Head in a Cloud

troposphere and stratosphere meet blogosphere

Head in a Cloud random header image

Convectively-detrained cloud ice and enhanced WV feedback

October 24th, 2006 by Sean Davis · 1 Comment

An interesting new article came out this week in GRL on clouds and water vapor feedback entitled Does convectively-detrained cloud ice enhance water vapor feedback?, by V. John and B. Soden at the University of Miami.

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.

figure 4 from john and soden
So why is this the case? Shouldn’t models that produce more cirrus clouds in a global warming scenario also have those clouds produce more water vapor (when they evaporate), thereby enhancing UTWV and WV feedback? This may be the case, but the issue is whether it is a significant mechanism compared to the UT moistening from convection of moist air from the warm sea surface… To understand why this is not a significant mechanism, one must recognize the scale differences between IWP and UTWV… In both models and observations by the Aura MLS, the IWP is an order of magnitude less than the UTWV. IWP values are about 30 g m-2, whereas UTWV values are ~300 g m-2. Thus, relative changes in IWP just don’t give the same bang for the buck as do changes in WV under a common constant relative humidity assumption.

Finally, what do the results from this paper mean in the larger scheme of things? As I understand it, one of the implications of this paper is partly that it provides evidence against the adaptive IR iris hypothesis put forth by Lindzen et al. (2001), which claims that increases in convective precipitation efficiency decrease convectively detrained cloud ice, thereby drying the UT as climate warms, and creating a negative (or perhaps diminished) WV feedback.

The results of this paper also seem to point towards a diminished role for cirrus clouds in sensitizing climate. One question I have about this result, though, is what, if any role, could changes in convectively-detrained cloud ice play in regional climate under future scenarios. John and Soden have compared changes in tropical UTWV to the global WV feedback parameter, but what about the possibility that changes in IWP/UTWV affect local climate? I understand that the global WV feedback parameter is more relevant in relation to the IR iris hypothesis, but it would be interesting to know if there are regional effects…

Tags: climate · modeling

1 response so far ↓

  • 1 seand // Oct 30, 2006 at 2:18 pm

    An update to this post from one of the authors, Viju John:


    After I got your mail about the blog, I noticed that Figure 4 in the paper does not show dln(IWP)/dSST, but it shows dln(UTWV)/dSST. Figure showing dln(IWP)/dSST is attached. Please note that the correct figure does not change conclusions of the paper. In fact, the text in paper was written based on the correct figure. This is a mistake which occurred while transferring the files for production. A letter has been sent to GRL editor for taking necessary actions to correct the figure.

    You may also correct this on your web page.

    Sorry for any inconvenience this may have caused.


    The updated figure can be found here

Leave a Comment