Arguments
Add Frame and Stone to the List of Papers Validating IPCC Warming Projections
Posted on 18 December 2012 by dana1981
Just a few weeks ago, a paper in Environmental Research Letters by Rahmstorf, Foster, and Cazenave (2012) confirmed the accuracy of the global surface warming projections made by climate models used in the 2001 and 2007 Intergovernmental Panel on Climate Change (IPCC) reports (the Third and Fourth Assessment Reports). Now a new paper published in Nature Climate Change, Frame and Stone (2012) has confirmed the accuracy of the temperature projections made by the climate models in the 1990 IPCC First Assessment Report through 2011 (Figure 1).
Figure 1: Observed changes in global mean surface temperature over the 1990–2010 period from HadCRUT3 and GISTEMP (red) vs. 1990 IPCC business-as-usual best estimate (dark blue), vs. projections using a one-dimensional energy balance model (EBM) with the measured greenhouse gas (GHG) radiative forcing since 1990 (light blue) and with the overall radiative forcing since 1990 (green). Natural variability from the ensemble of 587 21-year-long segments of control simulations (with constant external forcings) from 24 Coupled Model Intercomparison Project phase 3 (CMIP3) climate models is shown in black and gray. From Frame and Stone (2012).
Frame and Stone Methodology
The paper tests the IPCC warming projections using a simple one-dimensional energy balance model (EBM) comparable to the main model used to make the 1990 prediction, using similar input parameters. Frame and Stone then ran the model using just the radiative forcing (heat imbalance) caused by greenhouse gas (GHG) changes from 1990 through 2011, represented by the light blue line in Figure 1. Because the IPCC model projection is based on GHGs-only, this is the most applicable comparison. They also simulated other radiative forcings like changes in solar activity and particulates in the atmosphere, represented by the green line in Figure 1.
Results Similar to Prior Skeptical Science Analysis
We at Skeptical Science previously conducted a similar analysis to that in Frame and Stone (2012). The 1990 IPCC report ran simulations using models with climate sensitivities (the total amount of global surface warming in response to a doubling of atmospheric CO2 concentrations, including all feedbacks) of 1.5°C (low), 2.5°C (best), and 4.5°C (high) for doubled CO2 (Figure 2).
Figure 2: IPCC FAR projected global warming in the BAU emissions scenario using climate models with equilibrium climate sensitivities of 1.5°C (low), 2.5°C (best), and 4.5°C (high) for double atmospheric CO2
In reality, GHGs have increased about 20% slower than the IPCC's "business-as-usual" scenario, in part because of the success of the Montreal Protocol in reducing chlorofluorocarbon (CFC) emissions (CFCs are GHGs), and in part because of the collapse of the Soviet Union, among other reasons.
As noted above, the light blue line in Figure 1 is the most appropriate for comparison, and is very similar to our own previous analysis at Skeptical Science (Figure 3).
Figure 3: 1990 IPCC business-as-usual "best" global warming projection reflecting the observed GHG changes (blue) vs. observed average global surface temperature change from GISTEMP (red) since 1990.
Observed Warming Not Natural Variability
Frame and Stone (2012) also simulated the possible range of natural temperature variability since 1990 by using the ensemble of 587 21-year-long segments of control simulations with constant external forcings from 24 Coupled Model Intercomparison Project phase 3 (CMIP3) climate models. These give a 90% range of about ±0.19°C, and are shown in black and gray in Figure 1. The observed warming from 1990 through 2011 was approximately 0.39±0.20°C (95% confidence range); thus there is only a very small chance that the observed global surface warming over the past 21 years could be explained purely by natural variability.
IPCC Has Excelled at Global Warming Projections
Ultimately, Frame and Stone note that while there is a fairly large range in the envelope of all climate model projections, and while to some degree they may have gotten the right answer 'for the wrong reasons', the IPCC has thus far done quite well in projecting future temperature changes.
"...it seems highly likely that even in 1990 we understood the climate system well enough to make credible statements about how its aggregate properties would change on timescales out to a couple of decades, even in the presence of considerable uncertainty surrounding the exact forcing trajectory."
Rahmstorf, Foster, and Cazenave (2012) arrived at a similar conclusion by taking a very different approach, first using the statistical technique of multiple regression to filter out much of the short-term variability, then showing that the IPCC temperature projections have been very accurate (Figure 4).
Figure 4: Observed annual global temperature, unadjusted (pink) and adjusted for short-term variations due to solar variability, volcanoes and ENSO (red) as in Foster and Rahmstorf (2011). 12-month running averages are shown as well as linear trend lines, and compared to the scenarios of the IPCC (blue range and lines from the 2001 report, green from the 2007 report). Projections are aligned in the graph so that they start (in 1990 and 2000, respectively) on the linear trend line of the (adjusted) observational data. From Rahmstorf, Foster, and Cazenave (2012).
Frame and Stone have also shown it is very unlikely that natural variability alone can account for the observed global surface warming over the past two decades.
While it has underestimated many climate impacts, thus far the IPCC has done very well at projecting average global surface temperature changes.
The reason for them not doing so is clear, they do not have a robust theory (e.g., anthropogenic global warming) on which to base their predictions, so any predictions they make invariably falter after a very short time (e.g., McLean's disasterous attempt). Here are a few more failures by fake skeptics:
[Source]
Now it is OK to be wrong sometimes, but all the time? And to never learn from those mistakes, but rather just keep recycling the same old tired debunked nonsense about cycles etc.? No thanks, I'll stick with the real climate scientists.
In the interest of full disclosure, those words of mine were inspired by those of Dr. Pierrehumbert, whose AGU talk I had the good fortune to watch in person.
PS: Tks for fixing my broken link DB.
Since it's related to this post, I'm hoping that you would comment on Figure 1.4 in Chapter 1 of the IPPC draft. It seems to have the people at WUWT all worked up (and now my head hurts from reading the comments there).
If you wish to comment or query regarding one (or more) of the published papers the IPCC may refer to in the draft, in regards to the data and conclusions of those published works, that's another matter entirely - please do.
Sigh.
"...it is a tale, Told by an idiot, full of sound and fury, Signifying nothing." - Macbeth
ENSO is not really a forcing - it a mode of internal variability. There are models focussed specifically on predicting ENSO because of its meteorological importance but they have very limited success so far. Instead ENSO-like behaviour emerges from the models but this is a chaotic phenomena and so model outputs are not good predictors of actual ENSO. For this (and other issues with internal variability), models have very little skill at decadal or less prediction. They are after all climate models - targeting at predicting what will happen to 30 year averages. What you can expect for climate has be result of averaging many different runs with different initialisation and likely forcing scenarios.
AlexC, Yeah I read that section (& only that so far). Frankly I was hoping that Dana could provide more insight than I have, since it is directly related to the post he wrote above.
As for reduced GHG growth being the explanation, here are the numbers for CO2: forecast in 1990 from Figure 1.7 of Chapter 1 of FAR (approximately) 398 ppmv; actual for 2010 is 392. Is that drop in GHG emissions enough to drop the trend from 0.28 to 0.19?
Hard to say. Methane also was way below forecast. FAR doesn't include a methane future growth graph. However they do note in Table 1.1 of Chapter 1 it was 1720 ppbv and growing at 15 ppbv per year. Actual growth since 1990 is much much lower, growing at less than 4 ppbv between 1990 and 2010 inclusive.
Even if I agreed the actual to forecast miss for FAR is explainable by less than forecast GHG growth (I don't, at least not all of the forecast error), it will be harder to argue that going forward if La Nina conditions return in 2013 as currently forecast. (link attached)
http://www.cpc.ncep.noaa.gov/products/CDB/Forecast/figf4.shtml
The arguments coming back at me will be 2013 won't count because it is a La Nina year (if it does emerge as forecast). However, I have no doubt if you got an El Nino instead (and hence possibly another record) it absolutely would count and be further evidence of man-made climate change.
Fair enough. Read section 1.3.1 of that leaked report, which discusses the figure in question.
First,
Section 1.3.1 goes on to state:
They also do not, as a rule, include ENSO events except as emergent behavior in some of the models. Finally, Emphasis added.
---
The GWPF and WUWT arguments about this (still in review) figure ignore natural variability, ignore that the models did not and do not predict ENSO, that earlier models did not include Pinatubo, and that the model ranges shown are not error bars. Climate models are appropriate for predicting 30 year averages, not the next five years in detail, and the skeptics (should have) have been around long enough to know that.
In short, the GWPF and WUWT articles are a rather extensive misrepresentation of the figure discussed.
"...full of sound and fury, Signifying nothing."
Klapper @14 - CO2 is not the only GHG. CFCs play a big role, as noted in the above post. We've already analyzed the numbers - net forcing vs. temperature. I'm not sure why you're trying to parse it out by only looking at a couple of GHGs instead of looking at the net forcing. Frankly I'm not sure what point you're trying to make.
"the bars on the side represent the range of results for the scenarios at the end of the time period and are not error bars."
These are scenarios presumably for CO2 emission. But from Figure 1.6, the prior predictions for CO2 emission seem pretty good. Am I to conclude then, that if someone used the actual CO2 concentrations rather than the scenarios, the picture wouldn't have changed all that much? And if that's true, is that the reason why natural variablility, especially ENSO, is the dominating effect?
It was also interesting to see that the report says, "Analyses by Rahmstorf et al.(2012; submitted) show that accounting for ENSO events and solar cycle changes would enhance the comparison with the AR4 and earlier projections." Which Dana shows very nicely in Figure 4 above.
One issue especially bugs me. I can see that natural variability accounts for the discrepancy between the data and the scenarios. What i don't see is how "the globally-averaged surface temperatures ... generally are in the middle of the scenario ranges" after 2005 or so. They don't look like they are in the middle. Nor should they necessarily have to be.
The actual FAR scenario that best fits the modern realized climate forcing is Scenario 2.5(D). This was discussed in detail in "Wall Street Journal 'Skeptics' Misrepresent the IPCC" by Keith Pickering, 3 March 2012.
http://www.skepticalscience.com/wsj-skeptics-misrepresent-ipcc.html
As Dana said too, you're ignoring GHG reductions in other forms like CFCs, which were banned by the Montreal Protocol only just before FAR was released.
Furthermore, you're forgetting that since FAR was released there was an update to the forcing equations that the IPCC used, the new paper being Myhre et al 1998. The CO2 equation, ∆F = Aln(C/C_0), was A = 6.3 in FAR and A = 5.35 in Myhre et al 1998. In other words the IPCC was using a higher forcing per increase in CO2 than later research suggested to use. It's no wonder their prediction might be higher.
ENSO predictions right now are neutral with a slight positive bias. I myself wouldn't bet on another La Nina.
And there are ways, remember, to filter the ENSO signal anyway. Foster and Rahmstorf did this. If you really think that an El Nino will cause a spurious spike, then maybe we can just apply their methodology and remove it, and we'll see then.
JoeT @17:
They weren't just CO2 emission scenarios, they were total emission scenarios. The (1) and (3) points in Section 1.3.1 also tell why the projections aren't in line with the observations, which is because of natural variability that the projections did not model. So, if we include *only* present CO2 levels, of course they won't change much, but if we use the fuller 3D models and include the correct natural variability we have observed, the models will predict the temperature accurately.
As to "the globally-averaged surface temperatures ... generally are in the middle of the scenario ranges," that is quite clearly demonstrated by Figure 4. The statement doesn't mean that it tracks the mean all the time, but the observations fall within the first standard deviation range without seeming to be biased toward one side of it.
Re your claims concerning the FAR predictions, please look carefully at the figure inmy post @1 above and Figure 1 in the [main post]. You are applying a linear trend to a curve that is not linear, with the result that the resultant trend overestimates the amount of warming expected between the end points. Indeed, when done properly, the graphs above show excellent agreement between what was predicted for circa 2011 and what was observed.
I sense that you are reluctant to accept that predictions made over 20 years ago, using pretty simplistic models by today's standards, have done remarkably well. Now please compare those predictions with those of "skeptics" and those in denial that are shown above.
Your post contains at least one error. 2012 is probably officially going to be designated a La Nina year (yet it will very likely be the 9th warmest on record). As for 2013, it is still a little too soon to say, but the official forecast is for ENSO neutral conditions to persist into 2013. Your claim of a La Nina for 2013 misrepresents the official position of the CPC.
But here again we have "skeptics" arguing about the noise (i.e., ENSO) when it has already been pointed out by climate scientists ad nauseum that it is pointless (and misleading) to focus on periods of time that are unable to extract a statistically significant signal from the noisy temperature record. Yet "skeptics" continue to play this disingenuous and scientifically meaningless game.
FWIW, my simple box-type model is predicting a global surface air temperature anomaly for GISTEMP of about 0.61-0.65 C, so 2013 may tie or break the current global record (+0.64 C) without the boost of an El Nino. And think about this, barring a large tropical volcano blowing its top, 2013 will likely be warmer than 1998 was (+0.59 C), and that year followed a huge El Nino (a three plus sigma event if I recall correctly). If that happens it will impressive and speaks to just how much the positive radiative imbalance (from humans increasing greenhouse gases) is warming the planet.
[Edited to fix typos, see square brackets]
I suppose most of these talking points have been gone over before, but...
Lower CO2 is not the only factor that was not (could not?) be predicted in 1990 which has reduced the rate of increase in the surface temperature. There has been an increase in Asian aerosols, the sun has been less active than usual, and the ocean has been warming at depth - which may or may not be related to the dominance of La Nina in recent years. Not sure, but it appears you are trying to imply that, because the explanation is complicated, it must be wrong.
A model is just a way to approximate reality. Science works by replacing models with better ones, whether that is in chemistry, nuclear physics, what have you. So, if you are going to throw out the IPCC ensemble of models, what are you going to replace it with? If you are going to replace it with nothing, then you might as well be saying we don't know everything; therefore, we know nothing, and that conclusion really doesn't follow that premise.
If you can show a model which does a better job over the last 20 years, as well as the over the historical record, that would be very interesting. Pretty lines on a graph with no connection to the laws of physics would not be very interesting.
"As to "the globally-averaged surface temperatures ... generally are in the middle of the scenario ranges," that is quite clearly demonstrated by Figure 4. The statement doesn't mean that it tracks the mean all the time, but the observations fall within the first standard deviation range without seeming to be biased toward one side of it. "
It seems to me that Fig 1.4 does show a bias to one side, that the IPCC projections shown in the graph do side on the prediction of more warming than has occurred. Am I reading the graph incorrectly?
I understand that due to recent natural variation the global temperatures are "depressed" and therefore this is expected.
Maybe there was a bit of a misunderstanding? Sorry I was referring to Figure 1.4 in this blog post, not Figure 1.4 from AR5. Bad ambiguity on my part, no I mean that the results from Rahmstorf et al. (2012) show this quite clearly.
OK, my fault. I jumped to conclusions. I am confused about the draft report. The text of the draft and the placeholder figure 1.4 don't seem to say the same thing. This of course maybe because it is a draft.
You raised the same point that I did, which is that it is difficult to claim that Figure 1.4 of the draft shws the data right in the middle of the scenarios. If you're still following this thread Tamino has an excellent discussion of this exact point and shows that the problem is that the draft aligns all the projections at 1990, a warm year, rather than to a trend line to the data. In brief, this resolves the discrepancy between the draft report and what Dana is showing above.
If I can put in a request -- I'm very interested in understanding more about the physics of drought. More than just "the wet gets wetter and the dry dryer." I think this is a crucial issue that the world is going to have to face right now -- not some decades in the future. I've been reading the interesting papers of Isaac Held of the GFLD, but I'd like to know more about the assumptions that go into the various models, as well as the contradictions between models and the ramifications.