There’s been some hoopla in the blogosphere lately about comparing projections of temperature from the IPCC TAR (third assessment report), published in 2001, to observed temperature. The comparisons have been made to temperature data since 2001, on the basis of the claim that that’s when the projections start so that’s when the comparison should start. It appears such claims are in error.
In a previous post, I addressed some simple ways to account for the effect of autocorrelation on estimating trends (in particular, the probable errors in such estimates). I applied those methods to global temperature data from NASA GISS, from which it emerged that the data since 2001 do not contradict warming at a rate of 0.02 deg.C/yr (or 0.2 deg.C/decade), the warming rate claimed to be the projection of the IPCC TAR.
As a reader pointed out, the IPCC TAR projections actually don’t start in 2001. That’s when the TAR was published, but the individual studies on which they’re based were published before that, and of course the computations were done before publication. It turns out that the models used to make those projections begin their independent (from any observations) computations in 1990. A proper comparison of projections with subsequent observations appeared in Rahmstorf et al. (2007, Science, 316, 709, hereafter referred to as “R07″), which examines not only global temperature, but CO2 concentration and sea level as well. The authors clearly state, “Although published in 2001, these model projections are essentially independent from the observed climate data since 1990.” The temperature data used for comparison in R07 are annual average land-ocean surface temperature from GISS and HadCRU, from 1990 through 2006.
Pointing out that comparison of IPCC projections with observed temperature has already been done in the peer-reviewed literature, has led to questions about the origin of this temperature graph in R07 (I’ve added labels to the x-axis to indicate which years are graphed):
Lucia, for instance, makes a query on her blog which is well described as pleading for an explanation. This leads to considerable speculation in reader comments. I too was curious about the origin of this graph, but I took a different approach than asking my blog readers. Forgive the twisted logic of my bizarre approach, but the strategy I adopted was: ask Stefan Rahmstorf.
His reply was to the point: the graphs were made using actual data which was provided by the IPCC authors. Imagine that. The grey range and scenarios in R07 are exactly those famous future scenarios shown in the IPCC TAR Summary for Policy Makers in Figure 5. Here’s an updated version of the graph which includes more current observed temperature data (click the graph for a larger, clearer view):
The solid blue and red lines are the trends from GISS and HadCRU data, the dashed lines are the IPCC projections included in the TAR. The graph speaks for itself rather well.
The models used for the TAR were developed in the mid-1990s. They’re not statistical models based on fitting observed data, they’re models based on the equations of thermodynamics and hydrodynamics. They weren’t “tuned” or updated using any observed climate data subsequent to 1990. Furthermore, it’s just irrational to claim (as some have suggested) that model developers would subsequently have used observations post-1990 to change their models; in the mid-90s, the period after 1990 would have been way too short to be meaningful for model tuning, just as the period starting in 2001 is today too short for us to draw any conclusions about trends (which we’ll see shortly). The period 1990-2006 used in R07 is just about long enough for a meaningful comparison — that’s why R07 published it at the time.
The projections shown in R07 don’t follow the strict 0.2 deg.C/decade rate of increase claimed to represent IPCC projections. So what are the actual projections from IPCC TAR? The TAR states in more than one place that we can expect “about” 0.2 deg.C/decade warming over the next few decades. Note the qualifier “about,” and that the quoted figure is accurate to only 1 significant digit. The logical conclusion is: between 0.15 and 0.25 deg.C/decade, in which case flatly stating 0.2 deg.C/decade as though that were an ironclad exact figure isn’t an honest representation of the IPCC projection. But we needn’t be satisfied with so imprecise a figure. Using the actual data from model runs used in IPCC TAR and referred to in R07, we have these estimates for trend rates from 1990 to 2010, and for that matter from 2000 to 2010:
Note especially the final entry, which gives the average of the included scenarios. For the two decades following 1990 the average rate is 0.0165 deg.C/yr (0.165/decade), for the decade following 2000 the average is 0.0174 deg.C/yr (0.174/decade). Hence if we want to compare observation to projection starting in 1990, we should use the more precise figure 0.0165 for IPCC projection, and if we follow others by starting at 2001, we should use 0.0174. Considering how close the error ranges come to the purported projection of 0.02/yr, it makes quite a difference.
And what is the observed trend rate since 1990? R07 display both GISS and HadCRU data, so I’ll look at both of them. I’ll estimate the trend from ordinary least squares (OLS), but I won’t compensate OLS using a simplified estimate of the impact of AR(1) autoregression, or Cochrane-Orcutt estimation, both of which assume an AR(1) model for the random fluctuations. For the AR(1) model, the autocorrelation at lag j is given by
However, as a reader pointed out in comments to the preceding post, the autocorrelations at lag greater than 1 depart from this pattern, tending to be greater than their AR(1) values. The result is that although the AR(1) model certainly gets us in the ballpark, giving a realistic estimate, it underestimates the error range, which is larger than indicated by the AR(1) model. Since the final result of that analysis is rather a “close call,” it turns out that the AR(1) model, while sufficient for many purposes, won’t do for this job; it’s necessary to apply more realistic estimates of the autocorrelations. I’ll also use the exact formula for the impact of autocorrelation on the probable error in an estimated trend rate from OLS (see Lee & Lund 2004, Biometrika, 91, 240).
Results? For GISS data the trend estimate since 1990 is 
I’m sure many readers are interested in results using only post-2001 observations. GISS data indicate 
Finally, it must also be mentioned that even these error ranges, based on exact methodology and a realistic autocorrelation model, are still too small because they don’t include the uncertainty in estimating the autocorrelation parameters. So the error bars really should stretch even further than they do in this graph, which serves to emphasize the failure of observed data to falsify the IPCC TAR projection. It’s more work than I care to undertake to incorporate the impact of this final uncertainty into the numbers, but in this case it’s not necessary because we know that the error bars will get bigger — and they’re already big enough to rule out any falsification of IPCC TAR projection.
It also serves to emphasize the tremendous uncertainty is trend estimates from a mere 7 years of data with a signal-to-noise ratio as small as that we see in global temperature. Over a longer time span, the signal can emerge from the noise with sufficient precision to make meaningful comparison of projection to observation. But over such a short time span the long-term trend is swamped by the effect of the noise, so we effectively end up comparing random fluctuation in observed data to long-term trend from projections. That invites faulty conclusions — and in this case, unless one applies considerable rigor to the analysis it actually spawns faulty conclusions.
How, then, do some conclude that temperature data falsify IPCC TAR projections?
- Use the wrong start date, so that you
- Compare short-term fluctuation to long-term trend
- Make the rate projected by IPCC too high.
- Make the error range, even when compensated for autocorrelation, too small
I’ll close by quoting an important summary paragraph from R07:
Overall, these observational data underscore the concems about global climate change. Previous projections, as summarized by IPCC, have not exaggerated but may in some respects even have underestimated the change, in particular for sea level.
I couldn’t have said it better myself.
164 responses so far ↓
kim // March 26, 2008 at 3:51 pm |
How about comparing projections from AR4 with temperatures from 2001?
================================
[Response: The tremendous uncertainty in estimated trends based on only 7 years of data, with a signal-to-noise ratio as small as we observe in global temperature, makes such a comparison unlikely to reveal meaningful insights.]
kim // March 26, 2008 at 4:04 pm |
One insight it might reveal is that with stable or dropping temperatures for three more years, then the IPCC’s projection of 0.2 degrees centigrade temperature rise per decade would be falsified at the 95% confidence level. That would be a meaningful insight if a flipped PDO gives us stable or dropping temperatures over the next three years.
Shhh, don’t tell Pachauri, or any journalists or politicians. It’s a secret.
=============================
[Response: As this post (and Rahmstorf et al.) illustrate, characterizing the IPCC projection as 0.2 deg.C/decade is just not correct -- that figure may apply if one examines the entire 21st century or even up to 2050 or thereabouts, but not to the last couple of decades. That's one of the fatal flaws in other attempts I've seen to falsify IPCC projections. Perhaps the "best" characterization of this and the preceding decade is the average of the included models, which turns out to less than that figure. But as there are many models included, even characterizing it by a single value is an incomplete description.
Three more years of data will reveal more information. Whether or not that will falsify any projection depends on how strongly the signal emerges from the noise, but unless the results of simpler analyses are clear-cut, a rigorous analysis is necessary to draw reliable conclusions.]
JCH // March 26, 2008 at 4:21 pm |
Well, what if somebody isn’t looking for meaningful insights?
Adam // March 26, 2008 at 4:33 pm |
Would there be anything added if the range for the IPCC prediction (grey area) was added to the red line on the last plot (is it even possible?)?
I understand it’s not the same thing as the error range in the observed trends, but the grey area does suggest quite a wide range in predicted/possible rates of warming.
Adam // March 26, 2008 at 4:35 pm |
Oh, also great post, BTW.
kim // March 26, 2008 at 4:59 pm |
The fundamental problem, and a very obvious one, is that the models assume no natural climate variability. Then, any temperature variability from the ‘trend’ is assumed to be weather noise, and our dancing calculations angelically miss the real contribution of CO2 to temperature, and climate.
==============================
fred // March 26, 2008 at 5:15 pm |
tamino, isn’t the interesting and important question, whether the events since 2000 or whenever have falsified a prediction the IPCC did not make, namely, that temperatures would stay the same? Not much hangs on the question of whether the IPCC did or did not make a mistake in one particular estimate. A lot hangs on whether its warming or cooling.
[Response: Consider the data since 2001 (since I've already run those numbers). The GISS data are indeed consistent with (i.e., don't falsify) a warming rate of 0, or a sizeable negative (cooling) rate. But they're also consistent with a warming rate *twice* as high as the IPCC projection. And as I mentioned in the post, those error bars are still too small because they don't account for the uncertainty in autocorrelation estimates. Yet another factor, which I failed to mention, is that this analysis only considers random fluctuations as part of the genuine temperature evolution, it omits to consider the additional uncertainty due to the measurement and estimation process itself -- if there were no such uncertainty, then GISS and HadCRU (and NCDC) would be in complete agreement, which obviously they aren't. All of which again underscores how little we can determine from such a short time span, given the noise level.
In a previous post I outlined the conditions under which I'd accept that we have solid evidence of a cessation of warming, and noted that for these data the time span which is "likely but by no means certain" to allow a conclusion is 15 years (from start 2001 to end 2015). I've also posted about the fact that given the nature of the random fluctuations, encountering time spans which appear (without rigorous analysis) to reverse the warming trend isn't just expected, it's *inevitable*. Flip that coin often enough, you're gonna get 10 heads in a row.
I think the root of the idea that even very brief time spans must always falsify cooling, is a lack of sufficient comprehension that noise really exists, in both measurements and the climate system itself, that on subdecadal time scales it's considerably larger than the signal, and its effect isn't some rare noteworthy event, it's ubiquitous.
Physics, both analytical analyses and computer simulations (which are physics, not statistics), is quite clear: warming will continue. Statistical analysis has yet to contradict that.]
Mike B // March 26, 2008 at 6:14 pm |
Next year after all the 2008 temperature data is in, how many “interested parties” will there be running statistical analysis on the last 10 years (1999-2008) to see if that falsifies the IPCC predictions?
Thomas Huxley // March 26, 2008 at 6:50 pm |
Tamino
You are very careful to use the phrase “IPCC projection” rather than “prediction”. Is there a subtle difference?
Or are they the same thing?
[Response: That's the word chosen by Rahmstorf et al., so it seemed appropriate. Wiktionary defines "prediction" as "a statement about what will happen in the future," and "projection" as "a forecast or prognosis obtained by extrapolation," while "forecast" is defined as "an estimation." From that it would seem that "projection" doesn't strictly apply, because the IPCC TAR projection isn't based on extrapolation, but neither does "prediction" because clearly it's an estimation rather than a definitive statement.
I'll leave it to others to decide what word is most appropriate, and to the folks at Oxford to define words more precisely.]
sod // March 26, 2008 at 7:09 pm |
very good post, very well written. love this part:
How, then, do some conclude that temperature data falsify IPCC TAR projections?
* Use the wrong start date, so that you
* Compare short-term fluctuation to long-term trend
* Make the rate projected by IPCC too high.
* Make the error range, even when compensated for autocorrelation, too small
————————
Mike, nex t year 11-year averages will be the hype. you wouldn t want to lose 1998, the year that allows an eyeball “earth is cooling” analysis, would you?
TCO // March 26, 2008 at 8:21 pm |
I just scanned the stuff that was written in February (which others here linked to). It seemed there that Lucy had a comment about not accounting for annual temperature autocorrelation and just looking at a normal distribution based on the (implicitly iid) standard deviation. Did I read that wrong? Or did she change her analysis in some later posts (I know she had a lot of them). I still have not gotten a good description of how she modeled the autocorrelation. Saying AR1 defines a certain type of autocorr (not the most highly fitted), but it doesn’t say waht the single term is equal to (iow the AMOUNT of random walk).
BTW Tammy, it’s annoying that you don’t link to the specific Lucy posts. Here is the one that I saw, which does not talk about autocorrelation:
http://rankexploits.com/musings/2008/can-ipcc-projections-be-falsified-sample-calculation/
[Response: I see that post is based on analysis of annual data since 1998: 10 data points. Annual data show less autocorrelation that monthly, but it's still not zero. However, *estimating* autocorrelation coefficients using just 10 data points is next to impossible. In that case, it would be better to estimate the autocorrelation using a longer time span of data, and apply those estimates to analysis of the 10 data points.
It appears that her more recent attempts to falsify IPCC projections is based on monthly data, but in that case the AR(1) model isn't really good enough to do the job.]
Thomas Huxley // March 26, 2008 at 8:25 pm |
Re: prediction versus projection. Thanks very much for the clear distinction. I think you are right to prefer estimation or estimate. If only words had “error bars”, but then again maybe they do!
Paul Middents // March 26, 2008 at 8:25 pm |
Good post.
How many different ways do you have to explain the futility of using a short segment of a noisy time series to draw conclusions about a long term trend? The usual suspects are wasting no time pecking at your current effort.
Atmoz had a similar exercise which associated much of the noise in the temperature series to an idealized (spherical) El Nino Southern Oscillation (ENSO). He includes some nice graphics for the mathematically challenged.
http://atmoz.org/blog/2008/03/14/first-assume-a-spherical-enso/
kim (March 26 at 4:59 PM) makes the stunning assertion that “that the models assume no natural climate variability.” This might have been enough for you to moderate it into the wall but I’m glad you didn’t. It reveals a point which may be lost on some—that global climate model outputs show variability that mimics El Nino/Southern Oscillation (ENSO) like behavior. I think this is a measure of model “skill”. A very active current aspect of climate research is the effect of global warming on ENSO. Start with the references for the Wiki entry on El Nino. They include Merryfield, 2006, Changes to ENSO under CO2 Doubling in a Multimodel Ensemble, Journal of Climate, 19, 4009-4027.
http://www.ocgy.ubc.ca/~yzq/books/paper5_IPCC_revised/Merryfield2006.pdf
Following this paper forward via papers that reference it we find a recent one: Soon-Il An, Jong-Seong Kug, Yoo-Geun Ham, In-Sik Kang, 2007: Successive modulation of ENSO to the future greenhouse warming. Journal of Climate (Accepted)
http://climate.snu.ac.kr/2005_new/pub/papers/i61.pdf
Before some eagle eye points it out for me, I’ll quote from their summary:
“The sensitivity of ENSO to future greenhouse warming predicted by the various IPCC
AR4 Climate Model simulations is highly model-dependent (e.g., van Oldenborgh et al.
2005; Merryfield 2006). Thus it may be premature to conclude how ENSO may vary
during global warming. Nevertheless, it should be valuable to explore a fundamental
mechanism on the changes in ENSO variability modulated by greenhouse warming.”
These guys are not overreaching nor am I trying to over interpret the importance of their work. This is just a sample of work in this important area.
The whole subject of ENSO and other cyclic climate phenomena like the Pacific Decadal Oscillation (PDO) might be a rich area for our host to educate us. Our own University of Washington has a group dedicated to this work and a good web site chock full of data:
http://jisao.washington.edu/
They even discovered the Pacific Decadal Oscillation!
Aaron Lewis // March 26, 2008 at 8:59 pm |
The real question is, “How much air temperature change does it take to be significant?”
The atmosphere may be an important part of the climate system, but it is not the entire system, and the different parts of the system are not in equilibrium. GISS and other air temperature databases do not reflect heat taken up by deep ocean waters or warming ice or melting ice. There is a lot of ice that was “very cold,” and now is only “cold.” There is a lot of ocean water that is a bit warmer.
The IPCC models do not reflect the complex physics of warming ice. The melting of sea ice and permafrost has absorbed heat. There is a surprising amount of sub-glacial water that was ice. Nevertheless, this heat accumulated by global warming will eventually be accounted.
Remember that the difference between ice and water is zero degree of temperature. We can melt a lot of ice without changing the global temperature. Global warming and global temperature as measured by sampled air temperatures are NOT related by a linear function. The fact that these functions are highly non-linear will reduce the power of any statistical models that assume a linear relationship.
Falsification of global warming requires proof that heat is not accumulating, rather than simply a demonstration that air temperatures are not rising at some predicted rate. Proof of global warming rests on evidence of total heat accumulation, rather than on flux in some set of air temperatures.
TCO // March 26, 2008 at 9:24 pm |
Well of course, you would use a longer earlier set to give you the autocorr characteristics. Heck even were it iid, you would want to calculate the std deviation from year to year based on the overall data, not the in the cross hairs data. What a maroon. Why do we skeptics do this to ourseleves??
[Response: Perhaps Lucia's efforts represent a learning process rather than being a "maroon."]
TCO // March 26, 2008 at 10:03 pm |
Mebbe so. But how about the maroons on my side touting her work? They need to be disciplined. Spankings. Naked ones.
dhogaza // March 26, 2008 at 10:24 pm |
Not seeing much evidence of learning, here, since she’s insisting on her “proof”, telling WIlliam Connelly he doesn’t know what he’s talking about, etc etc.
I think it’s stubborness, not “marooness” at work, though.
David B. Benson // March 26, 2008 at 10:48 pm |
A reminder for an earlier poster:
From
http://en.wikipedia.org/wiki/Ice
When ice melts, it absorbs as much heat energy (the heat of fusion) as it would take to heat an equivalent mass of water by 80 °C, while its temperature remains a constant 0 °C.
kim // March 26, 2008 at 10:49 pm |
Paul, why do you fail to convince? Is it because the IPCC was tasked to explore anthropogenic climate change? And output isn’t the same as underlying assumptions.
Tamino, every lowering from 2 degrees centigrade per century means less cause for alarm. At what projection does climate change not become catastrophic? And sure, we know there are enough models to justify a multitude of sins.
Wait until it gets warmer outside, TCO, before you suggest naked spankings. Why don’t you try to figure out the real sensitivity of climate to CO2 so that honest advice can be given to policymakers?
=========================
[Response: This continual misrepresentation of IPCC projections is getting really annoying.
The same exact models which average to 0.174 deg.C from 2000 to 2010, also indicate 2 to 4.5 deg.C this century.]
TCO // March 26, 2008 at 10:52 pm |
Connely is nothing special. Tammy’s better than that. WC is a camp follower, wiki-weenie. Tammy can do math.
dhogaza // March 26, 2008 at 11:18 pm |
Strange to think that a mathematician from Oxford can’t do math. Or are you claiming that he’s lying about his background?
I’m sure the reason he got his job as a climate modeler at the BAR is because he’s mathematically illiterate.
Johan i Kanada // March 27, 2008 at 12:16 am |
Tamino,
Can you explain why you say that:
i) Having the two 1990’s trend values slightly higher than the IPCC model, “more likely than not” implies that the IPCC models underestimate the actual trend
ii) While, on the other hand, having two 2000’s trend values significantly lower than the IPCC projections “serves to emphasize the failure of observed data to falsify the IPCC TAR projection”.
Aren’t you sort of picking the conclusions you want to see here? I mean, you could have written that in case ii) the observations “more likely than not” overestimate the actual trend?
[Response: OK, I admit that those two statements represent arguing for my beliefs rather than a completely dispassionate appraisal of the results. I've made no secret of my belief in AGW. Perhaps the most objective statement is that both time spans are consistent with IPCC projections, neither indicates departure (either positive or negative) from the projected trend, and the error range from data post-2001 is so large as to prohibit meaningful insight.]
Heretic // March 27, 2008 at 12:48 am |
The hit taken by Arctic sea ice last year is enough to absorb a lot of heat from the system, methinks. Wit this year’s extra thin ice poised to possibly fare no better, some reassessment of the dynamics might be necessary. And a decadal average change is just that. Some decades will differ from average. Kim, your tone is getting annoying and does nothing to foster your views. I’m getting to think you deserve as much attention as Watts.
George // March 27, 2008 at 1:10 am |
Meanwhile, while some continue to argue that the IPCC has “overpredicted” the effects of AGW, all hell is breaking loose in Antarctica.
Apparently, Nature does not read blogs.
Alan Woods // March 27, 2008 at 2:39 am |
George, thats not hell breaking loose – its ice. And ice shelves are dynamic systems, so they can break off when it’s cold, or hot, or something in between (for different reasons). And looking at this, I’d say where at something in between:
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.anom.south.jpg
Johan i Kanada // March 27, 2008 at 3:59 am |
George,
Apparently Nature’s Ice does not read IPCC either:
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/global.daily.ice.area.withtrend.jpg
After decades of catastrophic AGW, the global sea ice coverage is above average?
(But this type of discussion is not what Tamino wants to see, so let’s get back to the statistics…)
[Response: Or take it to the open thread.]
dhogaza // March 27, 2008 at 4:42 am |
This one’s studied. 15 years ago, it was predicted that it would break off due to warming within 30 years (by the BAR).
So, who was on record predicting it would break off due to cooling, or something in between, due to being “dynamic systems”?
No one.
Why should we accept the “hypothesis” of an anonymous internet hand-waver over the scientists who’ve been studying the sea-ice-land system there for at least the last 15 years, if not more?
MarkR // March 27, 2008 at 6:15 am |
Tamino says:”we know that the error bars will get bigger — and they’re already big enough to rule out any falsification of IPCC TAR projection.”
What use is a projection that is impossible to falsify?
Steve Bloom // March 27, 2008 at 8:16 am |
Alan Woods, credibility would seem to require that you learn the difference between sea ice and ice shelves. The aforementioned Wikipedia would be a good place to do so.
BTW, the concern about the shelves is not because of the collapse of any one of them, but because the recent collapse is the latest in a series that has tracked from north to south consistent with the record of warming of the Antarctica Peninsula. It has been noted that the Wilkins shelf is 5 degrees south of the Larsen B shelf that collapsed in 2002, and that a further 5 degrees south is the West Antarctic Ice Sheet, the outlet ice streams of which have undergone a massive acceleration in recent years.
kim // March 27, 2008 at 8:59 am |
Sooner or later, most people get around to trying to falsify hypothetical projections. Some sooner. Some later. Some never. You can fool some of the people all of the time.
H/t MarkR
=================================
The Tuatara // March 27, 2008 at 10:08 am |
David BB makes an interesting point. Melting ice absorbs a lot of heat that might otherwise be observed in temperature increases. And the current global melt is substantial. Greenland mass loss, WA mass loss, glacier mass loss, Arctic multi-year ice (and volume) loss. There’s a lot of latent heat of fusion being used up… Perhaps enough to reduce the rate of global average temp increase? (Except of course where it’s released back to the atmosphere on winter freeze-up, viz last winter in Sweden and Finland).
Someone care to do the sums?
guthrie // March 27, 2008 at 11:32 am |
Kim is a well known poster from the mammoth dot.earth thread, who basically trolls merrily along asking annoying questions without showing any ability to learn, nor a deeper understanding of the problems. They undoubtedly have their deeply held belief about global warming being a con, but replying to them will get you caught up in the usual denialist maze.
I don’t think this is too harsh an assessment.
Alan Woods // March 27, 2008 at 11:43 am |
dhogaza, hand waving I can take. But anonymous? I think you might be projecting.
Bloom: I never said that ice shelves were sea ice. Rather, attribution to warming flies in the face of other evidence from the antarctic.
Phil. // March 27, 2008 at 1:02 pm |
“There’s been some hoopla in the blogosphere lately about comparing projections of temperature from the IPCC TAR (third assessment report), published in 2001, to observed temperature. The comparisons have been made to temperature data since 2001, on the basis of the claim that that’s when the projections start so that’s when the comparison should start. It appears such claims are in error.”
How do you square this with the fact that the scenarios for those calculations start from 2001?
http://www.grida.no/climate/ipcc_tar/vol4/english/wg1figts-17.htm
These scenarios were provided to the modellers in 1998 so that the results would be available for the TAR.
[Response: The models don't include, or depend on, nor were they tuned with, any climate data post-1990. Period.]
Phil. // March 27, 2008 at 1:21 pm |
[Response: The models don’t include, or depend on, nor were they tuned with, any climate data post-1990. Period.]
But the projections they make clearly start from 2001, and they do depend on data post 1990 since the starting point for the scenarios is 2001 and was provided in 1998. It’s not credible to assert that in 1998 they ignored changes in emissions since 1990 when devising them.
[Response: Wrong. As Rahmstorf et al. point out, the IPCC TAR projections for CO2 levels (beginning in 1990) turned out to be extremely accurate in spite of the fact that their projections for emissions weren't nearly so accurate, because errors in projected emissions and carbon sinks pretty much cancelled each other out.
And of course you're ignoring the real issue of this post, which is projections of temperature.]
Ken // March 27, 2008 at 1:41 pm |
Alan & Johan – Sea ice extent throughout the south isn’t representative of the substantial warming that’s been occurring on the Antarctic peninsula. That warming is indisputable and explains well the recent break of a (possibly) 1500 yr-old ice sheet.
Tamino – great post. I’m glad you dealt with this topic. I hope the error-bar size evidence helps readers understand the futility of jumping to conclusions based on short-term fluctuations.
George // March 27, 2008 at 1:42 pm |
Mark R says:
I believe HB’s point is that it is not possible to falsify the projection over the short period in this case, but that does not mean it could not be falsified (at least in principle) over a longer term. What that length of time is depends on how the temperature develops.
Some people are sure to cry “foul!” or “convenient excuse!” at having to wait any length of time to say that the IPCC got it wrong, but that does not change the reality.
The comment about the “error bars getting bigger” does not mean “bigger with time” it means doing a more detailed analysis of the noise over the short period would just tend to increase the error bars (not decrease them), so it would not change HB’s conclusion that “the IPCC (average) projection has not (yet) been falsified.”
So, what good is the IPCC projection? (or more precisely “what good are the projections?”, because there are actually many of them for different emissions and climate sensitivity scenarios)
For the very short term, I’d say not much, nor was it (were they) intended to be. It (they) was (were) intended to show policymakers the kind of temperature change to expect over the long (multi-decade) term with a given set of emissions assumptions, so they could plan accordingly.
What good is a (long-term) projection for the “financial health” of Social Security? (ie, telling you when it will probably become insolvent if nothing is done)
Tom C // March 27, 2008 at 2:47 pm |
OK Tamino – I agree that we are not anywhere near a point where the IPCC projections/predictions/scares/whatever can be falsified. But, as others have pointed out, if the aforementioned have any value they must be falsifiable in some way. So, what temperature trend over what time frame would constitute falsification? What is your answer to this?
[Response: I've repeatedly said that there's no fixed time frame which applies. If the actual trend since 2001 is zero and it stays that way (which I seriously doubt), then it won't take very much longer. If the actual trend since 2001 is 0.01 (rather than the average of IPCC TAR projections, about 0.0174) it'll take longer. And as IPCC projections make clear, the trend is expected to increase over the next several decades, and there are many individual projections given in the IPCC TAR so that by 2100 the projections themselves diverge.
It's impossible that IPCC projections are *exactly* correct; sooner or later they *will* be falisified. I just hope real temperature trends don't end up being much *higher* than projected.]
Phil. // March 27, 2008 at 3:21 pm |
“But the projections they make clearly start from 2001, and they do depend on data post 1990 since the starting point for the scenarios is 2001 and was provided in 1998. It’s not credible to assert that in 1998 they ignored changes in emissions since 1990 when devising them.
[Response: Wrong. As Rahmstorf et al. point out, the IPCC TAR projections for CO2 levels (beginning in 1990) turned out to be extremely accurate in spite of the fact that their projections for emissions weren’t nearly so accurate, because errors in projected emissions and carbon sinks pretty much cancelled each other out.
And of course you’re ignoring the real issue of this post, which is projections of temperature.]”
What’s wrong with my statement, nothing you stated refutes it?
Regarding your last comment I’m exactly on point, the projections are based on a starting point of 2001, something you suggested wasn’t the case in the OP:
“How, then, do some conclude that temperature data falsify IPCC TAR projections?
Use the wrong start date,”
Also Fig 17 makes it clear that the starting points are not based on their previous projections.
george // March 27, 2008 at 3:58 pm |
If one is talking about the trend over several years, is it better to trend the monthly data or the annual average data?
What potential advantage (if any) might monthly data provide in that case?
Also, from the perspective of trend analysis what are the advantages/disadvantages of averaging several data sets together before extracting the trend?
Is it better to average the data or to average the trends obtained from each data series and appropriately combine the errors ?
Or does which is better depend on specifics of the case?
For example, what if a particular data set has one or more data points that are in error (particularly systematic error) by a significant amount?
Is it better to do something else entirely (other than average the data sets or average the trends)?
Finally, what does it really mean to talk about “95% confidence” (or any level, for that matter) if it depends on the details of what was done?
No implication is intended with regard to the specific case at hand, but in general if one takes an approach that is just plain wrong, isn’t a “95% confidence” claim simply meaningless?
[Response: Generally it's better to analyze monthly rather than annual data, because there's more information there. However, when using monthly data the autocorrelation becomes much more prominent so it must be taken into account, and if the result is close then one has to use a realistic model for the autocorrelation. In fact, because of autocorrelation the additional information we get out of monthly as opposed to annual data is *much* less than would be the case for white noise, so the advantage isn't nearly so great as in the white-noise case. My intuition -- I haven't quantified this -- is that monthly is as much resolution as I'd want to use for estimating long-term trends (I wouldn't bother to get down to daily or hourly data).
And depending on how long the total time span is, there may be more than enough information in annual averages to answer the questions under consideration. Then the advantage of fewer numbers to crunch comes into play, calculations are faster (this doesn't matter for 30 years of temperature data, but it can make quite a difference for several gigabytes of satellite observations of the large magellanic cloud).
If you want to average data sets together, you have to be sure they're measuring the same thing (so averaging surface thermometer measures with satellite measurements of the lower troposphere isn't a good idea). You'd also want to be sure that the only differences between the data were truly random, and that the behavior of the randomness was the same for both. I'd say it's better to average the trends than to trend the averages. One should be very careful about computing the error of that average, the results may not be independent.
As for "95% confidence" (or whatever level), it should have little dependence on how the analysis is done, and in most cases that's true. But if one analysis is inherently more precise, then it's result should be preferred. But care should be taken when results are a close call. If a hypothesis is rejected at 94.999% confidence but not 95% confidence, we should definitely keep an open mind! Statistics gives probablistic rather than deterministic answers.
And yes, if an approach is just plain wrong, then "95% confidence" is meaningless.]
dhogaza // March 27, 2008 at 4:39 pm |
I’m not certain what Lucia did, but one of her justifications for averaging the multiple sources together was to narrow error bars.
But the HadCRU and GISS output is based on the same set of physical weather instruments, as I understand it, so I don’t see that one could treat them as though they’re independent.
But, then again, I don’t know how she computed her error bars.
chriscolose // March 27, 2008 at 5:13 pm |
Re: Prediction vs. Projection
I would define the latter as saying “if x, then y” (i.e., if we double CO2, we project 3 C of warming) whereas a “prediction” would be telling us that x will happen (i.e., we will get a doubling of CO2, and so we will get a 3 C increase).
chriscolose // March 27, 2008 at 5:30 pm |
By the way, climatological periods (for average weather data) are conventionally 30 years. This is enough time to see if the trend is stationary or if there is some underlying signal. Shorter scales are obviously harder to call because they are so noisy, and unless the signal of the warming exceeds the amplitude of noise in the temperature record, what you are seeing could simply be chance. Moreover, if the variability also increases (the standard deviation may increase along with the shift of the mean) then you can get temperatures just as cold as before the warming. It’s also of importance to risk assessment people to know the likelihood of extremes.
I’m in agreement with Tamino that we need to wait a while, maybe almost 10 years, to get a good idea of the trend. 2008 for example will probably not be very hot (meaning probably not in the top 5, but still above the 1951-80 baseline) because the first two months have been anomalously cool due to La Nina. If the rest of the year goes on like, say, 2005 then it is possible, but I don’t think very likely.
C
kim // March 27, 2008 at 6:51 pm |
Let’s try this. Forget for the time being 95% confidence, and TAR. Using your method, Tamino, and AR4 projections, and data from 2001 to the present, at what confidence level have those projections been falsified, already.
guthrie, more precision in language, please. Don’t you mean I ask irritating questions, rather than ennuying ones?
================================
steven mosher // March 27, 2008 at 7:26 pm |
Tamino, you wrote
“[Response: This continual misrepresentation of IPCC projections is getting really annoying.
The same exact models which average to 0.174 deg.C from 2000 to 2010, also indicate 2 to 4.5 deg.C this century.]”
This requires some clarification. you are talking about a spread due to changes in input. SRES.
Basically, here is what you have. In the TAR they said .2C per decade, NO MATTER what happens to emissions. Warming in the pipeline. In AR4 they said the same thing. 2000-2011 .2C NO MATTER what happens to emissions. Warming in the pipeline.
Basically, since the TAR the IPCC has said .2C per decade for the next couple of decades no matter what we do. shrugs.
Here’s a thought. Put proper errors bars on all the projections and you will tell a vastly different story. But you will not tell a story that motivates people to action and change because that story will be dominated by uncertainity.
In any case. I have decided to switch parties. I will now side with the people who want to wait a long time to confirm or disconfirm climate science. As many have noted we need thirty years of data. So, I dont want to hear any short term confirmations or disconfirmations.
no stories about ice and cold winters, or heat waves. 30 years. That’s about right. Then I will listen to the debates.
After 30 years, we can decide the matter. When all the data is in. Until then, everything we talk about is just the weather.
Seems fair.
[Response: I haven't checked AR4 for details, but in the TAR they say "about" 0.2. One significant digit. Only those who are looking for an excuse to falsify IPCC will insist that it's 0.2 on the nose. And they also say that's for the next several decades. But the trend is expected to increase over those decades, so it'd be less than that early and more than that late. And it's based on model simulations, which when studied to get numbers more precise than one significant digit, give less than 0.2 for this, and the preceding, decade. What part do you not get?
By the way -- have you retracted the mistaken claim about my suggested conditions for falsification, on those blogs where you cross-posted it? Seems fair.]
dhogaza // March 27, 2008 at 8:34 pm |
Only in the near term. It’s not that hard to understand.
It’s no different than the stock market, where you can quite confidently point out that money invested when you are young will have grown in value (in real dollar terms) by the time you reach your 60s, if past trends hold up.
Yet, we can’t promise that an investment made when you’re 21 will have grown by the time you’re 27.
Conservatives have no problem with that. Why the hair up the ass over a physical phenomena that is similarly noisy over the short term, but with a well-defined long-term trend?
Joel Shore // March 27, 2008 at 8:39 pm |
As I understood it, the IPCC uses the term “projections” rather than “predictions” because they are based on scenarios for what our future emissions might be. Although they have tried to make the scenarios realistic, one cannot predict what future course society will take and, most importantly, one might hope that the emissions scenarios don’t come to pass because we actually come to our senses and curtail our emissions. (The IPCC purposely made the scenarios essentially “business as usual”, i.e., not assuming that we realize our folly and curtail emissions.)
Another contributing factor might be that the IPCC does not attempt to account for natural variability due to natural forcings likechanges in the solar and volcanic forcings. This is justified for two reasons, one being that it is not now possible to predict these future natural events and hence the forcings, and the second being that the anthropogenic forcings are expected to dominate over the natural forcings barring some dramatic low-probability natural event such as a super-volcano or a major asteroid impact.
Hank Roberts // March 27, 2008 at 11:24 pm |
For those interested in how trends are perceived:
http://online.wsj.com/public/resources/documents/info-flash08.html?project=LOST_DECADE
http://www.wallstreetweather.net/2008/03/stocks-lost-decade-from-planetary.html
Hank Roberts // March 27, 2008 at 11:25 pm |
(And no, I am not endorsing the planetary-alignment theory of Wall Street, just pointing out how an astrologer is interpreting the trends so clearly pictured in an article from the WSJ)
George // March 27, 2008 at 11:58 pm |
Steven Mosher claims:
“In AR4 they said the same thing. 2000-2011 .2C NO MATTER what happens to emissions. Warming in the pipeline.”
You have misrepresented what the IPCC said, just as you previously misrepresented what HB said with regard to his previous post.
From AR4, Section 3
kim // March 28, 2008 at 3:42 am |
So let’s take AR4’s ‘about 0.2 degrees C per decade’, and the temperature record from 2001 to the present, and see at what confidence level that hypothesis is falsified, now.
C’mon, I dare you. You can do it again yearly if it makes you happy.
=======================
dhogaza // March 28, 2008 at 5:08 am |
Oh, Kim reads Climate Audit, and has read Lucia’s response that HB is wrong, because he used the TAR, not AR4, while she ignores the statistical analysis posted here!
Kim smart! Kim knows how to snip shit from CA and post it here (without attribution!). Kim wins, because he knows where the and keys are on his keyboard!
Kim, are you Lucia’s hero? Is this why she’s not posted her disagreements with HB here, rather than at CA, where she knows HB won’t respond?
Gavin's Pussycat // March 28, 2008 at 6:15 am |
Steve Mosher, do you ever learn anything from what has been pointed out to you in the past?
Like, what happened over the past 30 years?
Or is denialism like Alzheimers, and you have to explain everything from scratch again every time?
Gary Moran // March 28, 2008 at 9:00 am |
Tamino
RankExploits was testing AR4 since 2001 where the prediction was .2′C per decade, not TAR since 1990. So you’ve spent a lot of time and effort countering an argument they didn’t make!! Was that a deliberate ploy to cloud the issue?
fred // March 28, 2008 at 9:11 am |
HB, Lucia apparently was assessing not TAR but AR4. Does that change your mind about any of this? It evidently will not change your mind about the excessive shortness of the sampling period, understood. But some of the other surrounding stuff?
[Response: What's to change? The observed data since 2001 don't change, the size of the error range doesn't change, the error range still includes 0.2 deg.C/decade for both GISS and HadCRU, and it's still misrepresentative to take "about 0.2" and interpret it as an ironclad exact 0.2.
When I took up this subject, it was obvious that a 1-significant-digit statement complete with approximation qualifier, was not sufficient to characterize IPCC projections. So I tried to do better -- and it took me *less than a day*.]
kim // March 28, 2008 at 12:00 pm |
dhogaza, if you’d be a little smarter and look back through this thread, you’ll see that I’ve been yammering about AR4 since long before lucia posted that comment at CA. However, I did get it from her original posts at Blackboard.
Yes, I’m smart enough to understand that CO2’s role in climate is not precisely elucidated. And that the sun’s role in climate is not fully explicated, either. Frankly, I’m seeking understanding; I actually get some here.
=============================
kim // March 28, 2008 at 12:13 pm |
Truly, dhogaza, I’ve no idea why lucia doesn’t post here, or Tamino not at climateaudit. But it is obvious to me why you don’t post where you’d get eaten alive. With your attitude and rhetoric, I’ve wondered why you post anywhere. Have you no shame?
Be a scientist. Step back and look at the temperature data, and wonder, at long last, if your hypotheses might be in error.
================================
george // March 28, 2008 at 12:14 pm |
Lucia gives the following trend and confidence interval for the trend since 2001
Lucia also claims that
Compare that to the central value and range of warming for scenario B1 in the AR4 WG-1 report [in bold] (Chapter 10, executive summary, page 749)(see quoted text below)
In particular, note that the lower bound on the IPCC range for scenario B1 (1.1C/century) is the upper bound on Lucia’s trend based on the data since 2001.
In other words, even if one assumes that Lucia is justified in starting her analysis in 2001 and in using such a short period for comparison to IPCC projections and that her statistical analysis is correct (the uncertainty for the trend that she gives is actually narrower than the one HB arrived at for the same period for either GISS or HadCRU [the latter of which actually has the greatest negative slope since 2001 of ANY of the data sets]), based strictly on what is in the AR4, she has not “Falsified IPCC” , specifically not not the IPCC B1 projection at the claimed “95% level” (though it would be close to doing so)
Relevant text from the same AR4 document
AR4 WG-1 report [in bold] (Chapter 10, executive summary, page 749:
The meaning of the range attached to the scenarios is shown in Figure 10.29. (page 809 of the same document)
Harold Brooks // March 28, 2008 at 12:57 pm |
So let’s take AR4’s ‘about 0.2 degrees C per decade’, and the temperature record from 2001 to the present, and see at what confidence level that hypothesis is falsified, now.
If you take the monthly values from GISS for 2001-2007 and ignore the autocorrelation (including it would make the confidence level even lower), I get a little less than 75% for the 0.174/decade that Tamino gave for post-2000. For the annual values, the slope of the regression is 0.188/decade with a 95% CI of -.1 to .5/decade.
george // March 28, 2008 at 1:12 pm |
I neglected to provide the link to the relevant AR4 document above — the one that falsifies the “IPCC Falsified” claim.
kim // March 28, 2008 at 1:17 pm |
Drop back five, and punt. A bettor would be glad for those odds, and thank you, Harold, for your work.
========================
kim // March 28, 2008 at 1:40 pm |
It’s already falsified, george, but not at the 95% confidence level. However, we are in a cooling PDO phase, and with a quiescent sun. Every day the earth chills gets us closer to the 95% level.
It might behoove you to examine your assumptions and hypotheses about carbon dioxide. It is an important question, extremely so. Much money and many lives depend on the correct answers.
==============================
steven mosher // March 28, 2008 at 1:45 pm |
Tamino,
I apologize for the misrepresentation. I’ll post that at Lucias and you can link to it.
Also, Lucia’s analysis is NOT about the TAR, it’s about AR4. Read her most recent post for the clarification. Or look at the figures in her first post which clearly show her analysis versus AR4. so perhaps, we both can make mistakes and correct them.
[Response: I was mistaken believing that Lucia's analysis applied to TAR rather than AR4 projections.]
Dano // March 28, 2008 at 1:55 pm |
Be a scientist. Step back and look at the temperature data, and wonder, at long last, if your hypotheses might be in error.
Ah.
Do your own science if you don’t like the science that goes against your ideology.
Perfect.
Best,
D
steven mosher // March 28, 2008 at 1:58 pm |
Apology posted. AT lucia’s. The thread is called the tetter totter of temperature
george // March 28, 2008 at 2:41 pm |
kim says:
“It’s already falsified, george, but not at the 95% confidence level.”
Context, dearest Kim. Context,
Lucia (remember her?) was claiming “IPCC Falsification at the 95% confidence level”” not “falsification” at some other confidence level and that is what I was addressing, in my very last post and in the more detailed one that immediately preceded it.
That should have been obvious, but I apologize if it was not.
I neglected the “at the 95% confidence level” in my second post but included it in my longer post above:
With regard to what level the IPCC projections have actually been “falsified” at, I would first note that it is important to consider the error bar about the IPCC trends as well as about the temperature data. As I indicated, Lucia’s claimed “95% confidence” interval actually includes the lower value for the “95%” range given by IPCC for scenario B1 (1.1C)
I can not be sure where Lucia actually got her uncertainty range for the IPCC projected trends (from an “eyeballing” of the IPCC scenario projection graphic?), the one she used to claim
But the IPCC lays out the uncertainties for individual scenarios quite clearly in the text. (And even those uncertainties were intended for the full time period, not just a few short years and thus would not include the probable effect of short term noise like El Nino)
Second, and perhaps most important of all, at what confidence level is a “falsification” no longer a “falsification”?
95% is the level normally given for scientific purposes because something that is “rejected” at a lower confidence level may just be rejected too often due to chance.
The fact is, any way you look at it, Lucia was just plain wrong with her claim that “IPCC has been falsified at the 95% confidence level”, which is what started this whole broo-ha-ha.
george // March 28, 2008 at 2:56 pm |
Steve Mosher
…and, as explained in my comment above, according to the very clear text of the AR4 (For those who care to actually read it, see page 749 of WGI Chapter 10 , particularly as it pertains to scenario B1) her claim of “IPCC Falsification at the 95% confidence level” is just wrong.
steven mosher // March 28, 2008 at 3:02 pm |
Dialog is preferable to debate. When dialog is shut down debate ensues. When debate is shut down, diatribes develop. Then you’ll see real division. Two tribes talking past each other.
And power will define truth.
The best way to undo this is to return to dialog.
Dont feed hemlock to people who ask questions.
steven mosher // March 28, 2008 at 3:17 pm |
Hi Kim!
We both seem to read the same places. It’s funny that bloggers don’t engage each other on the same ice rink, so to speak. ATMOZ does a bit of this to his credit. For the most part it’s dialog at a distance. he said, she said. Which is very odd, but I understand the social dynamics very well.
When the truth has been politicized then admitting error is social death. Witness, for example, how people hang on to the Christy/Spencer error. Witness how they hang onto the Mann errors. Witness how people attack and defend AIT. To make an error in this climate is deadly
george // March 28, 2008 at 3:21 pm |
One more general comment and I am done (wasting my own time :) )
This whole thing illustrates perfectly what is most wrong with “science by blog”.
You get people making claims right and left when 95% of the time (95% confidence?) they have not even thoroughly read what they are “critiquing” — or if they have, they have not thoroughly understood it.
But some people seem to have the attitude “What the hey, if I am wrong about my claim, we will all learn something”.
Well, no.
They themselves might “learn something” (if the rest of us are lucky), but it is unlikely that “we” (the rest of us) will.
It’s kinda like saying “we have all learned something”, every time a spacecraft burns up or crashes due to a stupid metric/English conversion.
In the case of the incinerated spacecraft, ” ‘We’ have learned something” all right, but probably not what the one who made the mistake that led to the burnup thinks.
cce // March 28, 2008 at 5:28 pm |
This is what I would like to see.
Take the various temperature analyses and remove the influence of volcanic eruptions as was done in the “many factors” post. Calculate the rate of warming for the entire time period. Then apply the same statistical tests as Lucia has done to every 7 year period since 1975 (or 1979 in the case of the satellites). My guess is that there will be several 7 year periods that would, by the same standard, “falsify” the warming we know for a fact has occurred.
dhogaza // March 28, 2008 at 5:28 pm |
Well, usually Mosher angers me, but this time, he made me laugh hilariously.
Such an ironic line coming from an avowed libertarian …
steven mosher // March 28, 2008 at 11:39 pm |
Dhog,
Seriously, have I not made you laugh before? Seriously?
steven mosher // March 28, 2008 at 11:48 pm |
CCE has an interesting approach. Early on I asked Lucia if she could factor out Volcanoes.
And I pointed her at something Tammy had done ( cant recall where now, alzheimers again)
Atmoz also had something interesting … taking out enso ..
Much of this seems similiar to adjusting climate records for surface stations. similiar, but different.
basically, one could get a better estimate of the underlying trend by factoring out volcanic forcings and major weather periodic functions.
Sounds good in words. Probably a bitch in math.
steven mosher // March 29, 2008 at 12:58 am |
Ok, here is something everybody should like.
It’s work safe. No climate denialism. Just a fun demonstration.
http://www.youtube.com/watch?v=yldmXWu745w
hat Tip to Hans
Marion Delgado // March 29, 2008 at 2:25 am |
Where’s the profit in replying to the kims and TCOs? They ignore your replies and simply spam the same comments over and over. It’s like trying to argue science with parrots.
Once you’ve pointed out that there is too much noise determining the fit of models to data over very short time periods (especially cherry-picked short intervals a decade or two after the models for long periods were created), and the trolls make it abundantly clear they’re not going to address that, but simply re-spam their Inhofe speeches, that should end it, I hope.
steven mosher // March 29, 2008 at 2:44 am |
Kim is an aphorist of sorts. The comments Kim makes are not meant for you to argue with. They are meant to rattle your brain, to refocus your gaze.
You should not agree or disagree with Kim. You listen. If it strikes a chord you resonate.
Sometimes I resonate with Kim. This is good.
Other times not. I shrug and focus on those times when we see the same thing the same way.
I like Kim.
steven mosher // March 29, 2008 at 2:54 am |
Marion,
If you would actually engage TCO in a structured discussion ( point by point) you would find, as I have, that he is a fair but tough minded person. He will never cut you any slack.
and he might punch you in the nose when you dont pay attention, but you’ll probably learn more than you teach.
dhogaza // March 29, 2008 at 3:30 am |
I know a really good free, public, county-supported psyche clinic folks can visit here in PDX if, by any chance, this happens to you and you’re in the ‘hood.
Early diagnosis and aggressive treatment can’t hurt in such cases.
cce // March 29, 2008 at 6:35 am |
I don’t think the weather should be filtered out. Just the volcanoes, which are truly external. If the past (since 1975/1979) gives us periods (7 years) of non-warming equal to or more significant than the current non-warming, then it’s a bit hard to falsify model results using the same criteria.
steven mosher // March 29, 2008 at 1:02 pm |
dhog, now you made me laugh. we are even
steven mosher // March 29, 2008 at 1:07 pm |
cce, a while back I made a complilation of all 7 year trends from 1850 to today. If I recall the probabilty of randomly selctiong a 7 year time peroid and finding a negative trend was around .38. They have of course become more rare of late. based on niave stats, you have a 50/50 chance it will last until 10 years
George // March 29, 2008 at 2:52 pm |
cce says
Leaving the effects of weather in for the case at hand is the very point, of course because that is what tends to change the trend most significantly over short periods.
The only way to really compare the observed temperature trend over a very short period to a projected trend over the same short period is to include the probable effects of weather in the projected trend (or at the very least in the error bands) for the short term.
The IPCC uncertainties for the projected trends for their scenarios (TAR and AR4) do not reflect the effects of noise like that due to El Nino over the short term. They do not include weather induced uncertainty over the short term because they were not intended to accurately show what would happen over a short term like 7 years.
Anyone who wants to see a projection that takes into account the effects of short term noise like El Nino and volcanoes need only look at Hansen’s 88 projections. They go up and down about an upward-sloping “ramp” imposed by GHG increase. No one in his/her right mind would look at the IPCC projections and think they were intended to show the same thing as Hansen’s projections!
Comparing the trend from the temperature data over a very brief period (7years) with an “adjusted” (scaled for the short period in question) trend and error band intended to reflect what would happen over the long (multi-decade or even century) scale is simply folly.
If it shows anything at all, it is that one did not read the IPCC document carefully enough to actually understand what their projections mean.
As a former teacher, I can say that if one of my students had done this, they would have received a failing grade no matter how good their “analysis” was.
Timothy Chase // March 29, 2008 at 7:35 pm |
Kim wrote:
Not meaning to nit-pick, at least Kim in particular since I believe a number of people have been making the same mistake, but strictly speaking, you cannot “falsify” something to less than a 100% confidence level. A proposition or theory which is falsified receives falsification at a 100% confidence level. Anything less than 100% is disconfirmation, which means that the proposition (the first form in which Karl Popper proposed his theory) or theory (the second form in which Karl Popper proposed his theory) is still in play as something which is potentially true — although given his skepticism, whether or not it is true would be forever unknowable.
This, incidentally, is closely related to the reason why “falsifiability” belongs to the history of the philosophy of science rather than the philosophy of science — and has since about the 1950s. For those who are interested in learning more, you might want to an earlier post of mine at Real Climate. Incidentally, the criticism isn’t new — as it actually predates Karl Popper’s approach (1930s and 1940s) with Pierre Duhem (1892). As such, both the philosophy of science and science itself has moved on to a more Bayesian approach — and practitioners of climatology are perhaps generally more self-conscious of this than most disciplines.
Unbver // March 29, 2008 at 8:09 pm |
CCE asks about seven year trends since 1975. It is very easy to find the answer to his question, for example just put the annual averages into Excel and use the @Slope function.
During that time two major volcanoes interrupted the general warming trend, the 1982 El Chinchon and the 1991 Pinatubo. In the years following those eruptions there was a drop in temperature that caused the seven year trend to be negative.
Apart from those two volcano-influenced dips the seven-year trends were all positive until recently. So it does not appear correct to say that in the last 33 years there has been a non-volcano-influenced period of seven years with a negative trend before the present.
kim // March 29, 2008 at 8:12 pm |
I’ll happily accept disconfirmation instead of falsification, and thank you for the important lesson, TC.
Every day of chilling is disconfirming the hypotheses in the IPCC. Surely, real scientists must wonder about many of the assumptions underlying the proposition that CO2 has a larger effect on climate than the sun does, and atmospheric and oceanic circulations do.
================================
Timothy Chase // March 29, 2008 at 9:19 pm |
Kim wrote:
A little more to the point of your post…
This is like saying that the “hypothesis” of global warming becomes “increasingly falsified” or “disconfirmed” when we head into winter because things are getting colder but becomes increasingly confirmed as one heads into summer.
That’s not what the “thesis” of “global warming” is about. It is about the overall climate trend, and just as one would “detrend” for intrannual variability (remove the seasonal signal) one would detrend for the cool phase in the solar cycle or for the negative phase of El Nino or the Pacific Decadal Oscillation.
Thus, for example, the fact that 1998 is statistically tied with 2005 and 2007 for the warmest year in modern history should be discounted to some extent given the fact that it occured during a particularly strong El Nino and warm solar year, whereas one should given both 2005 and 2007 greater weight as 2005 was during a cool solar year and did not have the advantage of an El Nino, and 2007 was not simply during a cool solar year, but for the majority of the year was subject to a strong La Nina.
But one detrends for such quasi-cyclical behavior unless one were specifically interested in the dependence of the trend upon the cyclical phenomena or otherwise investigating their causal connections — assuming one thought that they were so connected.
… which actually seems to be the case in a rather interesting fashion:
During periods of global warming, ENSO tends to be in a positive phase (more El Ninos which are stronger and last longer) and the North Atlantic Oscillation and tends to be in a positive phase. The reason being? The system is chaotic and sensitive to its environment, particularly the forcing, whether it is solar in origin or due to greenhouse gases. But in the troposphere at least there is very little difference between the two and how they affect the climate system. However, with warming due to solar forcing, one expects both the troposphere and the stratosphere to warm (with the ultraviolet of increased sunlight being absorbed in the stratosphere by ozone), whereas with greenhouse gases one expects the stratosphere to cool while the troposphere warms — as the increased opacity of the atmosphere below the tropopause to infrared radiation reduces the thermal radiation reaching the stratosphere. A guess which way the trends have been going since 1979…
Not sure about the Pacific Decadal Oscillation (it is one of the longer period climate modes with a shallower amplitude than most — so I haven’t given it as much attention), but the La Nina has been weakening of recent, although we probably won’t see the beginning of the next El Nino until late next year. With regard to solar variability, we have seen beginning of the new solar cycle — with the first few sunspots showing up earlier this year.
I believe warmer times lie ahead — though this particular summer won’t be a real record-breaker.
dhogaza // March 29, 2008 at 9:21 pm |
Uh, Kim, you did notice that 2007, despite the strong onset of La Niña at the end of the year, was still the 7th warmest on record (IIRC), right?
And that 2008 will probably be similar, right?
kim // March 29, 2008 at 9:40 pm |
I understand your point, TC, but it is all dancing around the increasingly likely fact that the effect of CO2 on a warming trend has probably been exaggerated. If we don’t even know what natural variability is, how are we supposed to dissect out the effect of CO2? This is an important question, and it is not being asked, and certainly not being answered.
dhogaza, 7th warmest doesn’t sound like climate catastrophe to me.
==========================
Timothy Chase // March 29, 2008 at 10:27 pm |
kim wrote:
Actually, the question has been asked and repeatedly answered — albeit tentatively — by numerous studies.
See for example:
Annan, J. D., and J. C. Hargreaves (2006), Using multiple observationally-based constraints to estimate climate sensitivity, Geophys. Res. Lett., 33, L06704, doi:10.1029/2005GL025259.
Royer DL, Berner RA, Park J. (2007), Climate sensitivity constrained by CO2 concentrations over the past 420 million years. Nature, 446: 530-532.
It appears that climate sensitivity to a doubling of carbon dioxide is roughly 2.8 C — and has been for nearly half a million years. (It is partly a function of the distribution of the continents, and as such would have been different at different times in the earth’s history.) 2.8 C, give or take, but with something decidedly higher being more likely than something decidedly lower.
Oh — and has been pointed out recently by a number of us, other than the quasi-cyclical behavior, solar irradiance has been more or less flat to slightly declining pretty much since 1951. Not much of an alternative explanation for the modern period of global warming there I am afraid…
kim // March 29, 2008 at 10:59 pm |
Thanks, TC, but with temperatures dropping, and CO2 rising, James should re-visit some of his admittedly subjective assumptions.
=============================
kim // March 29, 2008 at 11:01 pm |
Also, it seems clear, that with TSI relatively constant, there must be a multiplier in the link between the sun and the climate. This mechanism, multiplier included, is what is presently unknown.
=================================
Timothy Chase // March 30, 2008 at 12:14 am |
kim wrote:
Well, I suppose January was rather cold…
kim wrote:
Please do let us know how the experiments turn out!
luminous beauty // March 30, 2008 at 12:18 am |
Kim,
The result of multiplying a trend with zero slope by an imaginary multiplier from an imaginary mechanism is zero.
George // March 30, 2008 at 12:38 am |
If there is a “multiplier,” (ie, feedback) it is most likely the same one that amplifies the temperature increase of a given atmospheric CO2 concentration increase — ie, water vapor increase.
The idea that there is some “separate multiplier” that only applies to solar irradiance change is stretching the limits of credulity.
Besides, if solar irradiance does not go up (or actually goes down) it does not matter how big the “multiplier” is, it is still not going to lead to an increase in temperature.
cce // March 30, 2008 at 12:54 am |
The answer to my question involves removing the influence of volcanoes, which is not so easy. It’s unlikely that the cool period between 1983 and 1987 can be entirely attributed to El Chichon. The aerosol levels had recovered by the end of 1985.
kim // March 30, 2008 at 1:26 am |
TC, please stop confusing weather and climate. The reliable thermometers, UAH and RSS, as well as the oceanic bouys have dropping temperatures for several years now.
Lum, there is correlation between relatively spotless times in the sun and cold weather. Also, the slope is not zero. If you don’t look, you’ll never find.
George, water vapor turns readily to clouds, and back. How water in the atmosphere works to determine climate is largely unknown, despite the false security inherent in the models. You can see yourself, that clouds at night warm the surface, clouds in the day keep it from warming.
I insist, the sun’s activities correlate well with climate. The mechanism(s) is unknown.
===============================
Hank Roberts // March 30, 2008 at 2:12 am |
>I insist
Why witness when you can look it up?
kim // March 30, 2008 at 2:24 am |
Don’t guess, Hank; look it up.
====================
steven mosher // March 30, 2008 at 2:55 am |
Thanks Timothy, I tried to suggest to Lucia, that she say disconfirmation rather than falsfication, but the verificationist and Popperian language seems very entrenched.
Now, I just shrug. people say falisify, they mean disconfirm, they just dont know it.
Ian // March 30, 2008 at 2:58 am |
kim,
Some of us think we have looked it up, many times, and we still don’t know what you’re talking about – can’t you give us just a little hint?
Timothy Chase // March 30, 2008 at 3:36 am |
kim wrote:
Do you mean that temperatures have been dropping since 2005? The year that was statistically tied with 1998 for the dubious honor of “Warmest Year”? (Hadley places 1998 higher, NASA puts 2005 higher, but both agree that they were very close) Well, given the climate system’s internal variability (with ocean heat content sinking due to weather patterns during La Ninas but rising to and spreading out along the surface during El Ninos), one can’t expect every year to be a record-breaker. But they have been close.
As I stated earlier:
Incidentally, if you need a fairly basic explanation of what an El Nino is, you might try:
*
kim wrote:
I know you presently prefer the satellites over landbased measurement, but Tamino has previously shown that despite the fact that satellites are measuring lower troposphere and and “surface” measurements are at roughly two meters (or at the surface itself in the case of water), the trends are in close agreement.
Please see:
What’s Up With That?
March 2, 2008
http://tamino.wordpress.com/2008/03/02/whats-up-with-that/
*
kim wrote:
Ocean Buoys? Several years? Are you thinking of the ARGO floats?
If so, please see:
… and it is quite clear from the following that there has been no statistically significant cooling since 1998 (NASA GISS has actually been showing a statistically significant warming trend), 2000 (still warming), or cherry-picked 2001…
You Bet!
January 31, 2008
http://tamino.wordpress.com/2008/01/31/you-bet/
Timothy Chase // March 30, 2008 at 3:55 am |
steven mosher wrote:
I am afraid “falsified” will always be a big pet peeve of mine. Well, actually it will be worse than that.
It is wrapped up in the view that scientific theories aren’t a form of knowledge as in Karl Popper’s view, theories aren’t capable of receiving justification. But the body of evidence that supports Quantum Mechanics, Special Relativity and General Relativity is vast. Scientific theories are in fact capable of being the most precise most strongly justified form of empirical knowledge available to us.
They are capable of receiving justification, a great deal of justification, but as with nearly everything we know, that justification isn’t absolute certainty, but a matter of degrees. And it may very well be the case that some of the most well-supported and exact theories will later prove to be approximations. But even approximations are a form of knowledge — within the bounds of their accuracy.
Games of cartesian doubt aside, scientific knowledge is knowledge.
kim // March 30, 2008 at 3:56 am |
Ian, sunspots were sparse or absent during the Grand Minima.
TC, temperatures have been dropping for several years. Minimally so, but truly. See reallyrealclimate.blogspot.com for a sample. You are simply out of date with the ARGO data.
Look, the warming at the end of the last century corresponded to an El Nino dominant PDO lasting around 30 years; look at all the El Ninos during that time. Now that the PDO has flipped, and La Ninas will predominate, we’ll probably cool, as is happening.
None of this is certain, but it is more likely than blaming CO2 for the temperature rise that is no more.
==========================
Ian // March 30, 2008 at 4:21 am |
kim, okay, I’ll buy a connection between the Maunder-era minimum and climate, but I don’t see the evidence for a connection with TSI/sunspots in the last 30-40 years.
dhogaza // March 30, 2008 at 4:52 am |
Why is ARGO magic, and other sources of data … “ignored”?
And why does ARGO say the sea is getting cooler, while other data says the sea is expanding in volume?
If ARGO showed that IPCC estimates were off in the other direction, would you love them? No, of course not.
Chris O'Neill // March 30, 2008 at 5:58 am |
kim:
Garbage. Until January this year, global temperatures were not dropping. If you want cherry-picking extraordinaire (January 1998 El Nino to January 2008 La Nina) then go to reallyrealclimate.blogspot.com .
We can only blame CO2 rise over the last 10 years for about 0.2 deg C. No-one disagrees that a big El Nino month minus an El Nina month will easily beat that. Trouble is, CO2 rise will add another 0.2 deg C every decade but perhaps you believe that we will experience no more El Ninos and perpetually stronger La Ninas as the decades go by.
Hank Roberts // March 30, 2008 at 6:19 am |
> Minimally so, but truly.
http://tamino.wordpress.com/2008/01/31/you-bet/
Unbver // March 30, 2008 at 2:00 pm |
CCE, I think you will find that orthodox climate science explains the dips in the temperature and slight cooling trends in the period from 1983 to 1985 and 1992 to 1995 on volcanic activity.
Of course it always possible that random weather “noise” was also a factor at that time, but Ockham’s Razor tends to favour the explanation of the volcanoes. This article from 1991 predicts cooling following the eruption of Pinatubo, and says that Chichon caused cooling from 1982 to 1986. http://query.nytimes.com/gst/fullpage.html?res=9D0CE2DB153EF933A05755C0A967958260&sec=&spon=&pagewanted=all
George // March 30, 2008 at 3:28 pm |
I would like to point out some significant differences between HB’s results and conclusions (above) based on the and
those of Lucia for the GISS and HadCRU trends since 2001 and ask some questions of HB related to that.
Here’s HB’s results and conclusions from above:
Here’s Lucia’s results from the table at the above link:
HB
I wonder if you would care to comment on the difference between the results you got
using
and the result Lucia got with Cochrane-Orcutt.
It is at least somewhat reassuring that the trend slopes obtained with OLS come out the same using your method and hers, but that’s where the similarity (and reassurance) ends!
In particular, why do the trends change so much with Cochrane-Orcutt (from OLS)? And why are the error bars obtained with your OLS (”exact”) method significantly greater (by about 50%) than the ones that Lucia got with Cochrane-Orcutt?
Finally, I wonder if you might speculate (don’t have to if you don’t want to, of course) on how the error that Lucia found for HadCrU (and the other data sets as well) could turn out to be virtually identical for the OLS and Cochrane Orcutt methods.
From what you indicated previously, I would have expected that Cochrane-Orcutt would tend to increase the probable error.
[Response: First, just so everyone's clear, the trend doesn't change -- but we don't actually know what the trend is. Our *estimate* of the trend changes.
Cochrane-Orcutt and OLS are different estimation methods, so they will necessarily give different estimates. That difference should be smaller than, but of the same order of magnitude as, the uncertainty in the trend. And that's what we observe; no mystery there.
My OLS errors are significantly bigger because Lucia uses an AR(1) model for the autocorrelation structure, but I use a more realistic model. The autocorrelations at lag greater than 1 are larger than their AR(1) values (a fact pointed out in reader comments even before I mentioned it), which has the effect of increasing the uncertainty in the trend estimate. And that's what we observe. As you can see, the effect is nontrivial.
Cochrane-Orcutt will indeed increase the error range over the OLS estimate using a *white-noise* error model. But Lucia bases her OLS estimate on an AR(1) error model -- which is exactly the same error model inherent in Cochrane-Orcutt. So it's no surprise they give comparable error ranges.]
Timothy Chase // March 30, 2008 at 5:17 pm |
kim wrote:
Sure — I had a brief exchange with Tilo Reber about the potential for cooling of the Southern Ocean over at Real Climate a little while ago.
At one point I stated:
… then quoted:
Tilo Reber owns reallyrealclimate.blogspot.com, so I found a way of commenting on it without actually mentioning it:
The bit that begins “one would be demonstrating” — that’s where I am talking about his blog.
*
Kim wrote:
Even if there were a slight cooling of the Southern Ocean over a 50-year period, this would be consistent with the model by Bryan and Manabe mentioned above. Ocean mixing — with deeper, cooler waters coming into play. But by the conservation of energy, the heat content goes somewhere. And deeper circulation of heat content actually implies a higher climate sensitivity, not a smaller one as it takes longer for the radiation imbalance at the top of the atmosphere to reach equilibrium.
*
Kim wrote:
I decided to look up the Pacific Decadal Oscillation. There is a good graphical representation of its distribution over here where you can see it compared to the El Nino here:
The Pacific Decadal Oscillation
http://cses.washington.edu/cig/pnwc/aboutpdo.shtml
They look quite similar, actually, except that the range over which the ENSO dominates is over a large swath of the tropics where PDO is weak, and the much smaller region over which the PDO dominates is in the Northern Pacific, where ENSO is weak.
In the article, they state:
They also state:
As such, it would appear that even though PDO may have gone negative in the late 1990s, we still do not know whether it has made a twenty or thirty year flip to the cool side — as there may be short-term flips on either side where it briefly reverses its polarity. They refer to another page where it states:
It appears that the Pacific Decadal Oscillation is big – at least if you live in the Pacific Northwest. Going to the data, I see that the most recent time that the index was positive was June through August of 2007, before that April of 2007, before that December 2006 through February 2007, January 2005 through July 2006, August 2002 through September 2004. It would appear a bit indecisive at the moment, and I would suggest that it is at best quasi-periodic.
On the above webpage, they state:
Seems like it could be chaotic. Your thirty-year period seems rather debatable — if you intend to suggest anything resembling clockwork behavior.
But does it “control” ENSO?
Well, the best I could find was:
However, looking at the left-bottom diagram on slide 10 of:
ENSO-forced variability of the Pacific Decadal Oscillation
Matt Newman, NOAA-CIRES CDC
http://www.cpc.noaa.gov/products/outreach/proceedings/cdw28_proceedings/mnewman_2003.ppt
… it would appear that ENSO does a rather good job of explaining the behavior of the PDO oscillation rather than the reverse. Then again, given the proximity of its center to the Arctic, one might expect a stronger teleconnection to either the North Atlantic Oscillation or possibly the Arctic Oscillation — but looking at the NAO index (see the chart showing 1860-2000)…
Welcome to the North Atlantic Oscillation www-page
http://www.ldeo.columbia.edu/res/pi/NAO/
… and the PDO index (see the chart showing 1900-January 2008)…
Pacific Decadal Oscillation (PDO)
http://www.jisao.washington.edu/pdo
… the correlation doesn’t seem as great.
So we seem to be left with ENSO, and as I stated:
*
Kim wrote:
Think about: we know that greenhouse gases are opaque to thermal radiation. We can measure their absorption spectra, the backradiation that they emit, and even their concentrations at various temperatures, pressures, and altitudes on account of this. Raising the opacity of the atmosphere means less radiation leaves the climate system, but if the same amount of sunlight enters the climate system, conservation of energy means that the heat content of the climate system has got to go up.
Now climate oscillations aren’t magic. They don’t cause energy to disappear or reappear. What they do is move this heat content around. Warm salty water sinks, cool fresh water rises — and perhaps it will have something to do with the trade winds. But the heat is still there. And until the surface temperature rises to a sufficient degree for the thermal radiation it emits to compensate for the increased opacity of the atmosphere, there will be an imbalance between the rate at which energy enters the climate system and the rate at which energy leaves the climate system. And the heat content will go up.
Kim wrote:
I believe both I and Hank Roberts suggested you check out:
You Bet!
January 31, 2008
http://tamino.wordpress.com/2008/01/31/you-bet/
George // March 30, 2008 at 5:30 pm |
HB says:
I messed up the indent above (sorry)
In light of this, I find the paragraph immediately after Lucia’s above table particularly ironic:
How about the following variation on the same theme?
I would also point out that the “95% range” associated with IPCC scenario B1 projection given in AR4 would actually not be rejected by (ie, would not be inconsistent with) the majority of Lucia’s results (for different data sets, based on her Cochrane Orcutt method).
Upper value for Lucias results
GISS -0.4 + 2.2 = 1.8C/century
NOAA -0.3 + 1.7 = 1.4C/century
UAH -0.8 + 2.9 = 2.1C/century
average, then fit -1.1+2.2 = 1.1C/century
Compare those to the “95% probability range” for the IPCC scenario B1
B1 +1.8°C (1.1°C to 2.9°C)
From IPCC AR4 WG-1 report (Chapter 10, executive summary, page 749:
Actually, some of Lucia’s results would not be inconsistent with the range for some of the other scenarios either.
At the very least, that would seem to be further reason for caution in making a “rejection at 95% level” claim (to say nothing of the fact that Lucia has compared one temperature trend with short term noise to another (IPCC projected) trend without it)
Hank Roberts // March 30, 2008 at 5:31 pm |
Awesome patience, Tim.
I hope Mr. Kim learns from your help.
David B. Benson // March 30, 2008 at 6:25 pm |
What Hank Roberts just said.
Heretic // March 30, 2008 at 9:01 pm |
Indeed, this is one informative post, Tim. It’s good that some do check into the existing real science out there. And the links and references do add to the value of it, contrarily to what Kim has suggested.
So far, Kim’s message can be summarized like this: we don’t know how CO2 affects climate (which some would say can be argued), so let’s prefer other explanations of which we know even less how they affect climate, and let’s not examine them as closely as we examine the CO2 idea. Right, makes perfect sense.
cce // March 31, 2008 at 12:03 am |
Unbver,
There is no doubt that explosive eruptions cause significant cooling, but I don’t think El Chichon is responsible for entire cool patch between 1982 and 1987. It’s hard to believe that there wasn’t any weather-related cooling in there, when just about every other equivalent period has some. I’ll try to post an example, or maybe Tamino can explain the method he used in his “Many Factors” post. The problem is that El Chichon and Pinatubo are placed such that it makes it difficult to look at the last 30-40 years without them mucking up any conclusions based on short 7 year periods.
Hank Roberts // March 31, 2008 at 1:22 am |
http://www.nature.com/nature/journal/v451/n7176/fig_tab/nature06590_F1.html
“… Until recently, the circulation of the ocean was thought to comprise two fairly independent parts. …. However, the dichotomy and the use of the term ‘thermohaline’ have almost disappeared from the oceanographic literature, because the circulation in the interior is now increasingly seen as being driven by turbulent mixing from the winds and tides5, 6 and directly by the winds themselves7.
The westerly winds over the Southern Ocean seem to be crucial in this regard ….
…. westerly winds in both hemispheres have been shifting polewards and getting stronger over the past 40 years9, 10, partly in response to the warming from higher atmospheric CO2 concentrations11, 12. Thus, the strongest westerlies are now more squarely over the ACC, and — as expected — they seem to be doing more work to drive the ACC and more work to draw deep water up to the surface than they were 40 years ago ….”
http://www.nature.com/nature/journal/v451/n7176/full/nature06590.html
Timothy Chase // March 31, 2008 at 2:28 am |
Hank Roberts wrote:
This would presumably be why circulation around Antarctica is bringing more organic material to the surface — releasing carbon dioxide and methane, and as such, why the Southern Ocean is becoming less of a carbon sink. However, it also brings to mind the rivers that run between the ice sheets and rock — below the ocean surface. If warmer water starts passing through those rivers, the ice sheets might prove less stable. And perhaps this is part of what explains the strength slow feedbacks which presumably aren’t included in Charney Climate Sensitivity.
Different pieces in the same puzzle — or so it would appear. But I honestly hadn’t thought of the poleward migration of the Westerlies in quite this light before.
Timothy Chase // March 31, 2008 at 3:01 am |
Jeez!
That article is open access — a great overview — and a definite keeper!
Incidentally, I didn’t know that the thermohaline belongs in the dustbin, either. And it answers some other questions.
Unbver // March 31, 2008 at 5:15 pm |
CCE if you look at the revised Hadley Centre graph of smoothed trends http://hadobs.metoffice.com/hadcrut3/diagnostics/global/nh+sh/
you see that the 21 point binomial smoothing is designed to remove general weather noise. With that smoothing there was a very clear strong warming trend from 1975 to 2002. After the Chichon and Pinatubo eruptions the trend line flattened, but it did not turn downwards. There is however, even on the revised graphic, a definite downward trend recently, which cannot be explained by weather noise.
Phil. // March 31, 2008 at 6:25 pm |
“After the Chichon and Pinatubo eruptions the trend line flattened, but it did not turn downwards.”
Are you sure it didn’t?
The whole point is dealing with the end point, terminate the temperature record after Chichon or Pinatubo, then smooth and see what you get.
Dave Andrews // March 31, 2008 at 8:33 pm |
Hi, this is my first post and perhaps it is a naive question, but early studies of CO2 concentration in glaciers found levels considerably higher than those we are worrying about today – up to 7400ppmv.
If the effect of CO2 is as great as now being claimed how did the earth recover from those earlier high values?
[Response: How early are we talking? And what's the source of the data?]
Dave Andrews // March 31, 2008 at 9:04 pm |
Studies from approx 1956 -82 quoted in Jaworowski et al, Atmospheric Co2 and Global Warming, Second Revised Edition, MEDDELELSER NR119, Oslo !((”
[Response: I don't have a copy of that book, or ready access to a good science library. What I'm really trying to find out is, at what time was CO2 concentration 7400 ppmv? So, which studies (referred to in Jaworoski's book)?
You say "studies from glaciers," which would imply some time in that last million years. Frankly, that just ain't so.]
Dave Andrews // March 31, 2008 at 9:05 pm |
Sorry, should be Oslo 1992
Hank Roberts // April 1, 2008 at 1:48 am |
Rushing, but I found one likely hit here by just pasting the question into Google:
– found about 297 English pages for
CO2 concentration 7400 ppmv?
One of the first hits is:
… (1992 b) compiled all such CO2 data available, finding that CO2 levels ranged from 140 to 7400 ppmv. However, such paleoatmospheric CO2 levels published …
folk.uio.no/tomvs/esef/ESEF3VO2.htm
Hank Roberts // April 1, 2008 at 4:51 am |
And where was it published? Google is your friend:
“… a non-refereed magazine published by Lyndon LaRouche. … Jaworowski, Z., T.V. Segalstad, and N. Ono, 1992, Do glaciers tell a true atmospheric CO2 …
en.wikipedia.org/wiki/Zbigniew_Jaworowski
Hank Roberts // April 1, 2008 at 4:54 am |
Oh, and Eli’s covered Segalstad long since, see
http://rabett.blogspot.com/2006/12/gift-for-john-h.html
Timothy Chase // April 1, 2008 at 7:01 am |
Hank and everybody, that comes from a piece that begins:
[Note: get rid of the extra spaces]
It is — as is obvious — a denialist tract in the creationist tradition of science as grand conspiracy. It claims that the figures were gathered in 1992, but nothing was ever published after 1985 showing high CO2 levels (I have seen 3000 ppm or higher before), and the results were edited with large amounts of data being thrown out. Right up there with man landing on the moon being filmed in Arizona. (I ran into some geocentrists out of England that still give that some credence.)
Timothy Chase // April 1, 2008 at 7:22 am |
Incidentally, I am not saying that glaciers don’t pose a problem for our understanding of hothouse high-CO2 paleoclimates. At least one recent paper suggests that they might — which gets referenced at RealClimate:
But this stuff doesn’t get swept under the rug. It is out there in the open. It gets debated. And if it turns out that there were glaciers under those conditions, someone is going to make a real name for himself when he explains how — or at least comes up with convincing evidence that there were such glaciers. (Personally, I don’t think there were, but if there were, perhaps those westerlies played a role.)
Dave Andrews // April 1, 2008 at 8:41 am |
According to Jaworowski the study that found concentrations ranging from 1700 – 7400 ppmv was Raynaud & Delmas 1977 “Composition des gaz contenus dans la glace polaire” In Isotopes and Impurities in Snow and Ice. International Association of Hydrological Sciences (Washington DC) Publ.,118:371-381
Jaworowski et al can be found at http://folk.uio.no/tomvs
Timothy Chase // April 1, 2008 at 2:37 pm |
I should have recognized the name of Segalstad’s expert. Jaworoski and Segalstad have authored at least one piece together:
Jaworowski, Z., Segalstad, T.V. and Hisdal, V., 1990. Atmospheric CO2 and global warming: a critical review., Rapportserie 59, p. 76, Norsk Polarinstitutt, Oslo.
Regarding Jaworowski:
Dave Andrews // April 1, 2008 at 3:25 pm |
You might also have noted that Jaworowski is a former Chair of the UN Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) and has held posts at The Norwegian Polar Research Institute and the National Institute for Polar Research in Tokyo, so presumably he has some scientific credentials.
However, his study references 16 papers that report CO2 concentrations in amounts considerably greater than those that pertain today and my original question stands – if these are correct how did the earth manage to come through these periods given the dire predictions that are currently being made about rising CO2 levels?
Hank Roberts // April 1, 2008 at 3:41 pm |
Yep. Dave, where did you come across this particular piece? You’ve got one of the real outliers there, and if you want to understand the science you’d do better to start with work that has more connection to the science generally.
You know how to look up citations, to see what use later scientists have made of any particular piece of work, to get an idea of whether it’s useful enough to be relied on by others? (If not a reference librarian will help gladly.)
Hank Roberts // April 1, 2008 at 3:43 pm |
Oh, and the point is — outliers don’t get discarded without some reason; but they do get discarded if there’s reason.
http://www.causeweb.org/resources/fun/watermarked/CityDumpOutliers.jpg
Hank Roberts // April 1, 2008 at 3:47 pm |
And, of course
http://www.nearingzero.net/screen_res/nz021.jpg
Phil. // April 1, 2008 at 3:49 pm |
“Jaworowski has also written that the movement to remove lead from gasoline was based on a “stupid and fraudulent myth,” and that lead levels in the human bloodstream are not significantly affected by the use of leaded gasoline”
Then he should stick to his own area of expertise because he’s flat out wrong on this! Lead was removed from gasoline because it poisoned catalysts, nothing to do with leads in the blood.
Dave Andrews // April 1, 2008 at 7:58 pm |
Phil,
In my recollection, here in the UK, the reason for banning lead in petrol was all about health issues and especially the effects on children.
Incidentally your post inadvertently supports Jaworowski because he said it wasn’t about lead levels in the human bloodstream and so do you.
Dave Andrews // April 1, 2008 at 8:40 pm |
Hank,
When you say outlier are you referring to Jaworowski?
But what about the 16 papers he cites, are they all outliers too?
JCH // April 1, 2008 at 9:34 pm |
They trumpet the reduction in the levels of lead in children today, but in the 1970s the presence of leaded gasoline at gas stations meant stubborn customers could buy it and ruin their pollution-control systems – which they did in hordes.
So it had to go. Even if it had been totally benign in humans, it still would have gone.
Phil. // April 1, 2008 at 9:40 pm |
“In my recollection, here in the UK, the reason for banning lead in petrol was all about health issues and especially the effects on children.”
No that’s all later propaganda, and is mainly related to lead-based paint anyway. Leaded gasoline (petrol) had to be removed to allow exhaust catalysts to be used so that the new clean air targets could be met.
“Incidentally your post inadvertently supports Jaworowski because he said it wasn’t about lead levels in the human bloodstream and so do you.”
No the quote attributed the introduction of lead free gasoline to a “stupid and fraudulent myth”, it was not as I’ve outlined above.
Hank Roberts // April 1, 2008 at 9:53 pm |
Dave, did you find that in a science journal somewhere? Did you find any papers citing that paper later? What source are you looking at, and why do you trust that source?
Phil, same question.
Both of you, when I go looking for information on what you post, I find the same thing you ought to find — but it doesn’t support what you’re posting, that’s why I’m asking where you’re getting your information.
Dave Andrews // April 1, 2008 at 10:10 pm |
JCH and Phil,
Perhaps what we are coming across here are cultural differences between the US and UK, although presumably the “clean air targets
” would have been introduced with at least some health implications in mind.
The legislation was introduced here much later than in the US and so the technological reason
that earlier applied was perhaps not a seen to be as important as before.
Hank Roberts // April 1, 2008 at 10:14 pm |
To save time, this may help; Phil, the problem was known in the 1920s. What happened? The industry could deny the epidemiology and did for decades; they couldn’t deny the failure of catalytic converters. So, yes, regulation was done to protect the auto industry, not to protect children. But the children were protected by the regulation, and more regulation is needed now that we know.
PDF: http://www.ajph.org/cgi/reprint/75/4/344.pdf
Am J Public Health. 1985 Apr;75(4):344-52.
A ‘gift of God’?: The public health controversy over leaded gasoline during the 1920s.
Dave Andrews // April 1, 2008 at 10:24 pm |
Hank,
Obviously I didn’t find the Jaworowski paper in a sciencejournal but it was published By the Norsk PolarInstitutt.
Does that count for a lot – I don’t know do you?
But he cites 16 papers – tell me if they are wrong and how. They may be older than the current crop but have they been shown to be wrong? I’m just trying to find some elucidation.
Hank Roberts // April 1, 2008 at 10:50 pm |
Dave, anyone can cite anything.
I don’t want to go way off track for this topic. There’s the Open Thread for such.
Hank Roberts // April 1, 2008 at 10:56 pm |
Dave, one suggestion — go to Google Scholar and search on the authors you want to know about. The default is “since 2003″ but you can work backwards.
I found doing this search
– Jaworowski Segalstad Hisdal 1990 –
did not match any articles published since 1999.
Phil. // April 2, 2008 at 1:38 am |
“Phil, same question.
Both of you, when I go looking for information on what you post, I find the same thing you ought to find — but it doesn’t support what you’re posting, that’s why I’m asking where you’re getting your information.”
Hank, I don’t need to look it up, I worked on the subject with research contracts from the Dept of Energy, Ford, GM, Volkswagen, Honda, Yamaha! Trust me, to cut emissions to the required levels needed catalytic converters which meant lead had to go, the stories about lead in the blood was some after the fact rationalizations. In Europe the changes came later but for the same reason. Sulphur has been removed from fuels for a similar reason, I was at a meeting between government, fuel refiners and car manufacturers: basically in order to meet regulations the car manufacturers said the only way to do it is to reduce sulphur, the refiners said we’re not going to change voluntarily because we don’t trust our competitors! So the government had to impose it and everyone was happy!
luminous beauty // April 2, 2008 at 2:41 am |
Phil.,
No after the fact rationalizations, but the industry in which you were working did try their best to deny and obfuscate epidemiological research on the health effects of air-borne lead at the time.
http://www.hsph.harvard.edu/review/review_fall_05/rvwfall05_schwartz.html
There is a pattern here.
Hank Roberts // April 2, 2008 at 3:09 am |
Yep.
It’s no surprise the government was willing to intervene at the request of the industry to make them all do the same thing, the “level playing field” is the argument.
And it’s no surprise that since the 1920s, the health problems of lead in gasoline had been ignored.
As you’ll see if you read some of the references, the level of lead known to do brain damage to children keeps being readjusted lower and lower, as the background level from gasoline and paint and battery refining and the other polluting uses decreases — and the epidemiologists are able to show the effects of small amounts against the lower background.
The small signal takes some time to emerge, with the reduction of the massive background level of the old industries.
There may be a parallel here.
cce // April 2, 2008 at 5:03 am |
Unbver,
The most significant negative 7 year trends in the HadCRU data ended September 1986 and April 1994. Unless something unusual happens in the next year or so, the current cooling will not equal those periods, but those were dominated by El Chichon and Pinatubo. In order to figure out which period was most significant in terms of weather noise, you’d have to remove the volcanoes.
Dave Andrews // April 2, 2008 at 2:10 pm |
Hank,
Thanks.
Chris O'Neill // April 2, 2008 at 3:59 pm |
Unbver:
The 21 point binomial smoothing works OK as long as you have 21 years of data available to give you the smoothed value for the middle year. However, after 1997, they can’t actually use a 21-year binomial smoothing because there aren’t 21 years available around the required year. This means that as the year becomes more and more recent after 1997, there is less and less smoothing possible and hence the “smoothed” figures are more and more affected by weather noise. So the downward trend recently (all 0.012 deg C that it is) can definitely be caused by weather noise.
JCH // April 2, 2008 at 4:06 pm |
When much of this happened I was the national training director for a carburetor company. Part of my job was to change the extremely negative attitudes mechanics and dealerships had toward pollution systems on the new cars.
Catalytic converters first appeared on cars around 1975, and they had to have unleaded gasoline. On that date you could paint your child’s room with lead paint. Leaded gasoline would foul a converter in just a few minutes. It’s been a long time, but after lead fouling, I think we were able to get some of them to work again, but I believe many had to be replaced. They continued to sell leaded gasoline for years to service the large fleet of older cars.
There were two major problems when converters first appeared. I personally witnessed mechanics at car dealerships illegally “convert” new cars back to leaded gasoline prior them being put out on the lot for sale. The owners of the dealerships were telling them to do that. The other problem was the accidental use of leaded gasoline in lead-free cars.
They introduced smaller nozzles for lead-free pump handles, and a mated restrictor in the car’s gasoline fill pipes. The leaded nozzles were larger, and they would not fit into the restrictor. Some dealerships got stern warnings from the EPA.
The man in the Harvard article did not start working for the EPA until, I believe, much of what I’m describing had already been worked out. In all the time I trained mechanics and dealerships, I never heard a word about children’s IQs. I wish I had as those guys were extremely resistant to the new pollution equipment. I could have used an additional argument.
The desire for clean air (reduction of CO, NOX, unburned hydrocarbons) killed the future of leaded gasoline. It was illegal to produce cars that used it, so it was doomed. Eliminating smog was the primary driver in eliminating leaded gasoline. Smog was about all they talked about in the late 60s and early 70s. Newsweek and Time said it obscured the sun in Los Angeles.
Lead’s toxicity kept it killed. That is where the man from Harvard did his excellent work. They were trying to reintroduce lead to make gasoline cheaper. Somebody was suggesting doing the the same thing just recently. The rats always want to come out and play.
Unbver // April 2, 2008 at 5:29 pm |
CCE, as I understand it weather “noise” should be randomly distributed on either side of an underlying trend. It seems a bit unlikely that genuine random “noise” would spend seven years heavily on one side of the trendline to the extent that the trend is obliterated. This leads me to think that either there has been some sort of temporary event similar to a volcano of which we are unaware, or that the underlying long term trend has changed since the turn of the century.
[Response: First, that's one of the misconceptions about *autocorrelated* noise. White noise (uncorrelated) has the property that even for brief time spans it tends to be evenly distributed on both sides of the trend, but autocorrelated noise tends to have long sequences which fall to one side. Only over the "long haul" will it be evenly distributed.
Second, and perhaps even more to the point, if you look at this graph you'll note that it has *not* been to one side of the underlying long-term trend for many years -- unless you're referring to GISS data being on the *high* side of the long-term trend for seven years.]
Hank Roberts // April 3, 2008 at 3:05 pm |
New site, familiar names:
http://landshape.org/enm/recent-climate-observations-compared-to-predictions-by-rahmstorf-etal-a-review/
george // April 4, 2008 at 2:36 pm |
Lucia did not read the AR4 document very carefully and apparently just “eye-balled” the IPCC uncertainty off the IPCC graphic, rather than to take, as she says
[bold added by me above]
Here’s what Lucia says in regard to what she has claimed are the IPCC numbers (trend, range) upon which she has based her “falsification at the 95% level”:
FALSE!
The IPCC does provide trends and “95% probability ranges” for the projections for the various scenarios in AR4 (albeit for the century time scale) and as I pointed out in this comment above, even her claimed results (trend + error bar) would not exclude (ie not reject) at 95% confidence level the results for the B1 projection given by IPCC in AR4.
Upper value for Lucias results
GISS -0.4 + 2.2 = 1.8C/century
NOAA -0.3 + 1.7 = 1.4C/century
UAH -0.8 + 2.9 = 2.1C/century
average, then fit -1.1+2.2 = 1.1C/century
Compare those to the trend +- “95% probability range” for the IPCC scenario
B1 +1.8°C (1.1°C to 2.9°C)
From IPCC AR4 WG-1 report (Chapter 10, executive summary, page 749:
The IPCC did not provide the kind of short term projection (including short term noise from El Nino, la Nina etc) required for Lucia to “falsify” (her word) at the 95% level based on a 7 year time span since 2001, but in direct contradiction to her claim (below), they did indeed provide (in AR4)
Full quote from Lucia:
I agree. She would actually have to read those values!! (in AR4)
For those who do not take the time to carefully read and understand what the IPCC has actually done, it will not matter how much information the IPCC provides: it will never be enough for some.
george // April 4, 2008 at 2:53 pm |
The above
should read
As indicated in my previous comment above, the numbers in the parens are ranges (ie, actual temperature bounds), not +- uncertainties to be added to the trend
Here’s the full quote from the IPCC AR4 document (that I also gave above):
Phil. // April 4, 2008 at 3:22 pm |
George I think you should read your IPCC report more carefully, the uncertainty ranges you quoted are the -40%/+60% values which are said to be ‘likely’, does that represent 95% probability?
Dano // April 4, 2008 at 3:29 pm |
Hank, Landshape was started by a commenter at CA. I’d say he’s the denialist Tamino, if you will.
;o)
Best,
D
Hank Roberts // April 4, 2008 at 4:19 pm |
Oh, I agree — it’s a cross-reference.
george // April 4, 2008 at 6:15 pm |
Phil said:
I stand corrected, the IPCC ranges in the brackets are not the same as the “5-95%” ranges shown in IPCC Fig 10.29 (Ar4, WG-1, Chap 10, p 809) which shows the mean trends and 5-95% probability ranges for several of the projections for the various scenarios (and also shows the overlap of these with -40% and + 60% values for the ranges)
But Lucia implied that the IPCC did not provide the
— lucia
You can see yourself from Fig 10.29 that this is simply not true.
In fact, there is more than ample information in Fig 10.29 to see that several of Lucia’s claims of IPCC falsification at 95% are simply not warranted (contradicted, in fact)
In particular, For scenario B1 (the scenario that I said above was not “falsified at 95%” by many of Lucia’s results), the lower value for the IPCC “5-95%” range is virtually identical to the lower value for the -40% range (ie, 1.1C).
AS I indicate above, this 1.1C value, in turn, is actually included within several of the ranges provided by Lucia (for the various data sets and trend estimation methods that she calculated) Note: Only the lower value of the IPCC range is relevant to the discussion.
We compare Lucia’s upper value to the lower value for the range of B1 (in which case 5-95% and -40% are the same)
Upper value for Lucias results
GISS -0.4 + 2.2 = 1.8C/century
NOAA -0.3 + 1.7 = 1.4C/century
UAH -0.8 + 2.9 = 2.1C/century
average, then fit -1.1+2.2 = 1.1C/century
So, here’s my corrected statement:
My conclusion does not change. Lucia was simply not warranted in her claim of “falsification at 95% level”, since 1.1C is actually less than or equal to the upper values of Lucia’s 95% confidence range (for many of the data sets and processing methods that she showed)
Dano // April 4, 2008 at 6:41 pm |
Lucia was simply not warranted in her claim of “falsification at 95% level”, since 1.1C is actually less than or equal to the upper values of Lucia’s 95% confidence range (for many of the data sets and processing methods that she showed)
If Lucia was correct, she’d be published by now, heralded as the next Galileo, on book tour, appearing on Jay Leno…
Best,
D
George // April 6, 2008 at 2:44 pm |
I can appreciate the motivation for starting with the 2001 data for the IPCC projection to avoid the appearance of cherry picking, but I have a question about how that may bias the result for the trend started in 2001.
This relates to a general question about the “best” way to calculate and “join” sequential trend lines — particularly a relatively short one following a significantly longer one.
For simplicity, let’s assume we are trending only annual average temperature data for just one data series.
If I trend the data (using OLS, for example) from 1975 – 2001 (inclusive), I get a line with a certain slope and intercept.
If I find the trend line from 2001 (inclusive) to the present, I get another line (with different slope and intercept).
In general, the ordinate corresponding to the year 2001 will be different for each of these lines .
This would appear (to me, at least) to have some relevance if one is trying to figure out the “most probable” slope for the trend line from 2001- present.
It would seem reasonable to assume that there should be some kind of “continuity” between the trend lines (even if not between their slopes). “Jumping” from one ordinate value in 2001 to another would seem to violate that.
This is actually related to a more general question: how does one interpret the ordinate corresponding to each year for a particular trend line?
Does it somehow represent the “most probable” temperature for that year in the absence of noise? If that is the case, can one justify a jump from one trend to the next?
In general, are there any statistical (or physical) requirements about how a short trend that follows a long one should be calculated in order to take the “ordinate in the overlap year” issue into account?
Is it really justifiable to calculate the second trend line completely independently from the first — ie, as if the first never occurred?
I’m sure there is probably a standard way of dealing with this issue, but I don’t know what it is.
Calculating the trend lines independently and then “sliding” the second trend line up or down (ie, parallel to itself) so that the ordinates correspond in the year 2001 would seem to be a less than optimal way of dealing with the issue.
One method comes to mind but I have no idea whether it is justified statistically: “fixing” the ordinate for 2001 for the second trend line at the value calculated from the trend line for the years from 1975-2001 and then estimating the slope of the trend line from 2001-present by “swinging” the second trend line up or down (using 2001 as a hinge) so that it has the slope that minimizes the sum of the squares for the data points beyond 2001. (in other words, mandate that the two trend lines start at the same point in 2001)
Paz // April 6, 2008 at 9:53 pm |
I have a related question. In my field of work (psychology & neuroscience) we are taught that we have to correct all significance tests for multiple comparisons, that is, every significance value gets multiplied with the number of tests we can run on the data. The critical point is that this number is not the actual number of tests that you have run on the data, but the number of tests you *could* run in principle
My intuition is that something similar should apply here as well. After all, Lucia does not test the overall prediction of the IPCC, which is, if I understand correctly, based on a longer timeframe. Because she could, in principle, have run the tests that she has run in any other year (2006, 2007, but also 2009, 2010, etc.) and get different results, wold she not have to correct her significances accordingly. To use the Bonferroni-style simple multiplications with the number of possible tests is probably to conservative, after all, the different possible tests are not statistically independent. Is there any agreed and tried methodology to this in cases like these?
If she does not do this, she could still claim that in 2007 the IPCC prediction was overestimating the rate of warming. But that conclusion would have roughly the same scientific status as the idea that there was no warming from 1998, because it is based, at least in part, on cherrypicking.