If the climate system is in a fairly stable, if chaotic orbit, as it has been in the late Holocene (say the last 7000 years) then there is some room for climate heuristics. An anomalously warm North Atlantic indicates how far north the tropical convergence zone will migrate in Brazil that July, hence whether the north coast of South America will have a dry year.Now I have to admit that I can't recall which way it works!
I avoided chemistry and biology as a student because I have trouble committing facts to memory. I had a great work-around for exams, mind you. I would create a dense single page of notes for any closed book exam. I would copy and recopy it about eight times. If the exam was a morning exam, I'd stay up all night doing so; for an afternoon exam I'd just do this in the morning. I would not vary the format in any way. Thus, I would commit the materials to short term memory, secure in the knowledge that I would forget them the next day. I did well as an undergraduate.
In grad school, I tended to expect being tested on understanding, not on memorization, but I did have to resort to the page-recopying trick for Dr. Hastenrath's class on Tropical Meteorology. His class and his book were a thick compendium of observations, rather light on principles. (There were a couple of very crude equations, concessions to basics like the thermal wind law, for instance). One of them was whether a warm north Atlantic was good news or bad for Northern Brazil. It was one or the other, and I'm sure I had the fact ready at hand during the exam.
This sort of heuristic has some modest value I suppose. If you can say this will be a dry year more likely than not, or a year with more tropical storms than usual, based on a collection of heuristics, then to the extent that is reliable, you can help people plan. And you can just comb through the data looking for correlations with some predictive value. It's an activity which looks like science. It has enough value that you can get grants for it. NOAA regularly puts out seasonal outlooks for the US which have a couple of regions marked as '>60% chance of above normal temperatures' or the like.
I've always had a couple of problems with this technique. First, I'm unconvinced it actually helps anybody do anything. 60% chance of above normal really isn't a very strong statement. Are there really activities which are rationally conducted differently in the light of such information? Second, it confuses people about climate science; this is neither climate science nor weather prediction, both of which are based in physics. It's pure heuristics. Yet it's usually called a "short-term climate prediction". Given the importance of climate change in the future, this is unfortunate and misleading nomenclature. Third, there is indeed accelerating climate change, so the value of heuristics will decline as the underlying conditions which supported the heuristics also change. The heuristic method is pretty much guaranteed to get worse.
(Now, if we ever get to predictive models of deep ocean circulation, maybe short term predictions of climate will be physically meaningful. At present this seems rather speculative, but people do seem to be working on it quite seriously.)
There's a another issue, though. The empiricists have always, to some extent, treated the physicists as a threat. Somebody who is good at digging into history should look into this, but I have several points of corroboration. First, there is the (from where I sit) unexplained tension between the famed empiricist Reid Bryson and the groundbreaking climate physicist Vern Suomi that is at the heart of the meteorology program at Wisconsin splitting into two institutions. Second, there was my brief meeting way back in 1993 or so with Bill Gray, the famous hurricane prediction guy, the classic empiricist. I hadn't even made a name for myself on sci.environment in those days; and it was really unlikely that he had heard of me. My advisor, John R. Anderson, was known more as a modeler than as a climate change guy. In those days there was little sense that the topics were tightly connected. John introduced me to Gray just as a grad student, period. It wasn't a minute into the conversation before Gray was ranting and raving about how hurricanes were more important than global warming, and how his money was drying up, etc.
So it's interesting to see that Gray is not alone among tropical storm researchers in this hostility. There are certainly signs of more than ordinary strife within the tropical storm science community. I can't say I have an inside view of what's going on but I note that Knappenberg and Curry both come out of the tropical storm community. I don't know how close they come to the Bill Gray tradition but I suspect there is some connection.
See, empiricism lacks consilience. When the science moves in a particular direction, they have nothing to offer. They can only read their tea leaves. Empiricists live in a world which is all correlation, and no causation.
This is the third installment of the empiricism series.
1) Empiricism
2) The Empiricist Fallacy
Next: Empiricism and Denialism
Image: NOAA. Extra credit: why do Mexicans and Canadians find this map irritating?
23 comments:
O/T but that's why, as a part of my job, I've never been great with metallurgy. Too much like chemistry (or cooking for the matter of that).
MT --- be working on it quite serioulsy.
Thanks, fixed along with two other typos in the same paragraph. The spell-check feature that works in Firefox fails in Safari.
Very minor points, hardly worth raising.
Tobis: > "60% chance of above normal really isn't a very strong statement... it confuses people about climate science; this is neither climate science nor weather prediction, both of which are based in physics. It's pure heuristics."
Not necessarily - granted this particular nugget in this particular time is empiricism/statistical+historical,but a physics simulation (or a series of runs) *might* output a distribution for a value instead of a single number, and some of those distributions could have high variance. There is information in the mean, and there is information is the scale of the variance.
And so, the denialists give the game away when they argue that high variance implies support for business-as-usual. Implying that you would be less compelled to remove the single bullet from a six chamber gun, than a five chamber gun, before holding the gun to your head and pulling the trigger. Foolishness - removing the bullet is equally compelling no matter how many the number of chambers, because the action of removing the single bullet is to eliminate the very real worst case, and the *abstract average over all cases* does not enter into it. (Assuming a low cost to remove the bullet. Some insurance is too expensive no matter how horrific the unlikely worst case, or there are too many competing disparate terrible cases, but that has to be argued.)
Tobis: > "60% chance of above normal really isn't a very strong statement. Are there really activities which are rationally conducted differently in the light of such information?"
You could secure a perverse thrill from a short against a business acting with certainty that weather will be mild. And you could secure a nice payoff if your short pays off. ;-)
Moe, yes, if this series goes where I am trying to get it to, that point about variance, which I ,ake often enough myself, will be addressed.
But the predictions (I added a recent one as an image above) don't say anything about variance. They just say there's a slighlty above or below average chance that the weather will be at least slighty above or below average. (I think it's a three-quantile ('tertile'?) system.) The map shows where there is more than a 1/3 chance of it being warmer than normal or colder than normal. A similar map is issued for wetter/drier.(The four month prediction right now is completely blank.) I suppose that in mathematical principle it's possible for a region to get into both colors but I've never seen it; likely the heuristics don't handle that.
Eli is having this discussion about the AMO, yeah, you can find it in the observations, but what is the cause, cause without a cause you have only chance
In models, AMO-like variability is associated with small changes in the North Atlantic branch of the Thermohaline Circulation, however historical oceanic observations are not sufficient to associate the derived AMO index to present day circulation anomalies.
from
http://en.wikipedia.org/wiki/Atlantic_Multidecadal_Oscillation
Curry was hardly in the Gray zone in 2005. And she had some differences with empiricist Landsea in a related discussion on PBS a couple of weeks later.
The Landsea discussion is referenced here.
Steve, you are messing with my narrative! But better early than late and better late than never, I suppose.
I will claim that the McIntyre point of view is essentially empiricist. Curry certainly seems to have adopted that position if she didn't start with it.
I think Steve Easterbrook has already pre-messed with this narrative. Maybe he's taking one horn, and you're taking the other. You've got the hat for it...
When I was a ten-year old displaced Canuck living in Texas (dad was a jet jockey teaching USAF student pilots to fly during the end of the VIetnam war) we had a big-assed map on the front board with North America in the centre, with a huge USA and a disproportionally sized (smaller) Canada and Mexico. I pointed out that Canada was bigger than the USA and my teacher got all huffy and stated that it was a well-known fact that the USA was bigger than Canada. Of course, I meant land area and she meant population.
By then, I was already known as a trouble maker for not wanting to say the pledge of allegiance, requiring the Principal to intervene with my parents, but that's another story...
The correct answer to only question on the first exam that I took as a chemistry major was: “there is a math error in the assigned reading!”
The professor wanted us to think, not to recite. Ultimately, he wanted us to think, and solve problems.
The professor’s first step in solving a problem was to define the system and the second step was to bound the system. This is also the core of statistical sampling technique.
Define and bound are steps that I do not see in empiricists’ work. Unless they define/ bound the systems, they never know what problem they have solved, or what they have proved statistically.
Peer review is at fault, for not standing up and whacking empiricists.
Empiricists like Curry, Landsea and Gray are more than capable of arguing amongst themselves.
It's a misreading of what I am saying to suggest "empiricism" is synonymous with "wrong" or "denialist".
What I am saying is that the data alone, in the absence of physical interpretation, cannot tell you much.
That is not to say they can tell you nothing or that there is anything wrong with getting what the data can tell you out of the data.
I am, however, pointing out that if there is a tendency among empiricists to presume a stable climate where human forcing is unimportant, it wouldn't be surprising. And that we do see some signs of this.
MT asked: "Extra credit: why do Mexicans and Canadians find this map irritating?"
Canadians may be irritated because it ignores the vast majority of the area of the country yet it still covers the part where the majority of us live; reminding us that:
"Seventy-five percent of Canada’s population resides in a narrow 150-kilometre band pushing up against the U.S., with close ties south of the line. We are a border people. The border is our livelihood. The border is our identity."
http://tinyurl.com/3x5y9bg
Most Mexicans don't snuggle up quite so close geographically but, even so, they consider the fact of their proximity to be an inconvenience, eg, President Porfirio Diaz: "Poor Mexico! So far from God, and so close to the United States."
I think you're taking an unfair swipe at seasonal ('short term climate') predictions. All the reputable empirical seasonal prediction schemes (IRI, Australian BoM, various pacific outlooks and the NOAA one you're picking on to name a few) are based on physical explanations of why the predictor (say ENSO) is associated with increased probability of certain conditions (say increased rainfall in the US). You'll find the professionals that work in this field have nothing but contempt for schemes that are based solely on statistical relationships in the absence of a compelling physical explanation. Could you take a closer look and then let us know if you still think it's all 'pure heuristics'?
Yes, most of the time the odds are close to climatology in the mid-latitudes and higher, because most of the time that is all that the empirical relationships justify. Water managers, agricultural producers and insurance companies are all able to incorporate this sort of information into their decision making.
"heuristic"... that word again...
"the IF application in that post was intended as a simple heuristic device to think about the the uncertainties in aspects of the IPCC and to raise issues of ambiguities in the ipcc’s confidence statements"
but the IFA as presented here is far from being a simple heuristic device
Alex, it is a swipe, but I am not convinced it is unwarranted.
First, showme evidence that the faith that these predictions are considered useful is warranted. I want to hear even one customer say "thank goodness we have this information to guide our decisions". Anyone. A water planner. A fuel planner. A farmer. I don't want theory. I have heard the theory. I want testimonials.
Second, yes, given the heuristic you can tell a plausible story about it. If nothing else it forms a useful mnemonic. But it's generally possible to come with a plausible story whichever way the answer comes out. That isn't science.
I realize this is harsh.
I have come to understand that my job is first and foremost to be interesting and honest, not to make friends.
But, as part of my honesty, I have to be willing to eat my words. So, hurt feelings noted. Now, evidence, please?
Michael, (I believe) I understand your points but I think you're oversimplifying your final conclusion. [Side point: you appear to use "heuristic" in a much different sense (= empirical facts = observations) than I think of it as (something that provides insight into how something works).]
Observational data is collected for different reasons. If you have little understanding of the relative roles (or even identities) of the different possible drivers of a system, or even a basic understanding of the types and magnitudes of change of that system over time or space, then observational data is designed just to help you have an overall sense of what is happening and give as many hints/clues at what is driving it as possible. This is the basic idea behind survey sampling, and if done well, has great power to provide insight into a system's nature.
As one's sense of what's going on accordingly gets stronger, one can (and should) then target one's data collection to help discriminate more powerfully between the more limited set of possible explanations that emerge from survey sampling. This of course is the basic idea behind hypothesis testing/discrimination (or theory building if you like).
Both of these approaches are necessary in the overall scientific process. So I don't think it's correct to say that "empiricism lacks consilience"; consilience is a synthesizing process in the human mind that relies on much observational data, some of it tied to specific hypotheses, some of it not.
Or maybe I misunderstood your points entirely...
Jim, I am talking about something here. I am not sure I am using the right words for it. Let me ponder.
"anthropogenic climate change may be unlike past variability; empirical relationships based on historical observations may be inappropriate for projecting ecosystem responses to future climate change."
Rykaczewski, R. R., and J. P. Dunne (2010), Enhanced nutrient supply to the California Current Ecosystem with global warming and increased stratification in an earth system model, Geophysical Research Letters, 37(L21606), doi:201010.1029/2010GL045019
It's not Alex, it's Andrew. Apologies for the slow reply, I've been busy fending off zombies.
This abstract gives a good summary of the current state of the use of seasonal forecasts in Oz:
http://www.publish.csiro.au/?paper=AR06200
You won't find many gushing testimonials from users of seasonal forecasts in developed mid-latitude areas because they are one of many sources of information that affect decisions, and because they are rarely emphatic. This is a new field. You will plenty of modelling studies demonstrating economic potential, but few studies of the actual use of seasonal forecasts. The period in which one could have used seasonal forecasts in an application is so short as to make it difficult to quantify actual utility robustly. Likely the next 5 years will see dynamical models dominating the forecast schemes, so it may be difficult to test any claims to the real utility of empirical seasonal forecasts.
Post a Comment