2010-12-10 09:35:26Dessler on cloud feedback...bad timing!
dana1981
Dana Nuccitelli
dana1981@yahoo...
38.223.231.252

Less than a week after my cloud feedback blog post, Dessler (2010) comes out.

I estimated the magnitude of the cloud feedback in response to short-term climate variations by analyzing the top-of-atmosphere radiation budget from March 2000 to February 2010. Over this period, the short-term cloud feedback had a magnitude of 0.54 T 0.74 (2s) watts per square meter per kelvin, meaning that it is likely positive.

Maybe I'll do an addendum cloud feedback blog post to discuss this study, although Dessler wrote about it at RC and Romm at CP.  Maybe I'll try to convert the W/m2 into an approximate temperature change.

2010-12-10 12:20:46A follow-up post would be great
John Cook

john@skepticalscience...
124.179.115.67

As it provides additional evidence for positive cloud feedback.

Andrew provides a nice overview at:

http://www.realclimate.org/index.php/archives/2010/12/feedback-on-cloud-feedback/

Roy Spencer provides a nice conspiracy theory laden rebuttal at:

http://www.drroyspencer.com/2010/12/the-dessler-cloud-feedback-paper-in-science-a-step-backward-for-climate-research/

2010-12-11 04:07:07
MarkR
Mark Richardson
m.t.richardson2@gmail...
192.171.166.144

My take (written up quickly, don't have time to fully reference, edit and put pics in):

 

 

Research has given us good confidence that if CO2 doubles in the atmosphere then there will be 1.2 oC of directly CO2 caused global warming and a combination of melting snow/ice and increased evaporation of water will act as positive feedbacks to increase this to around 2 oC[1].

This climate sensitivity of 2 oC plus would mean most recent global warming was human caused and more is in the pipeline; but some scientists believe that changes in clouds will cool us down, though the most public paper to claim this has been heavily criticised for using a method that can be fiddled to give any desired result and ignores much of the planet.

The strength of the feedback is calculated as:

(F = Δ R / Δ T - I'd put more explanation here)

Dr Roy Spencer believes these calculations are bad because we can’t tell whether the warming is changing the clouds or the clouds cause the warming through ‘internal radiative forcing’. If the change in temperature is caused by the clouds in the first place, then we could calculate a positive feedback even if it is actually negative!

A new paper by Dessler attempts to get around this and calculate the quick cloud feedback using satellite data since 2000. The satellite measures how much heat is coming from the Earth and Dessler accounts for greenhouse gases, humidity etc to determine how much of the heat flow is from clouds. He then looks at how far above or below the average it is for its month, and plots this against temperature. If the temperature is related to clouds, then you expect a slope in the graph thanks to the equation above. Figure 1 displays the results, and Dessler finds that the short term feedback is , i.e. far more likely to be positive than negative, although negative values can’t be ruled out based on this data.

Dessler notes that aerosol pollution isn't properly accounted for: however, provided there is no net relationship between temperature and aerosols then all it will do is increase the uncertainty but not significantly affect the result.

The author makes one more clever observation: that most of these short term temperature changes are caused by the El Nino Southern Oscillation (ENSO). If the temperature change is being caused by ENSO, then it’s likely not being caused by clouds and Dr Spencer’s hypothesis is potentially sidestepped.

This paper adds confidence that the cloud feedback is not negative and various climate models agree with the observations. However, the author is careful to point out that there are differences between ‘short term’ and ‘long term’ cloud feedbacks in models which suggests that these observations might not be a good guide for the future; but the models did pass this test.

Personally, I think the paper is well thought out but needs more work to rule out Dr Spencer’s hypothesis. It is possible that El Nino affects clouds not just through temperature: perhaps by changing circulation. Further research is required and if this ruled out circulation effects then we could have good confidence that cloud feedbacks is not negative and by extension that the IPCC's most important result of 2-4.5 oC climate sensitivity is correct.

2010-12-11 04:13:25
MarkR
Mark Richardson
m.t.richardson2@gmail...
192.171.166.144

 

Basically I think it's a decent study but we need some more work to ensure that it's not something else like circulation effects. Spencer has a point, but I don't think his certainty about being right has much justification especially when you consider the amount of geological evidence and things like volcanic eruptions which also support the 2-4.5 oC sensitivity.

 

 

One example of how this correlation research could be used in an obviously stupid way:

Hypothesis: warming will lead to more tropical storm activity

Test: compare 10-20 years of data, temperature vs ACE.

 

You might say your results were against the hypothesis because warmer years tended to have fewer hurricanes. This would be because El Nino years are warmer, but El Nino also reverses the Walker circulation which increases wind shear and breaks up storms before they can become hurricanes, leading to lower ACE. It's possible that something like this is happening here and Spencer is right.

 

It would be just another one in the long line of coincidences required to vindicate Spencer: Milankovitch & MWP/MCA calculated sensitivity, that clouds happen to be changing in track with forcing for the past 100 years of instrumental data, that climate models better recreate seasonal, volcanic and regional cloud and temperature changes with positive feedback etc etc.

2010-12-11 05:40:08nice job
dana1981
Dana Nuccitelli
dana1981@yahoo...
38.223.231.252

Good start Mark.  I'd like to add a couple things - how do you feel about doing a co-blog post on the Dessler paper?  I was going to try and draft one up tomorrow, but you've already got a lot of the work done here.

If you're interested, I could use what you've written here as the basis and add a bit of stuff tomorrow.  Then I'd post it in the blog posts forum, and I could email it to you too if you'd like, to get your feedback.  Let me know, and if you want the email, either post your address here or send it to dana1981@yahoo.com

2010-12-11 07:14:03Co-blog, eh?
John Cook

john@skepticalscience...
124.179.115.67
If you guys decide to co-blog, let me know, I may tweak the database to allow two authors to a blog post.

" If the temperature change is being caused by ENSO, then it’s likely not being caused by clouds"

If I understand the exchange between Dessler and Spencer correctly, isn't Spencer asserting that clouds cause ENSO? Dessler discusses this in the RC blog post.

2010-12-11 07:41:28yes
dana1981
Dana Nuccitelli
dana1981@yahoo...
38.223.231.252

I have a hard time understanding Spencer's argument, but as I understand it, he thinks cloud changes form this "internal forcing" which we're not accounting for.  On top of that, he's also hypothesizing that changes in clouds cause changes in ENSO - I'm not sure if he's invoking this same mysterious "internal forcing" again, or if these are two seperate issues.  Maybe Mark can clarify, since he seems to understand Spencer a lot better than me.

Either way Spencer seems to be covering his bases by arguing that clouds cause temp changes and clouds cause ENSO changes, which in turn cause temp changes.  Dessler's response is basically "prove it", since Spencer has nothing but an untested hypothesis to back up his rather aggressive comments.

2010-12-11 08:54:01comment
Robert Way

robert_way19@hotmail...
142.162.16.174
I think its important in the post to point out that lindzen and spencers arguments counteract one another.
2010-12-11 09:27:02do they?
dana1981
Dana Nuccitelli
dana1981@yahoo...
38.223.231.252
Could you explain, Robert?
2010-12-11 10:02:32
nealjking

nealjking@gmail...
91.33.124.42

There is some discussion of this controversy at RealClimate, in a guest-posting by Dessler.

 He links to an email exchange of views between himself and Spencer.

2010-12-11 10:05:21post
dana1981
Dana Nuccitelli
dana1981@yahoo...
38.223.231.252
Yes, I included the exchange in the blog post, which I've drafted up.  I still need to add Dessler's Figure 1 and maybe make a few more tweaks, but I think it's largely done (unless we add something about Lindzen vs. Spencer as suggested by Robert).  Comments would be appreciated, particularly from Mark.
2010-12-11 10:50:08
Riccardo

riccardoreitano@tiscali...
93.147.82.109
A few points
- leave aerosol out, it was important in the paper to say it's not important in the conclusions. In this discussion it's confusing.
- that ENSO causes almost all of the observed T change is essential, not just "one more clever observation". It is the strength of the paper and what allowed Dessler to get around the "reverse causation" invoked by Spencer. As a (welcome) side effect, it forced Spencer to deny the known science on ENSO, a quite uncomfortable position.
- not sure it's worth to add something about Lindzen as Robert says. Anyway, Dessler used basically the same methodology as Lindzen & Choi but correctly applied it globally. If skeptics claim that Spencer is right and Dessler wrong, they have to admit that Lindzen is also wrong. Spencer and Lindzen cannot be both right.

2010-12-11 11:02:23
MarkR
Mark Richardson
m.t.richardson2@gmail...
134.225.187.80

Dana, feel free to take this and edit as you want. Riccardo's point about cutting the aerosols out makes sense. email me and I can send you a few of the references I'd stored up (if I can find my USB stick :p )

m.t.richardson@pgr.reading.ac.uk

 

I think Spencer was trying to assert that it could be internal forcing, but regardless of what he says I'm not sure that it doesn't rule out ENSO -> cloud + T changes, even if it does comfortably avoid clouds->temperatures. Having said that, Dessler clearly points out that he's looking at short term cloud feedback, which is different from long term feedback...

 

 

Also, Riccardo, Dessler doesn't seem to apply the exact same methodology as Lindzen & Choi. L&C picked specific large changes, and fiddled with the time period to minimise the sensitivity they calculated (maybe they didn't fiddle it and it was just a coincidence that they ultimately cherry picked minimising values).

 As I understand Spencer's internal forcing, it's that relatively long term natural changes, such as AMO, NAO, etc can change circulation without changing global mean temperature. This can affect clouds which can then affect temperature. Regional changes can be very important: Milankovitch cycles are driven by redistributing the heat over the surface where albedo changes as a function of position, but they don't change the total insolation (at least, not significantly). If you can redistribute heat and moisture over the Earth then in principle I don't think it sounds impossible to change clouds such that they have a non-zero effect on temperature.

2010-12-11 18:45:10
Riccardo

riccardoreitano@tiscali...
93.147.82.183
Mark
you're right that Dessler and Lindzen and Choi do not use exactly the same methodology, but as far as Spencer criticism (causality and phase space striations) is concerned they both fall into the same broad category.

I'd like to stress one more point already highligted by others (here and at RC). Dessler paper is not on sensitivity but on short term cloud feedback. The first reason is that in the long run clouds respond to many other factors; the only good news here is that model parametrization do a good job. Another is that sensitivity can not be inferred from only 10 years of data; see for example Lin et al 2010.

2010-12-11 22:26:56
MarkR
Mark Richardson
m.t.richardson2@gmail...
134.225.187.80
Yeah, I suppose the onus should be changed to what this does show; current modelled cloud feedback responses are not in disagreement with the data. However, the data range is broad so other values could be accurate.
2010-12-12 11:10:57any more comments?
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.107.107
Okay I updated the article to incorporate the comments so far and add Table 1 from the study.  Any more comments before I publish it?
2010-12-12 12:39:57Looks good
John Cook

john@skepticalscience...
121.222.210.74
Fair, balanced treatment, good to go!
2010-12-12 17:38:46thanks
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.107.107
I strive to meet the stringent standards of Fox News.
2010-12-12 23:15:23
MarkR
Mark Richardson
m.t.richardson2@gmail...
134.225.187.80

Couple of clarifications I think!:

 

The water vapour/albedo feedbacks generally push climate sensitivity to around 2 C (although sometimes higher- there's a few problems with combining them iirc), I had a paper but I left it in the office so I'll send you the reference later.

 

Also, the equation description: the cloud feedback factor is the change in heat flow from clouds per change in temperature not caused by clouds. At least, that's how it's calculated in this paper and a few others in the literature, so it's not the change in temperature from clouds like your phrasing implies!

 

 

2010-12-13 06:30:42got it
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.107.107
So what pushes sensitivity from 2 to 3°C?
2010-12-13 09:13:57
MarkR
Mark Richardson
m.t.richardson2@gmail...
134.225.187.80

I need to re-read the paper, but iirc water vapour/albedo tends to push it into the 2-3 C range, but possible regional changes and cloud changes push us into the higher end. I'll get my references when I get back to the office...

 

 

 

 

I've been thinking a little more about how to explain the difference between short term/long term feedback and I came up with this:

Imagine the world split into 2 regions that start at 10 C and that cloud cover is a function of temperature in Celsius squared. So you start with 100 units of cloud in each region = 200 total.

 

Now say you warm 1 region to 17 C and the other to 11 C, for a mean T=14 C ('El Nino'). You now have 410 total units of cloud, whereas if you boosted both to 13 C ('global warming') you'd have 392 total units of cloud. The two aren't the same and I think this sort of thing is part of the source of the difference between 'short term' and 'long term' cloud feedbacks (as well as changing properties), but I'll try to check up on that.

2010-12-13 13:48:04Clouds argument
James Wight

jameswight@southernphone.com...
112.213.154.195

Dana, do you think you could turn your two posts on cloud feedbacks into a rebuttal of “Clouds provide too much uncertainty”? The site still doesn’t have a rebuttal for that one even though it is one of the key “lukewarmer” arguments.

2010-12-13 15:49:17sure
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.107.107
Yeah totally, except it's one of those pages that doesn't have a URL associated with it.  John will have to fix that first.
2010-12-13 21:19:08
MarkR
Mark Richardson
m.t.richardson2@gmail...
192.171.166.144

Dessler has responded to me about the climate model runs:

 

I analyze the models using exactly the same ENSO climate variations that are in the observations, so it's an apples-to-apples comparison.  As I say in the last few paragraphs of the paper, this does not tell us what the long-term cloud feedback is, but the agreement between the group of models and the observations provides some evidence that models are handling the climate variations correctly.

 

So this is a test of model parametrisation...

 

 

Although there are still potential problems if the models don't reproduce the ENSO patterns properly. A way around this might be to see what models produce as net cloudiness using the "known" pressure/temperature/humidity fields...