From Watts Up With That?
By Andy May
The yearly net impact of clouds on outgoing and incoming radiation varies over one W/m2 from year-to-year, according to CERES satellite data. AR6 tells us that cloud feedbacks, to GHG surface warming, are the largest source of uncertainty in their assessment of the warming in the past century as we can see from the AR6 Figure 7.10—our Figure 1.
AR6 page 926, reports that it is very likely global net cloud feedback is positive, but the uncertainty is much higher than for other surface temperature feedbacks. The very likely range they report is -0.1 to 0.94 W/m2/°C. Thus, the total uncertainty is over one W/m2 per degree of surface warming, this estimate is totally model based.
Andrew Dessler provides an observation-based cloud feedback estimate for the period from 2000 to 2010 of 0.54 ±0.7 W/m2/°C. Dessler reports a strong overall or net (Planck + lapse rate + water vapor + albedo + cloud) negative (or cooling) temperature feedback of –1.15 W/ m2/°C for the period with an uncertainty of ±0.88. Thus, the overall uncertainty is only slightly higher than the cloud uncertainty. He emphasizes that the CMIP models have a problem modeling the pattern of cloud feedback.
Figure 1. This is figure 7.10, from AR6, page 979. It shows the simulated feedbacks caused by CO2 emissions calculated for AR6 (red), CMIP6 models (brown) and AR5 (blue).
While cloud cover, which is different from cloud feedback, but related, varies a lot from year to year, it has no statistically valid trend from 2000 to 2021. AR6 claims high confidence that net cloud feedback (NCF) is positive and increases it by 20% over AR5, according to AR6, page 979. A positive cloud feedback means that as surface temperatures warm, clouds increase the warming. While the IPCC have adjusted the cloud feedback parameters to increase the feedback, the uncertainty has not improved, in fact it is worse in CMIP6 than in CMIP5, as shown in Figure 1. The AR6 CMIP6 models increase warming due to clouds, and they increase the model uncertainty and the net uncertainty, not a good sign.
Their explanation of how this was done is presented in pages 974-975 of Chapter 7 and is quite weak. Clouds have a warming effect at night, but reflect sunlight during the day, and on average, they cool the surface. The AR6 claim that they act as an overall positive feedback to surface warming, which is possible, but unlikely since the overall effect is a cooling one. AR6 states that “different types of cloud feedback may occur simultaneously in one cloud regime.” It follows that depending upon the time of day, the location, and local conditions, cloud feedback can be positive or negative for the same clouds. Determining an overall parameter or function to characterize cloud feedback may not be possible since the amount of feedback and the sign of the feedback constantly change.
As local meteorological conditions change, especially temperature, the amount and sign of cloud feedbacks can supposedly change according to AR6. That is, AR6 is reporting that feedbacks, themselves, can have feedbacks. Or, at least, the direction and magnitude of some feedbacks can change as surface temperature changes. This obviously makes predicting future climate changes very difficult. Climate models normally show feedbacks increase (become more positive or warming) as temperatures go up although some show them decreasing (becoming more negative or cooling). AR6 carefully explains that this change in feedbacks with conditions is poorly understood and cannot be quantified.
We also have satellite data that show high-level cirrus clouds in the tropical Pacific provide a negative feedback to sea surface temperatures (SST). Richard Lindzen and his colleagues introduced this concept, which they named the “iris effect,” in 2001. While the debate over the iris effect has been fierce, it has stood the test of time. AR6 discusses it on pages 972-973 and has low confidence it is a negative feedback as Lindzen has always claimed. They manage to present a two-page discussion of the iris effect without once mentioning Richard Lindzen, Ming-Dah Chou, and Arthur Hou, the original authors of the concept.
Lindzen and his colleagues hypothesize that rising surface temperatures increase the speed of the water cycle over oceans and this leaves less water vapor available high in cumulus clouds for the formation of high-level cirrus clouds. Thus, as surface temperatures increase, there are fewer cirrus clouds, and the sky opens like an eye’s iris. This allows more surface radiation to escape to space, cooling it. High-level clouds are very cold, made of ice, and persistent. They do not radiate much energy to space because of their low temperature, but they absorb, and block surface and low-level cloud emitted radiation. Low level clouds are warmer, so they emit more radiation, as well as more effectively block incoming sunlight than cirrus clouds. Thus, logically lower-level clouds are more likely to provide a negative feedback to surface warming, but both Sherwood, et al. and AR6 claim lower level clouds are a net positive feedback to surface warming. The debate on this issue continues, but AR6 does provide an uncertainty range that includes some small negative values, as shown in Figure 1.
Recently, Thorsten Mauritsen and Bjorn Stevens acknowledged that convection efficiency (water cycle speed) is only very crudely represented in current climate models. They programmed a very simple representation of the observed precipitation efficiency/iris effect into their climate model and found it caused the computed ECS to fall to the low end of the IPCC range, near a value of two. They think the models might be missing these important hydrological feedbacks and agree with Lindzen that dry and clear (cloud free) areas in the tropics expand with surface warming. Mauritsen and Stevens showed that including the iris effect in their model moved all model results closer to observations. Other model experiments disagree with Mauritsen and Stevens’ results, so model results of the effect are inconclusive at this time, however the data is not.
The amount of net cloud feedback (NCF) and its long-term trend are uncertain. AR6 is also uncertain about whether NCF is positive or negative, although they state it is likely positive. They acknowledge that even if NCF is positive today, it could still turn negative in the future. And, as Figure 1 shows, the uncertainty in the sign and magnitude of the total feedback to CO2-caused warming is mostly a result of the uncertainty in NCF.
While Lindzen and other researchers have relied on observations to estimate climate sensitivity to CO2, the IPCC relies mostly on models, theory, and climate process analysis for their estimate. AR6 does take into account observation-based studies, such as those by Richard Lindzen, but they are only one of five methods of constraining the range of climate sensitivity estimates. They used a subjective statistical analysis technique to combine all five methods and derive the ECS range provided in the report, but Nic Lewis has shown the methodology (taken from Sherwood, et al.) was flawed.
Paulo Ceppi and colleagues have examined modern climate model estimates of ECS. It turns out that model computed ECS is very dependent upon net cloud feedback (NCF). As we explained above, cloud feedback is only estimated and not known, and it cannot currently be modeled. It can only be “parameterized.” This is a fancy modeling term that translates into English as “assumed” or an educated guess. The relationship between model based ECS and NCF is shown in Figure 2, the data shown in the plot is from Ceppi, et al.
Figure 2. A plot of modeled cloud feedback vs. the AR5/CMIP5 model derived ECS from 28 models. Data source: (Ceppi, Brient, Zelinka, & Hartmann, 2017)
Figures 1 and 2 demonstrate that clouds are the major source of uncertainty in the calculation of ECS and future warming. Further, the R2 in figure 2 suggests that the cloud feedback parameters fed into the models explain 71% of the variability in ECS. Ceppi, et al. write:
“The net cloud feedback is strongly correlated with the total feedback parameter …”(Ceppi, Brient, Zelinka, & Hartmann, 2017).
If the most uncertain model parameter explains 71% of the model result, how confident can we be in the result? The models plotted in Figure 2 are not the same models used in AR6, but similar AR5 models. AR6 acknowledges the dependence of ECS on NCF:
“… CMIP6 models have higher mean ECS and TCR values than the CMIP5 generation of 50 models. They also have higher mean values and wider spreads than the assessed best estimates and very likely ranges within this Report. These higher ECS and TCR values can, in some models, be traced to changes in extra-tropical cloud feedbacks that have emerged from efforts to reduce biases in these clouds compared to satellite observations (medium confidence). The broader ECS and TCR ranges from CMIP6 also lead the models to project a range of future warming that is wider than the assessed warming range”(AR6, p 927).
Steven Koonin reports in his iconic book, Unsettled on page 93 that researchers at the Max Planck Institute tuned their MPI-ESM1.2 climate model to their desired ECS using adjustable cloud feedback parameters. Koonin’s comment on this was “Talk about cooking the books.”
Modeled ECS, projected warming, and modeled TCR are more uncertain in AR6 than in AR5, and the increased uncertainty is due to net cloud feedback uncertainty. The uncertainty in the impact of clouds has increased due to efforts to reduce bias in clouds relative to satellite observations. Clouds cannot currently be modeled, so the changes were caused by manual changes to adjustable model parameters. The parameters increased the assumed, but unknown, positive cloud feedback, which increased the strongly correlated modeled ECS. Then the models produced both a higher ECS and higher projected future warming. Big surprise. It looks like the IPCC is manufacturing a desired result, and not doing it very convincingly.
The bibliography can be downloaded here.
- May, A. (2021c, April 28). Clouds and Global Warming. From CERES data. Link. ↑
- AR6, Chapter 7, page 975 ↑
- Dessler, A. E. (2013). Observations of Climate Feedbacks over 2000-10 and Comparisions to Climate Models. J of Climate, 333-342. ↑
- May, A. (2021c, April 28). Clouds and Global Warming. Link. Figure 5. ↑
- AR6, page 979. Cloud feedback is 20% larger in AR6 than in AR5. On the same page, AR6 admits this why modelled ECS is larger in AR6 than in AR5. ↑
- (May, 2021c) and Ceppi, P., Brient, F., Zelinka, M., & Hartmann, D. (2017, July). Cloud feedback mechanisms and their representation in global climate models. WIRES Climater Change, 8(4). ↑
- AR6, page 975 ↑
- AR6, 975, 979, 980 ↑
- (IPCC, 2021, p. 981) ↑
- Lindzen, R., & Choi, Y.-S. (2021, April 1). The Iris Effect: A Review. Asia-Pacific Journal of Atmospheric Sciences. ↑
- Lindzen, R., Chou, M.-D., & Hou, A. (2001, March). Does the Earth have an Adaptive Iris. Bulletin of the American Meteorological Society, 82(3). ↑
- Mauritsen, T., & Stevens, B. (2015). Missing iris effect as a possible cause of muted hydrological change and high climate sensitivity in models. Nature Geoscience, 8, 346-351. ↑
- Lewis, N. (2022). Objectively combining climate sensitivity evidence. Climate Dynamics. ↑
- Ceppi, P., Brient, F., Zelinka, M., & Hartmann, D. (2017, July). Cloud feedback mechanisms and their representation in global climate models. WIRES Climater Change, 8(4). ↑