Head in a Cloud

troposphere and stratosphere meet blogosphere

Head in a Cloud random header image

Observations of the Fixed Anvil Temperature Hypothesis

March 20th, 2007 by Jamison A. Smith, Ph. D. · 2 Comments

Here’s a paper with observations to support (or possibly refute) the “fixed anvil temperature hypothesis” of Hartman and Larson:

Xu, K.-M. et al, “Statistical analyses of satellite cloud object data from CERS, part II: Tropical convective cloud objects during the 1998 El Nino and evidence for supporting the fixed anvil temperature hypothesis, ” J. Climate 20, 819-842 (2007).


According to the fixed anvil temperature (FAT) hypothesis, radiative cooling of the troposphere controls the height of convection. This radiative cooling is dominated by the abundance of water vapor, which is limited by the saturation vapor pressure (which declines exponentially with declining temperature). Above 200 mb, the atmosphere is so cold that the abundance of water vapor becomes vanishingly small (tens of ppm) and the atmosphere can not cool via radiation. Even with a few degrees of global warming from increased CO2, the altitude at which water vapor declines to the point of being radiatively ineffective won’t change very much. Thus, convective cloud top helights should be insensitive to sea surface temperature (SST), and global warming should not affect tropical convection significantly.

Current observational study

The authors use satellite observations to classify tropical convective clouds into ‘cloud objects’ based on the horizontal extent of the clouds. The cloud objects are labeled S, M or L (for small, medium or large), and the properties of the clouds are analyzed to see if sea surface temperatures (SSTs) affect the properties of the convective clouds.

First major result

L clouds are correlated with higher SSTs. These L clouds have higher cloud top heights, lower cloud top temperatures and radiate less outgoing longwave radiation (OLR) at cloud top.

My comment

Correlation is not causation, but this result would indicate a potential problem with the FAT hypothesis. Clouds with a large horizontal extent are associated with higher SST, so higher SST may produce clouds with a large horizontal extent, which also produce higher, colder cloud tops (and thus amplifying global warming by reducing the cloud-top OLR).

Second major result and validation of FAT

An El Nino event was peaking (high SST) at the beginning of the observational period and dissipated over the next eight months (approaching normal SST). During this time, the changes to the properties of the L group were small. These insignificant changes in L group properties during El Nino dissipation support the FAT hypothesis for the L group.

My comment

The changes in SST between adjacent time segments during the dissipation of El Nino were small as well. This makes the connection between SST and cloud-top OLR hard to detect. The first and last of the five observational periods have very distinct SSTs, and the cloud-top OLR is clearly changed as well, showing increased cloud-top OLR with colder SST. This link between SST and OLR is evidence against the FAT hypothesis.

Additional comments

- Radiative cooling is important for producing atmospheric instability to support convection, but latent heat release ought to be important, too. Warmer surface temperatures mean more latent heat. Keep in mind that the saturation water vapor pressure increases exponentially with increasing temperature. A small change in surface temperature can result in a large change of latent heat.

- Moisture detraining at a given altitude will be increased when following a moist adiabat from a warmer surface. This increased moisture in clear, subsiding air provides a positive feedback on surface temperature by radiating more heat back to the surface. Also, this increased moisture radiates more heats to space, generating increased instability and possibly allowing deeper convection.

- Someone could make a case that the observations in this study refute the FAT hypothesis, but a key part of this discussion would have to include a definition of ‘insensitive’. How large a change is necessary before convection is deemed ‘sensitive’ to global warming?

Tags: climate

2 responses so far ↓

  • 1 seand // Mar 22, 2007 at 7:06 pm

    So Jamie I’m a little confused. It seems as though both of the major points of the paper you outlined above run contra to the FAT hypothesis.
    …Given that the title of the article purports to support FAT, what data in the article are given that actually support FAT?

  • 2 Jamie // Apr 2, 2007 at 12:39 pm

    Sean, the observations from Xu et al [2007] that support the FAT hypothesis are…

    …the properties of the L cloud objects did not change much over time as El Nino dissipated (from one satellite precessional cycle to the next), and the fraction of cloud object pixels in the L catagory is larger during peak of El Nino.

    So, the thinking is that as SSTs rise, the cloud objects should become more-dominated by the L category and their properties are insensitive to the underlying SST.

    If you re-read my initial comments, you can see that I can interpret their graphs and tables entirely differently. I think the changes in the SST pdfs were really small from one precessional cycle to another, thus resulting in really small changes in cloud properties. If you compare two precessional cycles that have very different pdfs of SST, the cloud properties are very different.

Leave a Comment