Guest commentary from Andy Dessler (TAMU)
When a new scientific hypothesis is published, two questions always occur to me:
- Did the authors convincingly show the hypothesis was correct?
- If not, is the hypothesis actually correct?
The answers to these two questions may not be the same. A good example is Wegener’s theory of continental drift — his idea was fundamentally correct, but he lacked the data and physical mechanisms to convince the rest of scientific community. It would take several decades before enough data were gathered that the scientific community wholeheartedly endorsed plate tectonics.
In 2001, Prof. Richard Lindzen and colleagues published his “iris hypothesis” (Lindzen et al., 2001). The hypothesis has two parts: First, in a warmer climate, enhanced precipitation efficiency will lead to less cloud being detrained into the troposphere from convection. Second, with less cloud cover, more infrared radiation can escape to space, thereby creating a strong climate-stabilizing negative cloud feedback that prevents significant warming from increasing greenhouse gases.
Within a few years, a number of analyses made clear that the evidence provided by Lindzen et al. had problems [e.g., Hartmann and Michelsen, 2002; Lin et al., 2002; Lin et al., 2004; Su et al., 2008]. Lindzen and colleagues responded to these critiques, but few were convinced by their arguments. By 2006, when I submitted an analysis of tropospheric water vapor that investigated whether there was an iris in that, one of the reviewers pointedly questioned why anyone was still working on this issue. I subsequently withdrew the paper.
Nevertheless, just because Lindzen et al. did not convincingly demonstrate their case does not mean the iris hypothesis is wrong. With that idea in mind, a new paper by Mauritsen and Stevens (2015) revisits the iris hypothesis. The most important part of their work was to simulate the iris in a climate model by artificially tweaking the model’s convective parameterization. They do this by increasing the rate of conversion of cloud water to rain as the climate warms, thereby reducing the amount of detraining condensate in a warmer climate. In effect, this imposes a tweak that mimics the iris effect – it is not a demonstration that the iris effect emerges from any physical mechanisms.
What they find is that, even though cloud cover is reduced as the climate warms, it does not generate a strong negative cloud feedback. While reducing cloud cover does indeed let more infrared energy out, it also lets more sunlight in. These two effects, while independently large, act in opposite directions. The net effect is the small residual of their difference. For runs with the strongest “iris”, the model’s climate sensitivity is reduced from 2.8°C for doubled carbon dioxide to 2.2°C — still well within the IPCC’s canonical range.
It’s also worth pointing out what this study doesn’t prove. It doesn’t validate Lindzen et al.’s original hypothesis — in fact, it does the opposite – even with an iris effect, the sensitivity does not become negligible. Additionally, there is little evidence that the rate of conversion of cloud water to rain actually changes with temperature, although Mauritsen and Stevens show that incorporating the iris into the model does improve the model’s simulations of some aspects of the climate system (even though it doesn’t change climate sensitivity much).
I view this as a what-if calculation of the impact of such a process. Future research may validate this, or it may not. This kind of calculation is one of the reasons why we like using models, of course.
Another argument against the iris comes from my work looking at the cloud feedback in response to short-term climate variability. If the iris provided a strong negative feedback, then we would expect to see it in response to short-term climate fluctuations. Analysis of observations doesn’t show anything like that (Dessler, 2013).
Overall, I think the debate over the iris hypothesis is a testament to the efforts the scientific community goes through to evaluate challenges to theories and find ways to improve our understanding of the climate (for instance, see Bill Ruddiman’s post from last week). This is one of the most important reasons I have such high confidence in the scientific process for figuring out how the universe works.
- R.S. Lindzen, M. Chou, and A.Y. Hou, "Does the Earth Have an Adaptive Infrared Iris?", Bulletin of the American Meteorological Society, vol. 82, pp. 417-432, 2001. http://dx.doi.org/10.1175/1520-0477(2001)082<0417:DTEHAA>2.3.CO;2
- D.L. Hartmann, and M.L. Michelsen, "No Evidence for Iris", Bulletin of the American Meteorological Society, vol. 83, pp. 249-254, 2002. http://dx.doi.org/10.1175/1520-0477(2002)083<0249:NEFI>2.3.CO;2
- B. Lin, B.A. Wielicki, L.H. Chambers, Y. Hu, and K. Xu, "The Iris Hypothesis: A Negative or Positive Cloud Feedback?", Journal of Climate, vol. 15, pp. 3-7, 2002. http://dx.doi.org/10.1175/1520-0442(2002)015<0003:TIHANO>2.0.CO;2
- B. Lin, T. Wong, B.A. Wielicki, and Y. Hu, "Examination of the Decadal Tropical MeanERBSNonscanner Radiation Data for the Iris Hypothesis", Journal of Climate, vol. 17, pp. 1239-1246, 2004. http://dx.doi.org/10.1175/1520-0442(2004)017<1239:EOTDTM>2.0.CO;2
- H. Su, J.H. Jiang, Y. Gu, J.D. Neelin, B.H. Kahn, D. Feldman, Y.L. Yung, J.W. Waters, N.J. Livesey, M.L. Santee, and W.G. Read, "Variations of tropical upper tropospheric clouds with sea surface temperature and implications for radiative effects", Journal of Geophysical Research, vol. 113, 2008. http://dx.doi.org/10.1029/2007JD009624
- T. Mauritsen, and B. Stevens, "Missing iris effect as a possible cause of muted hydrological change and high climate sensitivity in models", Nature Geoscience, vol. 8, pp. 346-351, 2015. http://dx.doi.org/10.1038/ngeo2414
- A.E. Dessler, "Observations of Climate Feedbacks over 2000–10 and Comparisons to Climate Models*", Journal of Climate, vol. 26, pp. 333-342, 2013. http://dx.doi.org/10.1175/jcli-d-11-00640.1
33 Responses to "The return of the iris effect?"
Tony Noerpel says
Can it be argued given the paleoclimate evidence for abrupt climate changes that there is likely no strong negative feedback over any meaningful time scale?
richard pauli says
Thanks for this…It helped with my understanding to find a brief description of the iris effect. http://en.wikipedia.org/wiki/Iris_hypothesis
Andrew Dessler says
Tony: Yes, climate sensitivity from the paleoclimate evidence is inconsistent with a strong negative cloud feedback.
Chris Colose says
The issue with the Mauritsen and Stevens piece is that it tries to go well beyond a “what if” modeling experiment, and attempts to make contact with a lot of other issues related to historical climate change (the hiatus, changes in the hydrologic cycle, observed tropical lapse rate “hotspot” stuff, changes in the atmsopheric circulation, etc) by means of what the “iris” should look like in other climate signals.
There’s nothing inherently wrong with this, but none of these topics (in my view) require a critical, zeroth-order silver bullet to reconcile observations, models, and theory, at least beyond what has so far been extensively discussed in the literature (observational issues, natural variability, etc). So the whole motivation of the article isn’t compelling and makes it very misleading. Even the title makes no sense- how can an effect explain high climate sensitivity in models if removing it only reduces sensitivity by a few tenths of a degree, or in some models, maybe increases it? What’s more, the issue of convective self-aggregation (as far as I understand) is a bit removed from Lindzen’s original idea, even if it provides an “iris-like” longwave feedback at the expense of different shortwave feedbacks.
I guess the article will be successful in that it generates discussion and probably will be cited quite a bit. I think they crammed too much in there though and tried to come out with some profound, unifying idea about climate.
Thanks for this. Is the iris effect going to turn into yet another ‘zombie lie’? https://www.youtube.com/watch?v=MPQDqNGtycM
When you say, “incorporating the iris into the model does improve the model’s simulations of some aspects of the climate system,” could you be clearer on what ‘aspects’ are improved? Does this require cherry picking what parts of the model or what timescales the iris effect is applied to?
Jim Eager says
This Tenberth et al paper currently in press at the Journal of Geophysical Research is highly relevant:
Climate variability and relationships between top-of-atmosphere radiation and temperatures on Earth
Roberto Rondanelli says
I am glad that A. Dessler has continued working on the subject (Dessler, 2013) despite the apparent discouragement he suffered in 2006. This is hardly a settled issue in my opinion, many of the observational effects are “up for grabs” for a serious scientist to look over the data. For instance, the issue of the precipitation efficiency was taken up by us in a relatively simple fashion with the data available at the time (Rondanelli and Lindzen, 2008). What would be an ideal dataset to work this out? Something we don´t have yet (or ever) a continuous radar data over the tropical oceans for instance, in which we could study the statistical properties of the mesoscale convective systems over the oceans, calculate the water budget, including cloud detrainment and radiative effects. Our study has not been followed up by experts in the observational community, and even if there is no Iris it would be interesting to know what is the sea surface temperature dependence of the precipitation efficiency in tropical clouds. New GPM data could shed an new light on this issue alone.
For me, Mauritsen and Stevens´paper show that the Iris should be taken seriously, even if it does not happen in the original form suggested by Lindzen et al (2002) and more as a convective aggregation with increasing SST.
Jai Mitchell says
My impression of this is that there is a concerted effort to place a body of poor science within the published record that allows for a “second track” of study. It is a kind of scientific conspiracy where key participants are funded to produce poorly substantiated arguments, that are then passed through complicit editors and reviewers so that it enters the official body of knowledge.
Then future work can be referenced to these poor arguments and a second track is developed with layers of argument, all predicated on assumption, shoddy science and even outright falsification.
This was the strategy of the tobacco industry, proponents of methyl lead as a fuel additive and producers of CFCs and (I believe) is still a strategy today.
Pat Cassen says
Andy (or anyone else with full access) – The abstract states that “Inferences from the observational record…indicate that models underestimate some of the changes in the hydrological cycle.”
What observations specifically are they referring to? (Wentz et al., 2007 inferred a stronger-than-modeled hydro cycle, but I don’t see that work in the references.)
Thanx for the clear summary. What do you think of the proposal in Mauritsen (2015) that convective aggregation is a possible physical mechanism for the iris ?
“Something we don´t have yet (or ever) a continuous radar data over the tropical oceans for instance, in which we could study the statistical properties of the mesoscale convective systems over the oceans, calculate the water budget, including cloud detrainment and radiative effects. ”
Isnt TRMM supplying some now ?
Thorsten Mauritsen says
Dear Andy, thanks a lot for a very nice perspective on our paper, which I truly enjoyed reading. I hope that in the future we can have an open-minded discussion on this topic, your 2006-experience points to the possibility that it has not always been like that. We should allow ourselves, and our colleagues, to turn every stone on this.
I would like to point out that even with an iris-effect you can get a high ECS. Andrew Gettelman repeated the experiment with CAM5 and actually got over-compensation from shortwave, i.e. an increase of ECS. This is written in the paper. Thus, a sizeable ECS does not invalidate the core of the iris-hypothesis; that dry and clear regions of the tropics could expand under warming. It simply means that the original estimates of the impacts on ECS were overly simplistic, if not exaggerated. We go further and argue that if shortwave cloud feedback compensation is positive, then ECS is most likely above 1.5 K. We cannot address the upper bound in this study.
So why would anyone care about the possible existence of an iris-effect if it might not alter ECS, say, beyond our sincere wish to understand how nature works? We point to two reasons:
1) Attempts to constrain ECS with shortwave cloud feedback alone (e.g. Sherwood et al. 2014) miss the possibility that real-world longwave feedbacks could lie outside that spanned by the model ensemble (the infamous unknown unknowns). This means that potentially these approaches can be reconciled with ECS estimates from the historical record.
2) It is exceptionally difficult to reconcile the high observed hydrological sensitivity, based on various independent but incomplete observing systems, with present modeling (see an excellent paper on this topic that I unfortunately missed citing: http://link.springer.com/article/10.1007%2Fs00382-014-2174-9). It is difficult to think of another plausible mechanism whereby hydrological sensitivity could be high.
The puzzle we lay out in the paper is thus solvable with an iris-effect. That does not mean there couldn’t be any other solutions to the riddle.
Concerning paleoclimate (I assume you refer to PETM), this was not a topic of our paper. However, since I do have a habit of torturing our model with high CO2 (http://onlinelibrary.wiley.com/doi/10.1002/2013GL058118/abstract) I of course also did this for ECHAM6-Iris. It turns out to have even more non-linearity than ECHAM6, such that at 16x pre-industrial CO2 temperatures have risen more than in ECHAM6. These are things I would like to understand better, but my working hypothesis is that the model simply runs out of high-level cloudiness so the iris-effect stops working at high temperatures.
Pete Dunkelberg says
Thorsten Mauritsen, thanks for your explanations at #12 (and thanks to Andy Dessler for the top post). This may give you an idea of how to get some data of interest:
Scientists Turn to Drones For Closer Look at Sea Ice.
jai mitchell says
How well did your models capture Indonesian lower and mid troposphere point source aerosol deposition from burning peat fires? The 300mb humidity anomalies in this area are often astounding as shown here: http://goo.gl/LhS23W (ESRL 300 mb relative humidity anomaly map 7/22/13-8/30/13 )
As you can see, the lower altitude haze is significantly suppressing raincloud formation during this period.
Alex Harvey says
Dear Andy, it would be very interesting to also invite Prof. Lindzen himself to write up his thoughts on the new paper in a guest post.
Sharon Hawkins-Fauster says
I was stunned to read that the warmer it is the less cloud cover there is. As an amateur astronomer, I’m having trouble finding a cloud-free night to use my telescope in the last few years. I’ve assumed that the warming oceans are putting more and more water into the atmosphere thereby producing more low-level clouds. I know that if I were to stay up very late into the early morning hours there are fewer clouds due to the condensation with cooler temperatures leading to dew on the ground, but at 75 years, I need my sleep. I live in Upper Franconia in Bavaria where there are many carp fish ponds, lakes and rivers which are warming, too. I studied astronomy, physics and chemistry more than 45 years ago, so I’m not totally ignorant of the science. Please explain why I’m seeing more and more cloudy nights, and that’s a fact.
Mal Adapted says
You do know that anecdotes aren’t data, don’t you? Before an explanation can be offered for your alleged fact, you first need to show that it is a fact.
You may be able to do that, if there are reliable weather records for your location that include either cloudiness or hours of sunshine.
Elsewhere on the dogma front, the Heartland Institute has embarked on a pilgrimage to Rome to re-educate the Pope on the evils of climate modeling .
Hank Roberts says
I don’t have time or expertise myself to explain this to you, but I put your query into Google
and on the first page (out of ‘About 6,670 results’)
So — just from that much — the suggestion above from Mal seems apt. You can probably find records for your specific location collected over enough decades to be useful in verifying your observation.
Getting to _why_ will take looking at rather more. Has your area industrialized or had other changes in land cover and land use in 40 years? Are there sources of cloud condensation (burning coal, for example)?
I know my area of N. America has a pattern of more bad weather on weekends. That used to be said as a joke. Someone did the statistics and it turned out to be an accurate observation made over time by many people, each thinking it was personal bad fortune.
Thorsten Mauritsen says
@ Pete Dunkelberg, I am convinced drones will revolutionize how we collect observations, although at the moment I wouldn’t know how to use them in these questions, i.e. it is a bit of a solution looking for a problem. That is not to say there aren’t many useful applications of drones such as the one you link to. My brother pioneered using drones in archeology in Denmark.
@ Jai Mitchell, ECHAM does not simulate aerosols, they are prescribed (http://onlinelibrary.wiley.com/doi/10.1002/jame.20015/abstract).
@ Sharon Hawkins-Fauster, regardless of whether or not your observation is correct, the iris-effect is thought to act primarily in the tropics.
Jai Mitchell says
The ECMF record is pretty clear wrt south east asian aerosol loading in the lower and mid Trop. The seasonal variation is quite large with regard to indonesian annual peat fires (mostly Sept and Nov).
he cooling due to aerosols is more than 10 W m−2 at the top of the atmosphere, and more than 25 W m−2 at the surface in the vicinity of Indonesia.
The haze reduced the seasonal average solar radiation absorbed by the equatorial Indian ocean by as much as 30 to 60 W m−2 during September to November 1997, and increased the atmospheric solar heating by as much as 50% to 100% within the first 3 kilometers.
Martin Singh says
Thanks for your comments. My understanding is that you created an iris effect by making the autoconversion rate a strong function of temperature. Presumably this has its largest effect in the lower troposphere where there is a lot of liquid water. If you altered the rate of conversion from cloud ice to snow instead/in addition could this have a larger effect on upper-level cloud, and therefore a larger effect on climate sensitivity? Not sure if this is really consistent with Lindzen’s original argument, but it would be interesting if it had a stronger effect on the resulting climate sensitivity.
1. as people pointed out, this regards the tropics
2. the Iris effect is a hypothesis by a prominent climate contrarian. As this is one of the very few of them who actually managed to produce genuine science, the theory is rare and therefore interesting to examine.
As pointed out, in the work presented the Iris effect was not established as emerging from some underlying physics, but just assumed. Maybe researchers would not be too surprised if tomorrow it turned out that this theory is plain wrong.
Thorsten Mauritsen says
@ Jai Mitchell, this sounds interesting, in particular if it has an annual cycle. However, what we presented were de-seasonalized de-trended anomalies over the area 20S-20N for the period 2001-2013. The main discrepancy between models and observations is in longwave, whereas in shortwave observations is in the middle of the CMIP5 model ensemble (Figure 2b). I don’t know how aerosols would affect the longwave regression, although it cannot be ruled out.
@ Martin Singh, I would tend to think that you are right, that a change directed more towards the ice-phase would be more effective.
Dan H. says
I was under the impressions that reduced cloud cover leading to increased temperatures (as mentioned above) was generally accepted in the scientific community. Is there a large contigent challenging this?
The bigger question, and still seemingly unanswered, is whether increased warming will lead to more or less cloud cover. Increased evaporation will lead to increased cloud cover. However, what evidence do we have that the increased precipitation efficiency will lead to less (not more) cloud cover?
Hank Roberts says
I’m not Andy, not even close, but on cloud feedbacks, they are not as simple as Dan H. suggests — not even close.
Illustrated in this picture of them being worked out:
Hank Roberts says
> “Dan H…. I was under the impression ….”
What do you read? Be careful.
I just tried ‘oogle
cloud cover temperature climate
and looking just at the first six results, they come from these sources:
4) Betts et al. 2014, DOI: 10.1002/2014JD022511 (J.G.R. Atmospheres)
If you go by your impressions from that kind of reading, you be confused.
Chris Colose says
Cloud feedbacks may be complicated, but a simple rule of thumb that emerges from that complexity is that high clouds exert a strong greenhouse effect and low clouds don’t.
In either case, it doesn’t take much water at all to make a cloud optically thick in the infrared- insofar as you have a respectable cloud to begin with, the greenhouse effect is just basically determined by the radiating height of the cloud top. The temperatures at the tops of high clouds are much colder than the surface, and thus reduce the energy loss of the planet better than low clouds (which emit at temperatures rather close to that of the surface). Having more water in those high clouds or divvying it up smaller particles shifts the balance toward a cooling influence a bit. In contrast, the albedo effect of clouds goes up a bit slower with water content than the greenhouse effect. The low clouds tend to have more water content and are more reflective, while simultaneously incapable of generating much greenhouse effect.
In the tropics, where you have deep convection and vertically extensive cloud towers, you have reflective cloud tops that also emit energy to space very high up, resulting in two very large but opposing terms in the energy balance. So a reasonable null hypothesis is that reducing the spatial area in which such clouds reside will result in a near zero feedback, with the residual probably quite model and cloud property dependent. The paper discussed in this post doesn’t really get into whether or not this is/ought to be occurring in observations, but it does argue what we expect to see if it were occurring.
Rule of thumb is high altitude clouds warm the climate, low altidude clouds cool it.
Dan Archers coursera course sheds some light on this.
Peter Todd Williams says
Sorry to ask this here, but I have a general modeling question & not sure where to ask…. seemed like a post on modeling would be a good place.
I’ve been wondering, what’s the lower thermal boundary condition for full Global Climate Models (GCMs), atmospheric or, perhaps more importantly, oceanic?
The thermal flux from the relic fossil heat from the Earth’s interior plus radioactive decay, is about 50TW total, is only 0.03% of the energy budget according to wikipedia. That may be true, but that doesn’t mean the *heat capacity* is negligible. In fact, stored heat in the crust has a big effect on daily weather patterns; witness the buildup in convective activity and large currents driven over land versus over water (e.g. coastal areas like the Bay Area). So what do you do, include a few meters of soil and fix an adiabatic b.c. below this? Or just fix an adiabatic b.c. at ground level?
Barton Paul Levenson says
I use one meter of seawater in my RCMs and set the geothermal flux to 0. A pretty crude approximation, admittedly, but I get results that match the USSA-76 pretty closely (r^2 99%, rms error 2 K for 20 layers of air and one of surface).
Dan H. says
I understand the high/low cloud differences. The tropics are much more complex, with cloud convection thought to cancel out daytime heating. Your null hypothesis soudns reasonable. Do we have any evidence that either cloud formation will form selectively over the other? Still unanswered was my main question, which was what evidence do we have that warming will lead to reduced cloud cover? Was looking for something substantial there, as Hank’s references were rather thin.
Hank Roberts says
> references were rather thin
That was the point.