Thursday, May 24, 2012


cc: Tim Osborn <>, Jonathan Overpeck <>, Keith Briffa <>, Eystein Jansen <>,
date: Fri, 14 Jan 2005 16:26:52 +0100
from: Stefan Rahmstorf <>
subject: 6.5.8 on climate sensitivity and last millennium
to: David Rind <>

Hi friends,
I'm not working on this topic myself so I'm by no means an expert. But I am still quite
concerned with the wording in 6.5.8 on the last millennium.
First, to avoid misunderstandings, I would like to suggest again to describe forcings and
climate changes going forward in time, rather than going backwards in time. Even colleagues
here that I discuss it with misunderstand the present version with backwards reasoning - it
leads to phrases like "deforestation warming" (used by David in his last mail), although
deforestation caused cooling - backwards in time you can see this as a warming, but should
you call it "afforestation warming" if you look back in time? I suggest to use the
physical, forwards, time arrow in the discussion.
The section states:

If one takes mid-range estimates of solar and anthropogenic forcings, and assumes that
volcanic, tropospheric ozone and land albedo changes cancel out, the resulting radiative
forcing change is ~-2.4 Wm^-2.

I don't think we should give a "mid-range" of the forcing like this; the assumption that
ozone, land albedo and volcanic changes cancel is hard to justify in any case. For the
forcing we need to give a range in my opinion, not one number. If we give a range, it will
become clear that the forcing is too uncertain for drawing conclusions on climate
sensitivity from this time period.
The section contains the sentence:

Relating forcing to response, the sensitivity from the models is then on the order of
0.6�C/ Wm^-2 (or higher, depending on the model used); the sensitivity from the
observations, if taken at face value, would be considerably less.

Although this was toned down quite a bit, it still sounds like we can say that observations
suggest a considerably smaller climate sensitivity compared to models. Skeptics will
trumpet this sentence around the world. And I don't think we can justify such a statement.
I agree that apparently models have not used all forcings. But I think it is not as
clear-cut as you say that they would come to much larger temperature change as compared to
the data if they did. I talked to Martin Claussen, co-author of the Bauer et al. paper, as
he works on this period. Martin strongly disagrees with your statement

Bauer et al. used a large aerosol effect and still needed a large deforestation warming
to bring her results in line with the Mann et al. reconstruction (in fact, it was done
specifically for that reason)

The GHG, deforestation and orbital forcings are the known forcings, while solar and
volcanic aerosols rely on uncertain proxy reconstructions. To cover this uncertainty, Bauer
et al. did a sensitivity study and used different plausible scenarios, e.g., 0.24% and 0.32
% for solar change and a couple of scenarios for volcanic forcing, and they discuss the
results of all those.
I agree that for 2.4 W/m2 with best guess climate sensitivity you get 1.5 �C in equilibrium
- but much of this forcing is due to a rapid rise in the 20th Century, and the climate is
not in equilibrium with that - transient runs suggest that only 1/2 to 2/3 of the
equilibrium warming is realised (here the "looking back in time" issue is again a possible
source of misunderstanding - it is of lesser concern whether the MM is in equilibrium, but
rather to what extent the present is). That gets you down to below 1 �C from the 1.5 �C
equilibrium value.
Then you state the Mann et al. data are 0.5 �C below the 1990s in the Maunder Minimum. I
can see they are 0.4 �C below the reference level (I believe this is 1961-1990). The mean
of the 1990s is 0.3 �C above this level (I calculated this from the Jones data) - so I find
that the Mann data are in fact 0.7 �C below the 1990s in the MM. The difference between
model expectation for 2.4 W/m2 and the actual found in the Mann data is almost gone then.
Add to that the possibility that the Mann data may somewhat understimate the variability,
and I do not see any significant discrepancy between models and data, which we should
mention and which we could defend as real - even for "best guess" sensitivity and forcing,
let alone considering the uncertainty in those.
Cheers, Stefan

Stefan Rahmstorf

No comments:

Post a Comment