Head in a Cloud

troposphere and stratosphere meet blogosphere

Head in a Cloud random header image

Still a thorn in our side / Positive cloud feedback?

July 17th, 2006 by Sean Davis · 8 Comments

Cloud feedback still largest source of uncertainty in GCM’s
What did they say? Clouds a positive feedback?

Last week, a new article by Brian Soden of the University of Miami and Isaac Held of GFDL entitled “An Assessment of Climate Feedbacks in Couopled Ocean-Atmosphere Models” was published in the Journal of Climate. Among their findings were that the representation of clouds and their associated feedbacks within GCM’s still presents one of the largest sources of uncertainty in current predictions of climate sensitivity. In this paper, Soden and Held compare the most predominant feedback mechanisms in a group of state-of-the-art atmosphere-ocean GCM’s (GFDL, GISS, CCSM3, HADCM3, etc.). Although in many ways this study is a confirmation of what is already known (that clouds are still a thorn in the side of modelers), there were several interesting take home points I found from this article.


1. This study was one of the first (or at least, one of few) to incorporate a methodology for calculating feedbacks in a consistent manner for each of the various models. Previous work has largely focused on gathering feedback estimates from disparate parts of the scientific literature. Soden and Held find that

“…the range of feedback strengths computed here is smaller [than previous studies] for all feedbacks except clouds.”

and provide some evidence that reduction in feedback strength difference between models is due to inconsistencies in previous studies, not an improvement of model physics.
2. The standard deviation of cloud feedback strength among the various GCM’s is about 4 times greater than the next largest standard deviation, which is the lapse rate/water vapor feedback (the motivation for combining the lapse rate and water vapor feedbacks is explained in the article. In short, the lapse rate feedback affects the water vapor feedback, so combining them is more appropriate).
3. Cloud feedbacks are positive in all models !!!??? Ok… so I’m embarrassed to admit this, but before reading this article I didn’t know that clouds represented a positive feedback in climate models. Is there anyone else out there that was under the impression that clouds were a negative feedback? Any student who has ever taken a class in weather/climate has learned about the cloud-albedo feedback, where the presence of bright (high albedo) clouds reduces the TOA radiative forcing by reflecting sunlight back to space. I have always been under the impression that because of this, clouds are a negative feedback (i.e., warmer Earth -> more clouds -> cooling effect). So why does this simplistic explanation fail? Well, it must be wrong. To get a positive feedback, either warmer Earth -> LESS clouds -> warming effect, or warmer Earth -> more clouds -> warming effect. There is no evidence of which I’m aware that cloud coverage/amount will decrease in a warmed climate. I would be glad to see evidence in the peer-reviewed literature either way, if anyone would like to comment. So if it is the latter of these two methods for getting a positive cloud feedback that is correct, how do we reconcile this with the cloud-albedo effect? I assume that the answer lies primarily in the fact that clouds have a warming effect at night by trapping outgoing longwave radiation (OLR). Because the cloud feedback is a diurnal (and annual and geographical, for that matter) average, it is possible that other effects outweigh the cloud-albedo effect to produce a positive cloud feedback. Nonetheless, from an educational standpoint, I am a little stunned by why this is not made more clear in undergraduate/graduate classes. This was not clear at all to me as a Ph.D. candidate in the subject, and very certainly is not made clear (and likely often mistaught!) in the undergraduate classes. Perhaps it needs to be emphasized in educational settings that the NET cloud feedback is positive, even though some of the contributing feedbacks are strongly negative.
4. Finally, at the end of the article Soden and Held speculate that intermodel differences in low cloud coverage account for most of the discrepancies in cloud feedback, although they note that their methodology is unable to identify the source of these discrepancies. As someone who studies cirrus clouds, I’ve always liked to think that it is my type of clouds that are the most important to understand in order to improve GCM predictions of climate sensitivity :) … The speculation of Soden and Held, which cites a 2005 GRL paper by Bony and Dufresne and another by Webb et al. (in press, Climate Dyn.), is that it is in fact marine boundary layer clouds. Oh well… perhaps the authors or some other group will care to take this study a step further and quantify the portion of the discrepancy that can be attributed to various cloud types, or overage/amount. If cirrus don’t end up being the most important, I can always fall back on the old line that they are most important per weight

Tags: modeling

8 responses so far ↓

  • 1 Steven Howell // Jul 17, 2006 at 12:49 pm

    I have a vague (and probably outdated) recollection that low clouds have a cooling effect while high clouds tend to warm. That’s because their effect on outgoing longwave radiation is different. Both efficiently absorb upwelling IR, but low clouds (like marine stratocumulus) are close to surface temperatures and “reradiate” much like the ocean surface while high clouds are cold and thus radiate less out to space.

    I just checked Seinfeld and Pandis (1998). They say the same.

  • 2 Gannet Hallar // Jul 17, 2006 at 1:08 pm

    Also don’t forget about the semi-indirect effect, which is important for marine boundary layer clouds. The semi-direct effect is a consequence of the direct effect of absorbing aerosols, it changes cloud properties in response to aerosols.

    Check out —

    Global indirect aerosol effects: a review
    U. Lohmann and J. Feichter
    Atmos. Chem. Phys., 5, 715–737, 2005

    This is the best paper that I have ever seen regarding a clear explanation of the many indirect effects of clouds and aerosols.

  • 3 seand // Jul 17, 2006 at 1:42 pm

    Note: The above paper can be found online at
    http://irina.eas.gatech.edu/EAS_spring2006/Lohmann2005.pdf

  • 4 seand // Jul 17, 2006 at 10:05 pm

    Brian Soden was kind enough to write me this email regarding this post:

    Hi Sean – Thanks for your letter and for highlighting our paper on your blog. You’ve done a nice job of summarizing the main points of the paper.

    You are not alone in your surprise regarding the sign of cloud feedback. I think much of the confusion stems from the tendency to interpret changes
    in cloud forcing as “cloud feedback”. As we show in the paper, the change in cloud forcing for any particular model is well correlated with
    its cloud feedback. But because the change in cloud forcing depends on the changes in both clear sky and total sky fluxes, the two quantities are not
    the same. In fact, many models with (weak) positive cloud feedback actually simulate a reduction in net cloud forcing (which tend to get interpreted as a negative cloud feedback).

    As far as the role of low versus high clouds, I believe that the TOA fluxes are more sensitive to changes in low clouds, because there is much
    less cancellation of their SW and LW effects than for high clouds. On the other hand, the net radiative fluxes at the surface can be more sensitive
    to changes in high cloud than to low cloud for the same reasons, i.e. at the surface, the LW and SW effects of low clouds tend to be more offsetting than for high clouds. The change in radiative fluxes at the surface are important because they determine the rate of change in global precipitation. Thus, to first order, low clouds tend to be more important
    for climate sensitivity, whereas high clouds tend to be more important for the hydrologic sensitivity … at least that is my working hypothesis.

    Brian

    He also provides a link to a paper which discusses the difference between cloud feedback and a change
    in cloud forcing in more detail.

    http://www.gfdl.noaa.gov/reference/bibliography/2004/bjs0402.pdf

  • 5 seand // Aug 4, 2006 at 3:53 pm

    Brian recently wrote a blog entry on RealClimate, entitled “Climate Feedbacks”, that discusses some of these points in his paper.

  • 6 Ferdinand Engelbeen // Aug 27, 2006 at 2:15 pm

    Sorry for the late drop-in, I just learned about this blog today. Even missed the RealClimate discussion, as I was on a trip to Iceland…

    Simultaneous sent to RealClimate and ClimateAudit:

    There are a few recent studies about cloud behaviour which challenge the positive cloud feedback included in (near all) current climate models.

    Chen and Wielicki (2002) observed satellite based cloud changes in the period 1985-2000, where an increasing SST (+0.085 C/decade) was accompanied with higher insolation (2-3 W/m2), but also higher escape of heat to space (~5 W/m2), with as net result 2-3 W/m2 TOA loss to space for the 30N-30S band. This was caused (or accompanied) by faster Walker/Hadley cell circulation, drying up of the upper troposphere and less cirrus clouds.

    In 2005, these findings were expanded by J. Norris with surface based cloud observations in time (from 1952 on for clouds over the oceans, from 1971 on over land) and latitudes. There is a negative trend for upper-level clouds over these periods of 1.3-1.5%. As upper-level clouds have a warming effect, this seems to be an important negative feedback.

    J. Norris has a paper in preparation about cloud cover trends and global climate change.
    On page 58, there is a calculation of cloud feedback, assuming that the observed change in cloud cover is solely a response to increased forcing. The net response is -0.8, which is a very strong negative feedback… Of course this is the response, if nothing else is influencing cloud properties/cover, but important enough for further investigation.

    Even internal oscillations, like an El Nino (1998) leads to several extra W/m2 more net loss of energy to space, due to higher sea surface temperatures. Thus IMHO, if models include a (zero, small or large) positive feedback by clouds, they are not reflecting reality.

    In addition: clouds act as a positive feedback for solar radiation: the small change in TOA radiation is negatively correlated with low cloud cover, see Fig.1 of Kristjansson ea..
    This means that solar forcing has a different (and higher) effect on temperature than an equivalent forcing by GHGs. Which is not incorporated in any model (to the contrary, the HadCM3 model probably underestimates solar effects with a factor 2)…

  • 7 Ferdinand Engelbeen // Sep 18, 2006 at 10:55 am

    There is a recent correction of the ERBS satellite data (see Wong ea.) which does change the radiation budget in sign…
    Instead of the original 3.1/-2.4/-0.7 W/m2/decade (over the period 1985-2000) for LW/SW/net radiation (TOA), after correction of the data, this is changed to 0.7/-2.1/1.4 W/m2/decade for the 1990s vs. the 1980s and for the 20N-20S equatorial band.

    There still are a lot of questions left, like the difference of ~5 W/m2 between ERBS (1999) and CERES satellites which operate since 2000, with a gap of one year, which prevents intercalibration. The consequences for the radiation budget for the subtropics are not given. For the period 1990-2000, the radiation budget should equal the observed increase in ocean heat content (I wonder how that can be, as the satellites don’t cover the highest latitudes). Further, that still doesn’t explain the observed reduction in ocean heat content 1980-1990. And last but not least, the changes in radiation budget still are an order of magnitude higher than what can be expected from the increase in GHGs in the period of interest (be it now with a positive sign)… Thus this is a positive cloud feedback for increasing SST (or is it the cause of higher SSTs?) in the tropics?

    This doesn’t change the findings of J. Norris, who found a global decrease of 1.3% in upper-level clouds since 1952 over the oceans. Which normally indicates more cooling (the results for low-level clouds were ambiguous). But he too used the previous ERBS radiation budget data to calculate the overall radiation budget change.

    All together, quite confusing that a small error in the satellite data can change everything and invalidate any theory… We are obviously in need of better instruments which are far better than current, to detect the real changes in radiation balance which we need to know…

  • 8 Head in a Cloud » Convectively-detrained cloud ice and enhanced WV feedback // Oct 24, 2006 at 12:12 pm

    [...] Well, to cut to the chase, the answer is NO, according to this article. John and Soden show that despite strong positive correlations between cirrus ice water path (IWP), upper tropospheric water vapor (UTWV), and sea surface temperature (SST), the water vapor feedback parameter, λH2O (see previous post on WV feedback, here), is insensitive to the slope of the relation between IWP and SST. One might expect that models that produce more ice with increasing SST (i.e. greater slope, d ln(IWP)/dSST ) should have a water vapor feedback that is enhanced relative to models that don’t produce as much ice with increasing SST. John and Soden show that this just isn’t the case (figure 4) — their “crux” figure shows that there is no correlation between the strength of the water vapor feedback and the slope of the IWP vs. SST relation. [...]

Leave a Comment

*