer-reviewed research, were compiled by 80 scientists from 12 countries, experts said at the International Union for Conservation of Nature (IUCN) World Conservation Congress in Hawaii.

    “We all know that the oceans sustain this planet. We all know that the oceans provide every second breath we take,” IUCN director general Inger Andersen said at the meeting, which has drawn 9,000 leaders and environmentalists to Honolulu.

    “And yet we are making the oceans sick.”

    The report, Explaining Ocean Warming, is the “most comprehensive, most systematic study we have ever undertaken on the consequence of this warming on the ocean”, co-lead author Dan Laffoley said.

    Global warming making oceans sick, spreading disease in humans and animals, scientists warn, ABC News (Australia) , Sep 12, 2016

  42. BBD says:

    It does look as though the recent flooding events in Cumbria (UK) are extremely unusual on a multicentennial timescale. Something hotly denied by contrarians but apparently nevertheless true.

  43. BBD says:

    angech

    Coral bleaching has occurred for thousands of years in corals in shallow hot waters, not all coral fits this bill, some live in quite deep waters and cannot get bleached.

    But it can and will be destroyed by ocean pH change. See eg. Veron et al. (2009):

    Temperature-induced mass coral bleaching causing mortality on a wide geographic scale started when atmospheric CO2 levels exceeded ~320 ppm. When CO2 levels reached ~340 ppm, sporadic but highly destructive mass bleaching occurred in most reefs world-wide, often associated with El Niño events. Recovery was dependent on the vulnerability of individual reef areas and on the reef’s previous history and resilience. At today’s level of ~387 ppm, allowing a lag-time of 10 years for sea temperatures to respond, most reefs world-wide are committed to an irreversible decline. Mass bleaching will in future become annual, departing from the 4 to 7 years return-time of El Niño events. Bleaching will be exacerbated by the effects of degraded water-quality and increased severe weather events. In addition, the progressive onset of ocean acidification will cause reduction of coral growth and retardation of the growth of high magnesium calcite-secreting coralline algae. If CO2 levels are allowed to reach 450 ppm (due to occur by 2030–2040 at the current rates), reefs will be in rapid and terminal decline world-wide from multiple synergies arising from mass bleaching, ocean acidification, and other environmental impacts. Damage to shallow reef communities will become extensive with consequent reduction of biodiversity followed by extinctions. Reefs will cease to be large-scale nursery grounds for fish and will cease to have most of their current value to humanity. There will be knock-on effects to ecosystems associated with reefs, and to other pelagic and benthic ecosystems. Should CO2 levels reach 600 ppm reefs will be eroding geological structures with populations of surviving biota restricted to refuges. Domino effects will follow, affecting many other marine ecosystems. This is likely to have been the path of great mass extinctions of the past, adding to the case that anthropogenic CO2 emissions could trigger the Earth’s sixth mass extinction.

  44. In addition to predictability of atmospheric events, notice should be paid to events which, when solidly expected and predictable, are suddenly disrupted for no explicable reason. For example, Osprey, Butchart, Knight, Scaife, Hamilton, Anstey, Schenzinger, Zhang, “An unexpected disruption of the atmospheric quasi-biennial oscillation“, Science.

    I mention, again , Norton’s Causation as folk science. See also this and spontaneous symmetry breaking.

  45. izen says:

    @-“the future states of the atmosphere are dependent on the past states.”

    They are contingent on the historical thermodynamics. That strongly constrains the range of variation possible, for instance it prevents the states of the climate from being a random walk.

    @-” It does not appear to be the case that ENSO occurrence is predictable. It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.”

    Because the the future states the ENSO cycle are dependent on the past states they are much more predictable than a coin toss.
    Much more likely to get ten heads in a row than ten El Nino’s.

  46. John Hartz says:

    Is it fair to say that the accelerating greenhouse effect is creating a “new normal” for the Eath’s climate system on a daily basis?

    If so, our ability to forecast future extreme weather events based on historical trends seems to be rather complex.

    An example of a recent severe weather event that appears to be unprecedented is the focus of…

    We Haven’t Seen Many Storms Like Hermine by Eric Holthaus, FiveThirtyEight, Sep 3, 2016:

    The tease-line for Holthaus’s article is: Which is what makes its future harder to predict.

  47. Trying to predicting random weather/climate events when we do not yet understand the nature of the randomness involved is premature to say the least.

    Well, it depends on “how crazy” you expect the climate system to be in the absence of major forcing. I think it’s entirely reasonable to assume wide sense stationarity, equivalent to the hypothesis of stationarity with respect to the second moment as in kriging, where (a) the dependence of two random variables in time only depends upon their separation in time, and (b) that dependence decreases, typically exponentially, with separation in time. Accordingly, there’s only so long that the volcanic explosion at Pinatubo can affect climate.

    I think that at some level there’s an assumption of additivity in some of the modeling, such that the effects of radiative forcing from an additional dollop of greenhouse gases can be added atop of whatever the baseline effect was before the dollop. From a meteorological perspective, sure, the forcing is gradual compared to seasonal and some local effects, and internal variability (something which I still don’t quite understand what it is). However, radiative forcing from greenhouse gas emissions on a geological scale is a huge impulse hit to the climate system, and it’s not clear we have a good understanding of what non-linear effects from such an impulse might be based upon observations. Because it’s been so long since the last time anything like such an impulse has affected Earth’s climate, all we’ve got are (a) paleoclimate reconstructions, and (b) ab initio climate models, which are not widely used or available, unlike the otherwise excellent GCMs which, from what I understand, have recent calibrations built into them.

  48. TE,
    you seem to be confusing weather and climate. What we can test is whether a model is able to reproduce the observed cyclone frequency. The same is true for El Nino/La Nina (in fact it’s well documented in the literature how each CMIP5 model performs wrt ENSO behaviour).

    Karsten,

    Below is the IPCC chart demonstrating blocking ( one aspect circulation ) that CMIP5 models hindcast for 1961 through 1990 (AR5, WG. Clearly, all the models are in error, most, grossly so. The models cannot predict the general circulation of climate ( thirty years ). This is not surprising, the solutions to the equations of motion are chaotic and subtle differences between models mean the solutions necessarily diverge, just as Lorenz wrote. That’s true in the short term ( weather – 1-5 days of a synpotic wave passing ) and it’s true for climate ( the thirty years of wave statistics of synoptic waves from the CMIP5 runs ).

  49. TE,
    You should probably try reading Karsten’s comment again.

  50. TE,
    The only place I can find that figure is in comments made by you. What is the original source?

  51. John Hartz says:

    TE: What do you hope to accomplish by posting comments on this website?

  52. Okay, I found it. Box 14.2 in WGI AR5 (i.e., Chapter 14 if WG1). It ends with

    In summary, the increased ability in simulating blocking in some models indicate that there is medium confidence that the frequency of NH and SH blocking will not increase, while trends in blocking intensity and persistence remain uncertain. The implications for blocking related regional changes in North America, Europe and Mediterranean and Central and North Asia are therefore also uncertain.

  53. Willard,

    TE’s difference makes no obvious difference.

    It *could* make a difference.

    Start here: There is a seasonal component but the circulation changes of ENSOs are composed of different wave patterns and frequencies of wave patterns.

    Now the conclusion: It does not appear to be the case that ENSO occurrence is predictable. It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.

    So many butterflies. It’s fractals all the way down. I dare you to prove him wrong.

    Have I just made a saucer pass so that Senior shoots in the gap?

    I used to play hoops against a guy who called my deke because of the lopsided distribution. It was annoyingly effective.

    Could TE hit a stationary puck into an open net from inside the crease?

  54. TE,
    please don’t teach me on blocking frequency in models. I do generally appreciate the input of others (experts and non-experts alike) if their contribution appears to be useful and well thought-out. Your comments fail to impress. Engineering and atmospheric physics isn’t the same. The sooner you realise, the better your chances to understand how the atmosphere actually works.

    ATTP,
    I’m working with Fredi (and Peter + Myles) 😉 Fredi, Peter and I are the Oxford “branch” of WWA.

  55. Willard,

    Speaking of Senior, I just serendipitously stumbled on Pielke and Zeng (1993) while looking for something else:

    Obviously, the Lorenz model itself is too simple to be used directly for the study of long-term variability of the earth’s climate system. However, the results based on the Lorenz model may be quantitatively correct for different nonlinear systems (including the climate system), and our results lead to our speculation that the seasonal cycle or other prominent periodic variations (such as the 22-year sunspot cycle) may lead to long-term variability due to the internal nonlinear dynamics (especially nonlinear coupling processes) in the earth’s climate system. A positive (but indirect) support of this speculation is found in Winograd et al. (1992), where, based on new observational evidence, they found that ice ages may not be caused predominantly by the earth’s orbital cycle (i.e., the Milankovitch theory) but by the natural internal variablities of the climate system. Therefore, with respect to the study of the cause for the natural variability of the climate system, the main value of our study is not the quantitative results themselves but its use in indicating a direction for future research when a real climate model and more powerful computers are available.

    I’m not sure this constitutes an example of weather fallacies, but I can appreciate the due diligence of leaving no stone unturned when hunting the non-obvious. I wonder whether Senior thinks real climate models exist at present.

    Winograd (1992) reads as advertised. Moseley et al. (2016) disagrees, based on even newer new observational evidence.

  56. ertbops says:

    QBO and Richard Lindzen

  57. And a note … the latest work quoted by the IPCC WG1 there on blocking is from February 2013. New work on blocking, models, and all that are reported in:

    http://journals.ametsoc.org/doi/full/10.1175/JCLI-D-14-00822.1
    http://journals.ametsoc.org/doi/full/10.1175/JCLI-D-15-0493.1
    http://journals.ametsoc.org/doi/full/10.1175/JCLI-D-12-00466.1
    http://journals.ametsoc.org/doi/full/10.1175/MWR-D-12-00134.1
    http://journals.ametsoc.org/doi/abs/10.1175/JCLI-D-16-0242.1
    https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20110008410.pdf
    http://www.sparc-climate.org/fileadmin/customer/5_Meetings/2016_PDF/blocking_workshop_abstracts.pdf

    As of 2015, the field still appears to be in flux, per (Overland, Francis, Hall, Hanna, Kim, Vihma):

    This paper combines recent literature with new analyses to conclude that evidence for linkages between Arctic amplification and midlatitude weather patterns may be emerging, but that mechanisms are likely to be regional, episodic, and seasonally and interannually varying, making systematic detection a major challenge. Current synthesis suggests that there will be no net midlatitude cooling, only a potential for severe events (Barnes and Screen 2015). In north-central Asia, changes in baroclinicity due to the loss of sea ice in the BK forces a sustained regional response, but the impact farther east is the result of a propagating wave train, involving time scales from days to months. The ridge–trough system that persisted over North America throughout many of the early winters since 2007 is a climatologically favored pattern that can be amplified by internal dynamic processes, local surface temperature anomalies, low-level temperature advection, and remote influences. Analyses based on decadal temperature trends or zonal-mean dynamics will not detect such events. Models are important tools, but are currently providing conflicting results.

    BK and northeastern North America are of particular interest for Arctic–midlatitude linkages impacting skill in extended-range weather forecasts (Jung et al. 2014). New metrics and approaches are needed to identify regionally and seasonally varying mechanisms linking substantial thermodynamic changes in the Arctic to shifts in the chaotic atmospheric circulation of the Northern Hemisphere. Additional challenges emerge as other changes in the climate system occur simultaneously; the interactions of these effects are complicated and may be different from anything humans have witnessed before.

    Given the complexity of the physics and the multiple approaches taken to study these phenomena, it is not surprising to find diversity, disagreements, and fragmentation of the scientific community, as research in this field gains attention and momentum. A warmer climate is predicted for the future (Wallace et al. 2014; Screen 2014), yet warm Arctic–cold continent linkages of a regional and episodic nature have a potential to amplify extreme midlatitude weather events, and therefore affect billions of people, in coming decades.

    In an attempt to summarize what I have read, results vary on how well GCMs predict blocking (not great, but …), RCMs predict blocking (better), and differences of opinion on what indices are better for it, Pelly and Hoskins vs Tibaldi and Molteni (used for instance by Whan, Zwiers, and Sillman) vs ?. The exception seems to be Matsueda’s 2016 conference report which asserts in part “Generally, models simulate the frequency of blocking in both NH and SH well, even at a lead time of 15 days, except some extreme blocking events. Verifications based on Brier Skill Scores (BSS) for probabilistic blocking forecasts show that the probabilistic blocking forecasts over NH are better than those over SH, although there are some year‐to‐year variabilities of the forecast skills.”

    Also, there’s an interesting split in the meteorological literature I’ve seen. (I cannot speak much to GRL and the like: I don’t subscribe.) There seems to be a set of scholars which pursue atmosphere-only explanations, a set who pursue strong ocean-and-atmosphere explanations, and a set who use models which, presumably, must incorporate some ocean links and thus include them indirectly. I don’t know if that’s due to personal preference of the scholars or some judgment, based upon rough physics on the part of some, that oceans are a side show and, to first order, decoupled.

  58. Turbulent Eddie,

    The models cannot predict the general circulation of climate ( thirty years ). This is not surprising, the solutions to the equations of motion are chaotic and subtle differences between models mean the solutions necessarily diverge, just as Lorenz wrote.

    It might help if you directly quoted Lorenz on divergence. In the meantime, here’s the caption:

    Box 14.2, Figure 1 | Annual mean blocking frequency in the NH (expressed in % of time, that is, 1% means about 4 days per year) as simulated by a set of CMIP5 models (colour lines) for the 1961–1990 period of one run of the historical simulation. Grey shading shows the mean model result plus/minus one standard deviation. Black thick line indicates the observed blocking frequency derived from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis. Only CMIP5 models with available 500 hPa geopotential height daily data at http://pcmdi3.llnl.gov/esgcet/home.htm have been used. Blocking is defined as in Barriopedro et al. (2006), which uses a modified version of the(Tibaldi and Molteni, 1990) index. Daily data was interpolated to a common regular 2.5° × 2.5° longitude–latitude grid before detecting blocking.

    While you’re skating around talking about divergence, the IPCC’s +/- 1-sigma ensemble goalie is waiting for you to notice that the puck is sitting on the ice doing nothing.

  59. hypergeometric,

    I think it’s entirely reasonable to assume wide sense stationarity, equivalent to the hypothesis of stationarity with respect to the second moment as in kriging, where (a) the dependence of two random variables in time only depends upon their separation in time, and (b) that dependence decreases, typically exponentially, with separation in time.

    Thank you for that concise description. A million other arguments suddenly began to make a lot more sense to me.

  60. Just a few more thoughts and I’ll be gone.

    1. There is a distinction between the predictability of global average temperature and the predictability many other phenomena, including precipitation, which depend upon motion. Temperature depends somewhat on motion ( cold fronts, precipitation, and the like ) but in the global mean, radiative forcing at the tropopause from 2xCO2 is only slightly affected by the variations of weather below. Consequently, the solutions to equations governing thermal content of the global troposphere are not unstable and are mostly predictable. Precipitation, on the other hand, is discrete and almost entirely dependent on motions which are governed by equations the solutions of which are unstable and chaotic.

    2. That doesn’t mean there is no predictive capacity. Thermal gradients increase from hemispheric summer to hemispheric winter and the jet stream intensity and location varies with the change in gradient. Similarly, the mountain ranges and ocean basins channel the flow imposing statistically preferred locations of troughs and ridges. Given that orbits and geography will not likely change significantly over the next century, we can predict that neither will the occurrence, relative intensity, or seasonal location ( within the range of chaotic fluctuations ) of the jet streams change.

    3. The first principles of meridional gradient determining the jet stream intensity and location are valid. Should AGW change meridional temperature gradients, that would induce change in the intensity and average latitude band of storm tracks, though within any intensity there still remains an infinite number of equally valid wave patterns ( of both short and long duration ).

    4. Jenifer Francis has published a hypothesis that AGW will lead toward reduced thermal gradients because of Arctic Amplification that will significantly reduce the jet stream winds and so stagnate general circulation patterns. James Hansen has published a hypothesis that Arctic Amplification will induce polar ice melt, which will cool high latitude waters, and increase thermal gradients leading to a super storm inducing increase in jet stream strength. Both cannot be correct. Both can be incorrect if the changes occur but are not significant.

    5. Both Francis and Hansen invoke Arctic Amplification as a causal factor, but AA is greatest near the surface ( latent heat of freezing ) and diminishes with height. Further, there is some question as to the relevant latitudes to consider for thermal gradients. Rossby considered surface pressure gradients ( it was 1939, after all, before even RadioSondes were available ) from 35 to 55 degrees. This is well south of AA. Should he Hot Spot occur as modeled, it would definitely increase the thermal gradient, including the 35 to 55 range. Radiosonde observations don’t indicate much change at 300mb. This will be interesting to watch, but to date, the models have failed to validate.

  61. TE,
    It’s almost as if you think that either we can make accurate (almost precise?) predictions or we can do nothing. Is that about it?

  62. Turbulent Eddie,

    There is a distinction between the predictability of global average temperature and the predictability many other phenomena, including precipitation, which depend upon motion.

    And there’s a distinction between divergence of a weather forecast due to poorly constrained initial conditions, and climatology errors due to poorly constrained boundary values. Chaotic behaviour does not in and of itself entail long-term unpredictability of the attractor. Think about that. It’s definitional.

  63. angech says:

    K.a.r.S.t.e.N says:
    “TE, Engineering and atmospheric physics isn’t the same.”
    Stating the obvious as we all know Engineers don’t know any physics’.
    Could not resist. I had a lot of friends who are engineers.

  64. angech says:

    “Willard, I always anticipate peddling when angech comes to play. I can’t predict when he’ll peddle.”
    Just paddling at the moment.
    I believe you and ATTP are conversant with probabilities and understand the sheer scale of overreach in the 175 times figure.
    Choosing to defend it is understandable it is like doctors and lawyers looking after their colleagues backs.
    The article stated that El Nino’s, the putative cause of coral bleaching events, worsened [attributable] by climate change due to CO2 increase are going to be occurring every year in 18 years time.
    “izen says: [Explaining the problem in reply to ENSO predictability]
    September 11, 2016 at 6:31 pm
    @-” It does not appear to be the case that ENSO occurrence is predictable. It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.”
    Because the the future states the ENSO cycle are dependent on the past states they are much more predictable than a coin toss.
    Much more likely to get ten heads in a row than ten El Nino’s.”

  65. angech wrote “Causality is not that straightforward a concept to define,”
    A little disagreement in the ranks.”

    If you think I am going to rise to the bait after the display of disingenuous behaviour on the other thread (the most charitable interpretation of which is that you aren’t even honest with yourself), you are mistaken. Especially if you are going to employ a ruse as transparent as “A little disagreement in the ranks”.

  66. TE wrote “One difference is this: coin flips are independent of one another the future states of the atmosphere are dependent on the past states.”

    The difference is irrelevant to the point Andrew was making, it just means

    “I can’t accurately predict a single fair coin toss, but I’m pretty sure about the distribution of 100 such tosses.”

    becomes

    “I can’t accurately predict a single fair coin toss, but I’m pretty sure about the conditional distribution of 100 such tosses.”

    The future states of speech signals are dependent on past states, but that doesn’t mean statistical approaches to speech recognition don’t work (they certainly do).

  67. steverichards1984 says: “Trying to predicting random weather/climate events when we do not yet understand the nature of the randomness involved is premature to say the least.

    Arguably there are no truly random phenomena (except perhaps at the quantum level, and perhaps not even there), and what we call randomness is just a lack of knowledge, so this comment actually doesn’t mean anything as it is the physics that we don’t fully understand, but probability is just the framework for dealing with that lack of knowledge. Which is why the best available approach is to combine our best knowledge of the physics, with the best data available use probability for the analysis, which is (oddly enough) just what the climatologists do.

  68. verytallguy says:

    Well, I agree with TE.

    Predictability of the impacts of climate change is poor due to the lack of reliable regional scale projections [this is in the AR5 somewhere I think, but I’m too lazy to reference it right now].

    As a result, money spent on adaptation is likely to be wasted – we don’t know what to adapt to!

    So TE’s conclusion that there is an urgent need for mitigation is one that I’d strongly support.

    (TE seems to have accidentally missed off the conclusion, but we can be sure he would follow the logic of his argument 😉 )

  69. angech,

    Choosing to defend it is understandable it is like doctors and lawyers looking after their colleagues backs.

    Okay, you appear to be engaging in bad faith, which I had thought you were less likely to do than others. I’m not defending it; I’m simply pointing out that assuming it is wrong because it is a big number is a poor argument against it.

    vtg,
    Maybe I’ve misunderstood TE’s point 😉

  70. TE wrote “It also does not appear to be the case that ENSO frequency over decades or centuries is predictable.”

    ATTP (quite reasonably) responded: “Citation needed!”

    TE later said “Just a few more thoughts and I’ll be gone.”

    I don’t see where TE responded to ATTP’s request (or provided evidence that the statistical properties of ENSO were sufficiently unpredictable to have an impact on centennial scale climate impacts). It would be good is TE could respond to this, or at least admit that he has no such evidence.

  71. JCH says:

    ENSO’s past is relevant… maybe. ACO2 is at 400ppm. There may be no relevant past for this reality. We just experienced a warmest year when ENSO neutral conditions existed in all 12 months: no La Niña and no El Niño. Next up, a warmest year when some of the 12 months have La Niña conditions and the rest of the months have ENSO neutral conditions. It’s the only ENSO virgin left. Or, why ENSO can’t warm or cool our planet; that’s mostly sunlight and CO2’s thing.

  72. > I believe you and ATTP are conversant with probabilities and understand the sheer scale of overreach in the […]

    I asked you a question, dear angech: coral bleaching – how do we “record” it?

    I believe someone who is conversant with the distinction between declaring and recording bleaching knows how we record it.

    I also believe that someone who answers questions has more leverage to get room service.

  73. lerpo says:

    I think a large part of the disagreement here stems from the fact that angech has misunderstood the article that she is quoting. She says:

    “The article stated that El Nino’s, the putative cause of coral bleaching events, worsened [attributable] by climate change due to CO2 increase are going to be occurring every year in 18 years time.

    But that’s not what the article says:

    “They found the record warm temperatures in the Coral Sea that drove the bleaching this year were driven by a combination of 1C of warming since 1900 caused by greenhouse gas emissions, and about a 0.5C jump in temperature driven by natural variability.

    Using climate model simulations, the team found such an event was incredibly unusual in models that did not include the effects of greenhouse gasses. But when those effects were put back in, temperatures such as those seen this March were 175 times more likely.”

    It says nothing about whether or how El Nino’s have changed or will change due to global warming.

    It also does not seem the least bit surprising that a rise of 1.5C above average would be extremely unlikely without the 1C boost from global warming. Natural variability causing a 0.5C rise is already uncommon. A rise of 1.5C during an El Nino seems (to me) unthinkable, but this article indicates that it is only 175 times less likely.

    Regarding “going to be occurring every year in 18 years time,” she has also misunderstood:

    “In a world without humans, it’s not quite impossible that you’d get March sea surface temperatures as warm as this year but it’s extremely unlikely,” King told Guardian Australia. “Whereas in the current climate it’s unusual but not exceptional. By the mid 2030s it will be average. And beyond that it will be cooler than normal if it was as warm as this year.”

  74. John Hartz says:

    As documented in tne two articles below, meterologists are not of one mind when it comes to forecasting something as seemingly straightforward as the duration of the current La Nina event.

    Japan sees 70 pct chance of La Nina continuing through winter by Osamu Tsukimori, Reuters, Sep 9, 2016

    La Niña fizzles, making record warm global temperatures more likely by Andrew Freedman, Mashable, Sep 8, 2016

  75. anoilman says:

    John,

    Both your articles show a spread of like 10% probability between 2 models…. There are lots of models.

    I think that predicting El Nino/La Nina isn’t too precise. Most decisions are based on the average of what all the models are showing, which right now, is no more La Nina. (Its in there somewhere… )
    http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/enso.shtml

  76. Eli Rabett says:

    VTG one adapts to the current situation, which is why mitigation is needed given a trend

  77. verytallguy says:

    VTG one adapts to the current situation,

    Linky?

    I thought it referred to both:

    Adaptation measures may be planned in advance or put in place spontaneously in response to a local pressure.

    https://www.theguardian.com/environment/2012/feb/27/climate-change-adaptation

    But often seems to be discussed in terms of infrastructSture design eg flood defences should be designed for what is likely in 30 years time, rather than the current climate.

  78. angech says:

    ATTP Sorry to give that impression
    dikranmarsupial said
    angech wrote “Causality is not that straightforward a concept to define,” A little disagreement in the ranks.”
    If you think I am going to rise to the bait after the display of disingenuous behaviour on the other thread (the most charitable interpretation of which is that you aren’t even honest with yourself), you are mistaken. Especially if you are going to employ a ruse as transparent as “A little disagreement in the ranks”.
    Dikran, I remember playing Bridge in Melbourne 30 years ago, novices [still] from the country. In a prolonged bidding scenario by the other side I put my 4th pass on the next row of lines down instead of the fourth square on the top line.
    ” Director! “This person is giving information to his partner.”
    I had no clue.
    A similar episode in high school 13 year old when the teacher asked “”Anyone else have anything to say?
    A visit to the headmaster for a caning left Mr Disingenuous a little wiser to the ways of the world.
    I apologize, did not realize I was using a ruse I am truly naive at times.

  79. John Hartz says:

    anoilman: I was illustrating how complex the weather systems governing the ENSO cycle is.

  80. angech says:

    The hot water temperature that drove the devastating bleaching on the Great Barrier Reef this year was made 175 times more likely by human-caused climate change, and could be normal in just 18 years.
    The scientists said they took the unusual step of releasing the work prior to peer-review, because the methods used to reach the findings are now accepted in the climate science community and the alarming results needed to be released as quickly as possible”.
    Note –“We are confident in the results because these kind of attribution studies are well established”
    But they also said
    “Coral reefs have had a crucial role in shaping the ecosys-tems that have dominated tropical oceans over the past 200 million years.coral reefs are critical to the survival of tropical marine ecosystems ”
    Look at the big picture. Coral reefs have existed for 200 million years according to this article and one could imagine have existed for a billion years The earliest life forms, stromatolites are 3.5 billion years old and similar life style to corals.
    The sea gets hotter in Queensland it becomes more tropical in Sydney, the corals move down the coast. When they spawn the coral can travel 50 to 100 kilometers in days .
    Getting hotter in one locale does not stop them moving and adapting. Never has.
    The current reef has only been there 10,000 years doe to sea level rise.
    Coral reefs are a 90 billion dollar source of income in the Caribbean, 1 Billion in Australia, tourism alone, they are an essential source of protein for many millions of the worlds poorer societies.
    This article focuses on damage only at one spot and not on benefits and natural mitigation that also occur elsewhere.

  81. I’ve stayed out of this piece of the debate, but the only really meaningful way that the 175x could be treated seriously is if someone calculated an odds ratio against the counterfactual of no-warming. This is a point estimate, so is contingent upon setting and dataset. This would be a logistic regression of the response “bleaching” in a set of experimental conditions where climate change is present versus the incidence of “bleaching” in a set of conditions where climate change is not present. And there would be some kind of highest probability density interval on the log odds to convey uncertainty.
    I clearly cannot address what the results of the study say about that, but that’s how it would be set up.

    Odds ratios can be calculated in other ways as well, not simply logistic regression, but it is the most conceptually direct.

  82. > Look at the big picture.

    From “Let’s start with” an irrelevant Grauniad article to burdening otters with defending that irrelevant article, back to the big picture.

    Sigh.

    The big picture is the attribution question, angech. Peddling’s over.

  83. hypergeometric,

    See what you can make of:

    King, A. D., M. T. Black, S.-K. Min, E. M. Fischer, D. M. Mitchell, L. J. Harrington, and S. E. Perkins-Kirkpatrick: Emergence of heat extremes attributable to anthropogenic influences. Geophys. Res. Lett., doi: 10.1002/2015GL067448.

    … which is by the same lead author in question. Under the submitted bucket we find:

    King, A. D.: Attributing changing rates of record-breaking temperatures to anthropogenic influences. Environ. Res. Lett., submitted.

    … which may be the one for angech to watch for given his level of interest in its validity.

  84. I’ll have a look, but it won’t be done tomorrow.

  85. Eli Rabett says:

    VTG, it’s Eli’s four laws of climate change

    1. Adaptation responds to current losses.
    2. Mitigation responds to future losses
    3. Adaptation plus future costs is more expensive than mitigation,
    4. Adaptation without mitigation drives procrastination penalties to infinity.

    The nature of the thing is that you have to have a very good idea of what losses are to adapt to them, and anticipating losses is at best over a short time horizon.

  86. I read it and I’m satisfied that they’re not pulling numbers out of a hat, so I consider it more your level of interest in the topic as the thing which drives how deeply you want to look into it. I could ding the Graun for not giving more context for the how the probability was calculated, but I think the more salient question is how the GBR will fare in 30 years. The Gruan cited Ove Hoegh-Guldberg, who is, like, *the* corals guy, and his work is all over the place for interested parties to read about.

  87. Dangit, Eli sneaked in there out of turn. My previous comment was directed at hypergeometric.

  88. @Eli Rabbet, “Eli’s Four Laws” are an excellent summary of the situation. Eli, did you talk about those at greater length some place on Rabbet Run?

  89. angech wrote “I apologize, did not realize I was using a ruse I am truly naive at times.”

    an apology for your disingenuous behaviour on the other thread would be more appropriate. It is a shame that attempts to discuss climate inevitably result in this kind of bullshit.

  90. Marco says:

    Angech, you are aware that the coral reefs of today are not quite the coral reefs of 200 million years ago, right?

    And you are also aware that it takes a LOT of time, as in many hundreds to thousands of years, before a coral reef is built up? In the meantime, there’s lots of trouble.

    A very recent paper discussing on of the troubles of coral reefs: http://advances.sciencemag.org/content/advances/2/4/e1500850.full.pdf

  91. angech says:

    hypergeometric says:
    “I’ve stayed out of this piece of the debate, but the only really meaningful way that the 175x could be treated seriously is if someone calculated an odds ratio against the counterfactual of no-warming.”
    Thanks for this considered input
    willard (@nevaudit) says:From “Let’s start with” an irrelevant Grauniad article”
    Irreverent, irrelevant [thanks] and as you know extremely improbable

    ” to burdening otters with defending that irrelevant article”
    I am quite happy for you to give an opinion on the likelihood of an 175 x more likely attribution in 14 years for this particularly or any other bit of climate attribution?

    “The big picture is the attribution question, angech”
    which is not helped by “irrelevant” articles peddling such a high attribution possibility when all other scenarios use much longer times and much less attribution.
    “King, A. D.: Attributing changing rates of record-breaking temperatures to anthropogenic influences. Environ. Res. Lett., submitted.… which may be the one for angech to watch for given his level of interest in its validity.”
    Thanks Brandon. Let us put it down as number 3.
    Another 7 different ones please if anyone can help out.
    10 would be a good sample to work from

  92. angech,

    I am quite happy for you to give an opinion on the likelihood of an 175 x more likely attribution in 14 years for this particularly or any other bit of climate attribution?

    I think Willard’s point is that you introduced it, and your main argument appears to be simply that it seems too big to you. Now, however, you seem to expect others to defend it. I don’t particularly care to defend, or attack, it. If you want to attack it then ideally you should have more than simply a sense that the number is too big.

  93. Andrew Dodds says:

    Angech –

    Coral reefs may have existed in one form or another for most of the Phanerozoic, but they have also been subject to some pretty drastic extinction events; once they were formed by coralline algae, once by calcified sponges ‘Stromatoporids’, sometimes by bivalves and several times by true corals; the final ecosystem of carbonate mounds may be similar but the species assemblage creating it totally different. And – importantly – completely wiped out inbetween, such that a completely different set of reef-builders can evolve. The ‘coal reef’ biome-niche seems to be intrinsically fragile with regard to mass extinction.

    Which means that if humans strip mine the fish from them, acidify the oceans, heat the oceans up and perhaps raise them by 10 meters, there’s a good chance that there won’t be anything left of them.

  94. izen says:

    It might be instructive to compare how attribution is treated in another field.
    In a century of the Tour de France bicycle race it is very rare for a competitor to win it more than once. Only four people have won it more than three 5 times, and then not always in consecutive years. ( In 2 cases there are suspicions it was not entirely the result of natural ability)

    So when Lance Armstrong stated a winning streak of seven years in 1999 it was an extremely rare and as it progressed, a unique, event.
    Armstrong is clearly an exceptional and strong athlete. An El Nino of the cycling world. However from the first year win there were doubts raised that Armstrong was just the extreme end of natural athletic variation. That trace chemicals may have enhanced his natural strengths was always suspected. As the wins accumulated the suspicions grew stronger until close investigation revealed that there was at least SOME evidence that additional chemical assistance was involved to some degree.

    It is likely that Armstrong could have won some of the time on natural ability. Perhaps the chemical boost was only a small percentage of his success. But tellingly the official determination was that ALL his results are at least to some degree attributable to doping and therefore ALL his wins have been declared void.

    We have now had eleven months in a row of record breaking temperatures. something not seen in over a century of records. Some of that can be attributed to natural variation, the recent El Nino. But the exceptional and extreme string of records should, like lance Armstrong’s results, ALL be attributed to the climate on steroids.

    I can understand that those that resist the understanding science has developed of the impact of trace chemicals on the climate may dislike the blanket attribution of ALL events to the influence of extrinsic effects. But it is consistent with how we treat attribution in other fields when distinguishing between natural variation and how the probabilities are skewed by outside influences.

  95. verytallguy says:

    Eli,

    I agree entirely with your assessment, but from everything I read (I’ve just had a skim of WG2), adaptation is commonly termed as investment to protect against against future expected change as well as current losses.

  96. Andrew Dodds says:

    Building a sea wall to cope with sea level rise that has just happened would be adaptation, then, and building several meters higher than currently would be mitigation. Not quite sure that this is the common usage, there’s a grey area. I’d normally think that building an extra high sea wall is adapting to sea level rise that we have now made avoidable.

    Whereas mitigation would be things like ‘Not emitting more CO2 in the first place’ or ‘Putting a solar shade at L1’ or CCS.

    All theoretical anyway, at least in the UK, we are barely adapting to standard 20th century weather and still building on floodplains and at sea level..

  97. Willard says:

    > I am quite happy for you to give an opinion on the likelihood of an 175 x more likely attribution in 14 years for this particularly or any other bit of climate attribution?

    I’m sure you would, angech. Now, please stop peddling that “but 175 X” line. Not paddling, peddling. I’d be happy to call it lobbing:

    Would I be quite happy if you’d stick to the attribution question that Friederike Otto presented instead of “any other bit” of climate attribution? You bet I would. Not as happy as if Nature of Climate Change was Open access.

  98. lerpo says:

    Hi Angech,

    > “I am quite happy for you to give an opinion on the likelihood of an 175 x more likely attribution in 14 years for this particularly or any other bit of climate attribution?”

    It may all be beside the point as Willard says, but it still irks me. You can’t really dismiss an article over claims that it doesn’t make. it says nothing about “175 x more likely attribution in 14 years”. If you want to discuss that article you need to read and understand what it is saying. You may find (once you’ve understood it) that there is nothing really that surprising or unlikely.

  99. Ehhhh, so how is that definition of “adaptation” inconsistent with what Eli wrote? Current losses is generally current projected losses. A loss is a loss irrespective of when it is incurred. Companies and governments generally do accrual accounting these days, not cash. Sure, they may be some discounting.

    The point is, as you may know, that calibrating adaptation for what the losses are expected to be is hard because, as Eli implied, estimating those are among the most difficult of challenges and because the nature of the threat is corresponding hard.

    Professor Robert Young, director of the Program for the Study of Developed Shorelines, had a bit of a laugh at Boston’s HUB Week in 2015 when there were grandiose plans presented for relocating Boston down the road and making the downtown a kind of ocean-resistent park. His point was and is that we are failing to pay attention to the most valuable information we have: Nature is telling us quite explicitly where we should not build or rebuild. According to reports from him and others, the Hurricane/Post-tropical Storm Sandy experience is horrible: People are reconstructing in place. I’d be loved to be proved wrong, but from the before-show scuttlebutt has it that rather than replan the severely-exposed-to-weather-sea-and-sea-level-rise South Boston development plans, Boston, on the 15th, is going to attempt to propose the municipal equivalent of shelter in place or this, too, will pass. That’s not even adaptation. That’s Wishful Thinking.

  100. angech,

    Thanks Brandon. Let us put it down as number 3.
    Another 7 different ones please if anyone can help out.
    10 would be a good sample to work from

    You’re welcome. As Anders, Willard and others have already pointed out, Dr. Otto’s paper and comments are an awfully topical example from which to work. Can you not discuss her specific arguments on their merits or lack thereof? Would they not stand (or fall) by themselves without needing to appeal to someone else’s arguments other than hers?

    Please discuss.

  101. verytallguy says:

    Current losses is generally current projected losses.

    Surely current is now, projected is future. But I’m obviously missing some subtlety.

  102. If someone told a company they would incur $1 million of losses in 2023, and the source was highly reliable, they would anticipate that loss in their bookkeeping. This is how possible (large) legal settlements are modeled. They cannot allow a big hit to destroy the company, so they plan for it. If it turns out it does not happen, so much the better, but if the source is reliable, they aren’t going to ignore it: That’s just foolish management. Of course, if the source isn’t so reliable, they might ignore it. So, even if the losses are projected in 2023, they are put on the books now, as I said, perhaps with a discount for the time into the future.

    The point from Eli’s Four Laws, whether companies are being disingenuous or not, is that by asking for concrete costs to adapt, they are effectively ignoring the problem, since no set of shareholders will be happy with management’s decision to adapt based upon the large degree of uncertainty in size and kind which near term climate projections have. Ditto, I think, politicians. Are they really going to sink $15 billion into a sea wall which might be overtopped by some accounts in 20 years?

  103. verytallguy says:

    OK, if that’s yours and Eli’s meaning then we are in complete agreement

  104. angech says:

    brandonrgates says:
    As Anders, Willard and others have already pointed out, Dr. Otto’s paper and comments are an awfully topical example from which to work. Can you not discuss her specific arguments on their merits or lack thereof?
    Sure.
    Would they not stand (or fall) by themselves without needing to appeal to someone else’s arguments other than hers?
    Anders is asking the attribution question.
    It is much bigger than Dr Otto on her own.
    The more examples we have of provable or probable attributable climate events, the more support for her arguments.

  105. angech says:

    brandonrgates “This paper by Friederike Otto and colleagues points out, regionally it depends on both thermodynamic changes (more energy) and changes in atmospheric circulation”.
    Basically, on the video, She points out that attribution of climate events to AGW is very ,very hard because the atmospheric circulation changes are too chaotic and unpredictable for us to have working models on and that they interfere with the expected effects of the thermodynamic effects hence attribution in any one locale [otherwise known as a climate event] is strictly not actually able to be calculated effectively.
    It is a known unknown which makes anyone offering specific attributions have to really work for it.

  106. angech,

    It is a known unknown which makes anyone offering specific attributions have to really work for it.

    We agree. Was that it then?

  107. angech says:

    Willard says:
    ” I am quite happy for you to give an opinion on the likelihood of an 175 x more likely attribution in 14 years for this particularly or any other bit of climate attribution?”
    I’m sure you would, angech. Now, please stop peddling that “but 175 X” line. Not paddling, peddling. I’d be happy to call it lobbing:
    Looking back you are correct. I did mention the “but 175 X” line from the Guardian first.
    BTW that cartoon seems to mirror Pogo ‘”We have met the enemy and he is us” in saying we deduced where we would dig in if we were over there. Confirmation bias.
    Perhaps the enemy had that Italian fellow from PB on their side.
    I reiterate that this discussion needs some examples put up.
    Would you care to lob some over or put some up.
    If you cannot or will not that is fine but it leaves the playing field open to “that” claim which you wisely refuse to touch and did sort of condemn.
    Are there no concrete examples of attribution to discuss?
    Brandon does not want to engage either.
    “We agree. Was that it then?”

  108. Willard says:

    > I reiterate that this discussion needs some examples put up.

    Not really. Otto’s paper stands on its own.

    A Grauniad article ain’t even a valid example.

    ***

    > and did sort of condemn.

    Where did I “sort of condemn,” again?

  109. angech,

    Brandon does not want to engage either.

    What Willard said.

  110. lerpo says:

    Angech: If you cannot or will not that is fine but it leaves the playing field open to “that” claim which you wisely refuse to touch and did sort of condemn.

    Which claim? That “175 times rate of occurrence of El Niño events?”

    That “175 x more likely attribution in 14 years””

    No one has made these claims but you.

    The claim in the article is that a temperature rise of 1.5C in the Coral Sea is 175 times more likely when you start with a 1C head start due to anthropogenic warming. I’m not sure why we should be shocked by that.

    Maybe you could detrend the temperature record for the Coral Sea (to remove the anthropogenic warming signal) and see how often a rise of 1.5C occurs relative to a rise of 0.5C? That may give you a ballpark ratio. If you find something an order of magnitude different from 175x then maybe you have a point. My guess is that you will find that a rise of 1.5C above the trend has never occurred. The current (rather large) El Nino apparently caused a rise of only 0.5C above the trend.

  111. angech wrote “Brandon does not want to engage either.”

    This is where attempts at engagement with angech end up. If angech can’t even be honest with him/herself, what is the point in trying to have a rational discussing with him/her? Stop playing games angech (“Brandon does not want to engage either” appears to be just another one), and maybe people might be willing to engage, it is your choice.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s