I thought this post on clouds and climate modeling below from Steve McIntyre’s Climate Audit was interesting, because it highlights the dreaded “negative feedbacks” that many climate modelers say don’t exist. Dr. Richard Lindzen highlighted the importance of negative feedback in a recent WUWT post.
One of the comments to the CA article shows the simplicity and obviousness of the existence of negative feedback in one of our most common weather events. Willis Eschenbach writes:
Cloud positive feedback is one of the most foolish and anti-common sense claims of the models.
This is particularly true of cumulus and cumulonimbus, which increase with the temperature during the day, move huge amounts of energy from the surface aloft, reflect huge amounts of energy to space, and fade away and disappear at night.
Spot on Willis, I couldn’t agree more. This is especially well demonstrated in the Inter Tropical Convergence Zone (ITCZ) The ITCZ has been in the news recently because early analysis of the flight path of Air France 447 suggests flying through an intense thunderstorm cell in the ITCZ may have been the fatal mistake. There is a huge amount of energy being transported into the upper atmosphere by these storms.
Here are some diagrams and photographs to help visualize the ITCZ heat transport process. First, here is what the ITCZ looks like from space. Note the bright band of cumulonimbus clouds from left to right.
Here is a pictorial showing a cross section of the ITCZ with a cumulonimbus cloud in the center.
And finally, a 3D pictorial showing ITCZ circulation and heat transport. Note the cloud tops produce a bright albedo, reflecting solar radiation.
And here is the post on Climate Audit
Cloud Super-Parameterization and Low Climate Sensitivity
“Superparameterization” is described by the Climate Process Team on Low-Latitude Cloud Feedbacks on Climate Sensitivity in an online meeting report (Bretherton, 2006) as:
a recently developed form of global modeling in which the parameterized moist physics in each grid column of an AGCM is replaced by a small cloud-resolving model (CRM). It holds the promise of much more realistic simulations of cloud fields associated with moist convection and turbulence.
Clouds have, of course, been the primary source of uncertainty in climate models since the 1970s. Some of the conclusions from cloud parameterization studies are quite startling.
The Climate Process Team on Low-Latitude Cloud Feedbacks on Climate Sensitivity reported that:
The world’s first superparameterization climate sensitivity results show strong negative cloud feedbacks driven by enhancement of boundary layer clouds in a warmer climate.
These strong negative cloud feedbacks resulted in a low climate sensitivity of only 0.41 K/(W m-2), described as being at the “low end” of traditional GCMS (i.e. around 1.5 deg C/doubled CO2.):
The CAM-SP shows strongly negative net cloud feedback in both the tropics and in the extratropics, resulting in a global climate sensitivity of only 0.41 K/(W m-2), at the low end of traditional AGCMs (e.g. Cess et al. 1996), but in accord with an analysis of 30-day SST/SST+2K climatologies from a global aquaplanet CRM run on the Earth Simulator (Miura et al. 2005). The conventional AGCMs differ greatly from each other but all have less negative net cloud forcings and correspondingly larger climate sensitivities than the superparameterization
They analyzed the generation of clouds in a few leading GCMs, finding that a GCM’s mean behavior can “reflect unanticipated and unphysical interactions between its component parameterizations”:
A diagnosis of the CAM3 SCM showed the cloud layer was maintained by a complex cycle with a few hour period in which different moist physics parameterizations take over at different times in ways unintended by their developers. A surprise was the unexpectedly large role of parameterized deep convection parameterization even though the cloud layer does not extend above 800 hPa. This emphasizes that an AGCM is a system whose mean behavior can reflect unanticipated and unphysical interactions between its component parameterizations.
Wyant et al (GRL 2006) reported some of these findings. Its abstract stated:
The model has weaker climate sensitivity than most GCMs, but comparable climate sensitivity to recent aqua-planet simulations of a global cloud-resolving model. The weak sensitivity is primarily due to an increase in low cloud fraction and liquid water in tropical regions of moderate subsidence as well as substantial increases in high-latitude cloud fraction.
They report the low end sensitivities noted in the workshop as follows:
We have performed similar +2 K perturbation experiments with CAM 3.0 with a semi-Lagrangian dynamical core, CAM 3.0 with an Eulerian dynamical core, and with the GFDL AM2.12b. These have λ’s of 0.41, 0.54, and 0.65 respectively; SP-CAM is about as sensitive or less sensitive than these GCMs. In fact, SPCAM has only slightly higher climate sensitivity than the least sensitive of the models presented in C89 (The C89 values are based on July simulations)…
The global annual mean changes in shortwave cloud forcing (SWCF) and longwave cloud forcing (LWCF) and net cloud forcing for SP-CAM are _1.94 W m_2, 0.17 W m_2, and _1.77 W m_2, respectively. The negative change in net cloud forcing increases G and makes λ smaller than it would be in the absence of cloud changes.
Wyant et al (GRL 2006) is not cited in IPCC AR4 chapter 8, though a companion study (Wyant et al Clim Dyn 2006) is, but only in the most general terms, no mention being made of low sensitivity being associated with superparameterization:
Recent analyses suggest that the response of boundary-layer clouds constitutes the largest contributor to the range of climate change cloud feedbacks among current GCMs (Bony and Dufresne, 2005; Webb et al., 2006; Wyant et al., 2006). It is due both to large discrepancies in the radiative response simulated by models in regions dominated by lowlevel cloud cover (Figure 8.15), and to the large areas of the globe covered by these regions…
the evaluation of simulated cloud fi elds is increasingly done in terms of cloud types and cloud optical properties (Klein and Jakob, 1999; Webb et al., 2001; Williams et al., 2003; Lin and Zhang, 2004; Weare, 2004; Zhang et al., 2005; Wyant et al., 2006).
(Bretherton 2006)
Dessler et al (GRL 2008) made no mention of strong negative cloud feedbacks under superparamterization, stating that sensitivity is “virtually guaranteed” to be at least several degrees C, unless “a strong, negative, and currently unknown feedback is discovered somewhere in our climate system”:
The existence of a strong and positive water-vapor feedback means that projected business-as-usual greenhouse gas emissions over the next century are virtually guaranteed to produce warming of several degrees Celsius. The only way that will not happen is if a strong, negative, and currently unknown feedback is discovered somewhere in our climate system.
There are a limited number of possibilities for such a possibility, but it is interesting that cloud super-parameterizations indicate a strong negative cloud feedback (contra the standard Soden and Held results.)
This is not an area that I’ve studied at length and I do have no personal views or opinions on the matters discussed in this thread.