Climate Audit

I’m making some changes to this post in light of comments from Gavin Schmift.

The difference between Hansen Scenarios A and B occasioned a controversy involving Hansen and Pat Michaels in 1998, which reverberates to this day. In previous posts, I’ve provided some of the links. Hansen described Scenario A as “Business as Usual” in his 1988 testimony and only showed Scenario A in Hansen et al 1988 Figure 3b for the period 2030-2050, but elsewhere in the same article said that Scenario B was the “more plausible” due to resource limitations. The arguably inconsistent statements thus gave ammunition for both supporters and critics.

As I introduced myself to the data, I became interested in a different aspect of the problem – although partisans, as so often in climate science, seem to think that I’m taking sides in an old dispute that I was not party to. As I parsed through the data, I became increasingly puzzled by what actually accounted for the difference between Scenario A and B – the “active ingredient”.

Gavin Schmidt said that Scenario A was “exponential” and Scenario B was “linear”, but in the period of time to 1997 (or 2007), the differences between the two are very slight and it seemed very implausible to me that these slight differences could account for the differences between Scenario A and B temperatures.

In summer 2007, Gavin Schmidt posted up digital versions of the GHG concentrations used in the Hansen et al 1988 projections in a relalcimate thread; Lucia directed me to this thread yesterday and I believe that I have identified the salient differences between Scenarios A and B. As so often in climate science, the true explanation is different from what either side thought and completely unexpected based on the written description in Hansen et al 1988.

First, here is a diagram showing CO2 concentrations for Scenarios A,B and C (per realclimate version), as compared with observations to 2006 (GISS). Obviously, there is no material difference between Scenario A and Scenario B concentrations or, for that matter, with observations up to 2006 or so. CO2 concentrations start to diverge between the two scenarios in the next decade, but, for the near time analysis, calling one graph “exponential” and another “linear” is not really salient to the quantitative analysis (contra Schmidt’s characterization at realclimate.)

Figure 1. CO2 concentrations. A,B, C from realclimate; observed from NASA GISS.

I’ll discuss methane below, as there are important and interesting aspects to methane projections, but the explanation of the Scenario A-B problem does not lie with methane or N2O, but with a very peculiar handling in Hansen et al 1988 of CFC11 and CFC12. Here’s how Hansen et al 1988 described their CFC scenarios:

In Scenario A, … CCl3F (F-11) and CCl2F2 (F-12) emissions are from reported rates (Chemical Manufacturers Association (CMA) 1982] and assume 3% yr-1 increased emission in the future with atmospheric lifetimes for the gases of 75 and 150 years respectively. ,, In Scenario B… the annual growth of CFC11 and CFC12 is reduced from 3% yr-1 today to 2% yr-1 in 1990, 1% yr-1 in 2000 and 0 in 2010…. In Scenario C, CFC11 and CFC12 abundances are the same as scenarios A and B until 1990; thereafter CFC11 and CFC12 emissions decrease linearly to zero in 2000.

Yesterday I applied this description in what I thought was a straightforward way and obtained negligible difference between CFC11/12 concentrations between Scenarios A and B in the period to 2006 (and identical values to 1990) – Hansen clearly says that CFC concentrations for Scenario A and B are identical up to 1990 (since both are identical to Scenario C concentrations).

The problem as of yesterday’s post was that there was negligible difference for any of the GHGs up to 2006 and it seemed impossible to account for the Scenario A-B difference. In that post, I observed:

In Scenario A, the “other” trace gases are accounted for by doubling the impact of CFC11 and CFC12, while in Scenarios B and C, they are held to have no incremental impact. So there is a difference between Scenarios A and B here, but it doesn’t look like it should have much impact.

This hypothesis appears to have been incorrect. In the Figures below, I’ve plotted CFC11 and CFC12 concentrations from the May 2007 archive at realclimate as compared to observed levels, showing that (1) observed CFCs have even lower concentrations than Scenario C. In my earlier version, I observed the large differences between Scenarios A and B, but failed to connect to the observation made in yesterday’s post about doubling in Scenario A. In my implementation of the doubling method, I had created another column for Other Trace Gases and assumed that they would keep this effect separate. In fact, what they did was to physically double the CFC11 and CFC12 values in Scenario A. I can now see how this ties into the description, but it seems an odd way of doing it. Nonetheless, my observation that the difference between Scenarios A and B rests with the CFCs was not incorrect.

Figure 3. CFC11 and CFC12 concentrations by scenario. In Scenario A, CFC11 and CFC12 values have been arbitrarily doubled as a catchall for other trace gases.

Yesterday, I had incorrectly assumed that differences in CFC would not be consequential, but it turns out that the handling of Other Trace Gases accounts for the difference between Scenario A and Scenario B.

Using the forcing formula at NOAA (without taking any position at present on the meaningfulness of his particular calculation), one can see that substantial differences between Scenario A and B forcings arose by doubling the CFC11 and CFC12 concentrations to account for other trace gases. This effect is cumulative to 1990. It accounts for the difference between Scenarios A and B – not the difference between exponential and linear forms.

Back to methane. The left panel shows methane levels in Hansen et al 1988 as compared with present estimates. Two things pop out: the levels in Hansen et al 1988 reported for 1987 are noticeably higher than levels presently believed to have existed in 1987. Is this because of changing ways of measuring methane content? At present, I don’t know, but the differences are surprisingly high, as compared to CO2. There’s a little separation in near-time between Scenarios A and B as one gets to 2005, but the differences are not large (confirming that it’s the CFCs that account for the difference.) As noted yesterday, methane concentrations have not increased nearly as much as Hansen projected, following a course parallel to Scenario C at a lower concentration. The forcing estimates obviously parallel the concentration estimates.

So where does this leave us? To clarify one point, I don’t have any problem with properly articulated scenarios as a way of analyzing projections. How else are you going to do it? With respect to projections of CO2 concentration made in the 1980s, the projections are pretty much on the money. But even if they were off to some degree, all it would do is defer or accelerate the date at which CO2 doubled somewhat. And I agree with concerned climate scientists that doubling CO2 is an unplanned real-life experiment that could have very serious consequences. If the impact is 2.5 deg C, then we should probably be glad that that’s all it is, as the results could have been much worse.

If Hansen’s 1988 model didn’t work very well, then that does not mean for me that some other model mightn’t work, even some other model by Hansen. Many readers here take much more hard-edged positions on these things than I believe to be justified.

Having said that, I’m also trying to figure out how one gets from point A to point B in all of this, a path made more difficult because the field is all too often characterized by sloppy practices.

In this particular case, I’m not saying that any of the errors or mis-descriptions necessarily “matter”, other than to a person actually trying to understand what the authors are doing. And once again, we have a situation all too familiar in climate science.

Given that the handling of other trace gases has such a material impact on the difference between Scenarios A and B, Hansen et al 1988 did not describe the matter very clearly, Differing historical CFC concentrations between Scenarios A and B was material to Scenarios A and B outcomes, but mentioned only in the fine print and it was not at all obvious that its impact on the two Scenarios was as large as it was.

While my earlier post did not quite analyze the problem correctly, the information now provided by Gavin Schmidt raises more problems.

The major difference between Scenario A and Scenario B in near time (e.g. up to the Michaels-Hansen dispute) is the inclusion of an effect for other CFCs and trace gases through doubling the CFC11 and CFC12 impact. In longer extrapolations, the difference between exponential and linear extrapolations bites, but the effect does not appear material in the period prior to 2007. So the argument between Michaels and Hansen on Scenario A and B pertains entirely to the handling of other CFCs.

So there are two aspects to Hansen’s assertion that Scenario B was “more plausible” – the handling of CFCs in near time and exponential vs linear in deeper time. In the vituperative debate in 1998 and thereafter, no one seems to have focused on the impact of CFC doubling. Does it make sense to say that doubling CFC11 and CFC12 concentrations as a way of modeling the OTGs is “Business as Usual”? Not doubling these concentrations may very well be “more plausible”. Indeed, as events turned out, it seems that it was “more plausible” not to double CFC concentrations.

In the 1998 dispute between Michaels and Hansen, in my opinion, neither party comes out very well. Michaels’ should have illustrated all three scenarios – something that I’ve consistently done without even thinking about it. But Hansen shouldn’t have said in his 1988 testimony that Scenario A was “Business as Usual” and should not have made the misleading claim in 1998 that Scenario B was the one used in his 1988 testimony (where Scenario A was the one actually described as Business-as-usual).

As to how Hansen’s model is faring, I need to do some more analysis. But it looks to me like forcings are coming in below even Scenario B projections. Thus, if Hansen’s projections over-state what we’re experiencing in 2005-2010, then this does not per se invalidate Hansen’s model, as some people are too quick to conclude. To the extent that the differences lie in lower than projected forcings, I’m not prepared to place fault on the model for those defects. Whether the forcings account for the defects is a different matter and it would be interesting to run the 1988 model with actual GHG concentration changes – an experiment that’s long overdue and which would end much speculation about the merit or lack of merit of Hansen’s 1988 projections.