ELICITATION OF EXPERT JUDGMENTS OF AEROSOL FORCING
M. GRANGER MORGAN, PETER J. ADAMS, AND DAVID W. KEITH
Abstract: A group of twenty-four leading atmospheric and climate scientists provided subjective probability distributions that represent their current judgment about the value of planetary average direct and indirect radiative forcing from anthropogenic aerosols at the top of the atmosphere. Separate estimates were obtained for the direct aerosol effect, the semi-direct aerosol effect, cloud brightness (first aerosol indirect effect), and cloud lifetime/distribution (second aerosol indirect effect). Estimates were also obtained for total planetary average forcing at the top of the atmosphere and for surface forcing. Consensus was strongest among the experts in their assessments of the direct aerosol effect and the cloud brightness indirect effect. Forcing from the semi-direct effect was thought to be small (absolute values of all but one of the experts’ best estimates were < 0.5 W/m2). There was not agreement about the sign of the best estimate of the semi-direct effect, and the uncertainty ranges some experts gave for this effect did not overlap those given by others. All best estimates of total aerosol forcing were negative, with values ranging between – 0.25 W/m2 and – 2.1 W/m2. The range of uncertainty that a number of experts associated with their estimates, especially those for total aerosol forcing and for surface forcing, was often much larger than that suggested in 2001 by the IPCC Working Group 1 summary figure (IPCC, 2001).
Comment by Roger Pielke Jr. — 8 Feb 2006 @ 4:47 PM
I don’t get it. The three forcings are equivalent at the two sigma level (if you trust the error bars). Note that the Chung et al. (2005) and Yu et al. (2005) have errors bars so large that the forcing could even be positive. One could claim that there was a problem if there were a 3-sigma difference or higher. I don’t see any issue with these estimates. The Bellouin value, if you trust its error bar, suggests that the true value should be higher 0.6 W/m^2.
The problem arises in the incident radiation at the surface. How is it possible that three contributions that are consistent give estimates with error bars that are clearly disparate? Either these papers are getting there error estimates completely wrong or the same assumptions do not apply in the three studies. I don’t have time to read the papers in detail now but I’m quite interested to know how you explain this.
[Response: The uncertainty presented by Bellouin et al., Yu et al., and the IPCC represents one standard deviation. In contrast, Chung et al.'s uncertainty value represents the range of forcing estimates derived from their different sensitivity calculations (see their Table 2).
Whether or not the central forcing estimates of Bellouin and Chung are within two standard deviations of each other depends upon whose estimate of the standard deviation you accept. The different standard deviations raise the question of whether the full range of uncertainty is really sampled by any single study. The larger point is that although each study narrows the uncertainty calculated by the IPCC, the studies when considered in aggregate span much of the original IPCC range. Nonetheless, each of these studies deserves credit for incorporating actual measurements of aerosol amount into the forcing estimate, as opposed to the IPCC value, which is largely based upon models.
As for forcing at the surface, the Bellouin and Chung clear-sky values are nearly identical; the all-sky values differ only because Bellouin et al. assume for simplicity that forcing in cloudy regions is zero.
Hongbin Yu has informed us that in the final version of their article (in press), they replace their all-sky TOA value with a clear-sky value, citing the uncertainty in estimating aerosol forcing in cloudy regions. - ron]
I am not receiving e-mail notices of new postinga. Can you help me?
Comment by William A. Atchley, M.D. — 9 Feb 2006 @ 2:08 AM
I’m wondering about “naturally occurring” forest fires (mentioned as a source of soot + CO2). Of course such fires have occurred, I guess, ever since there have been forests, long before humans entered the scene. But could some of these today be in part attributed to global warming? Such as when GW causes more drought & aridity or tree disease, and then fiercer winds?
Comment by Lynn Vincentnathan — 9 Feb 2006 @ 5:42 PM
As I have commented many times on aerosols here on RealClimate (and on the UKweatherworld discussion group), it is not easy to give a complete oversight of all the doubts I have about aerosols… In general, I have the impression that the cooling effect of human induced aerosols is largely overestimated.
To begin with, a back-of-the-envelope calculation: The largest cooling effect is supposed to be from sulphate aerosols. Here we have a good example, the Pinatubo, which ejected 20 Mt SO2 directly into the stratosphere (see for more details here. That lasted 2-3 years, until growing sulphate/water drops fell out. In contrast, humans emit some 80 Mt SO2/yr, lasting average only 4 days. The Pinatubo caused a global temperature drop (including water vapour feedback) of app. 0.6 K. If one expects that there is virtually no difference in direct effect for stratospheric and tropospheric aerosols, then the net primary effect of human SO2 emissions would be not more than 0.025 K. But stratospheric changes may have a larger impact than tropospheric (like changes in the jet stream position)…
The effect of aerosols should be measurable in the regions with the largest change, but they are not. Not in Europe, with an over 50% reduction since 1975 (neither did Philipona ea. find a positive change in insolation in their 2005 GRL paper), neither in India, where the tip is warming faster than the only station of the SH in the neighbourhood, not under the smoke of increasing emissions. Neither in ocean heat content, where all oceans in the NH are warming faster than the SH parts (if corrected for area), while the aerosol load in the NH is larger.
The influence of (sulphate) aerosols probably is overestimated, and/or the influence of other (soot) aerosols is underestimated. Which leads to questioning even the sign of the total aerosol effect…
Last but not least, the Bellouin ea. paper (I have not read – yet – the other papers in detail) need to be seen as a “worst case” scenario, and probably was intended to give a maximum (negative) influence of aerosols to be used in climate models. In fact, interpreting all fine aerosols over land as anthropogenic by them is way too high.
From the IPCC gaseous precursors and solid aerosols, the quantities involved are:
Anthropogenic: around 560 Mt/y less than 1 micron
Natural: around 350 Mt/y less than 1 micron, around 5300 Mt/yr over 1 micron.
Thus even if these are not underestimates of natural VOC emissions and/or natural fires, the annual natural emissions leading to aerosols present already 38% of the total fraction of fine aerosols. This is higher than the 28% error estimate of the authors.
Even more interesting are the recent findings that the aerosols found over land in the free troposphere are mainly of natural origin. See the 2005 GRL paper of Heald ea..
The main points:
- natural SOA (secondary organic aerosols) in the free troposphere are some factor 7 higher than anthropogenic.
- the mass ratio SOA/SOx (SO2+sulfate) aerosol is app. 2:1 to >10:1, between 0.5 and 5.5 km altitude.
- chemical transport models underestimate SOA’s with a factor 2 at the boundary layer and up to 10-100 times in the free troposphere.
The natural free troposphere SOA already counts for some 10% of the total aerosol optical depth. Add to that the amount of natural VOC aerosols formed below the boundary layer and other natural (fine and coarse) aerosols, and also the sea induced SO2 and salt aerosols over land, then we may safely conclude that the Bellouin study is a huge overestimate of anthropogenic aerosol influence.
In addition to restrictions of the upper bound influence of aerosols in climate models, the upper bound needs to be reduced further (probably more than halved, more like what is found in the Chung ea. study), based on the presence of natural small size aerosols. That has repercussions for GHG sensitivity too, as aerosol cooling and GHG warming are tightly coupled (see RC here), which results in appreciable differences in projections of future climate.
Thanks Ron, I finally got time to read the papers, and there are indeed differences in the way uncertainties are computed, but it’s obvious that all estimates can be put close to -0.6 W/m^2. So at the TOA the three approaches seem quite robust and the differences between the authors do not hint at a serious problem. Well, at least not until we get studies with smaller standard deviations.
And as you point out, clear-sky values are nearly identical, this clears my doubts. Incorporating the actual measurements seems to actually be working quite well (well there is still the problem of cloudy skies). Thank you for taking the time to present these interesting results. I don’t work in this subject and although interested miss some of the relevant literature (and do not have time to read each paper).
The effect of tropospheric aerosols on global surface temperature cannot be deduced so easily from the temperature effect of volcanic eruptions. The comparison just of the amount of SO2 and summing up over the lifetime is too simplistic. A good review about the volcano climate effect can be found at http://www.space.edu/papers/Volcanoclimate.pdf.
Two major points:
- the optical depth depends on the size of the aerosols (smaller aerosols have larger optical depth per unit mass). Larger eruptions tend to produce larger aerosols. Larger aerosols sediment out faster. Thus the temperature effect of a volcanic eruption is not proportional to the amount of SO2 emitted into the stratosphere.
- Volcanic eruptions lead to a heating of the stratosphere and alter stratospheric chemistry (ozone depletion). The alteration of stratospheric temperature distribution can alter atmospheric circulation patterns (El Nino, NAO), which also influence global temperature.
Further it is neglected, that in the stratosphere the aerosols are distributed over a much bigger area (for geometrical reasons, roughly about a factor of 3-5) than in the troposphere so you need more aerosols to get the same shielding effect (it’s not the same thing to put a parasol 2 m above ground or the same parasol 50 m above ground, you won’t get the same cooling effect).
Why should the same climate models that include the effect of sulphate aerosols, reproduce the cooling effect of volcanic eruptions very well, overestimate the effect of tropospheric aerosols by a factor of about 10?
The claim that the effect of aerosol reductions should be seen one-to-one in the measurements, seems strange. The temperature effect of aerosols (and ozone in the case of the Europe link) alone is compared to the measured temperature, which is the result of all effects together (natural, greenhouse gases, etc.). Apples are compared to fruit salad. The same for India and the oceans. For the radiation measurements which show increasing shortwave radiation since the 1980s, see http://www.realclimate.org/index.php?p=154.
For the calculation of anthropogenic climate forcing the exact value of natural emissions (or the discovery of new natural sources) is not that much relevant, because these emissions have been there before and do not change climate unless it is found that they have changed, too on the discussed timescale.
With reference to Jones and Cox’s (UKMO) attempt to broadly gauge the effect of sulphate aerosols. “Impact of uncertainties in sulphate forcing, climate sensitivity and carbon cycle feedbacks on climate projections for the 21st century.” See http://camels.metoffice.com/MiscReport01.html
I assume that these results allow us to constrain the 2xCO2 temperature increase to the range +2 to +4 deg C (Or in fact less? As sulphate forcing is only a part of the overall aerosol effect). So this seems to be further reason to doubt the possibility of the apocalyptic scenario presented by Peter Cox at the end of BBC’s Horizon Global Dimming (Although 2 to 4 deg C for 2xCO2 will probably still be very bad).
1. There is no difference in how sulphate aerosols are formed from volcanoes in the stratosphere or from industrial and natural SO2/H2S/DMS in the troposphere. It only takes more time to form in (and drop out of) the stratosphere, in part due to less water vapour. Other volcanic aerosols may be helping, but in general are heavier and drop out in a few days to a few months. The Pinatubo delivered long lasting aerosols of mainly 0.5 micron and another fraction below 0.18 micron, which may be defined as (very) fine. See: NOAA
2. Agreed, as the effects on stratospheric circulation enhances the cooling effect of volcanic aerosols.
3. Disagree, the quantity of reflected sunlight by the same amount of widespread aerosols is at least as high as for less spread aerosols. Dense packed aerosols might even be less effective, as more drops are in the same light path. But there will be differences if the spread is changing in latitude.
The effect of a dark umbrella before a large window, is identical to cutting the umbrella and glueing the pieces over the whole window…
(Btw, the diameter of the earth is 12.756 km, adding 2 times 12 km to include the lower stratosphere does change the total surface a little bit, but that is not very relevant here).
4. Good question, should be answered by the guest commenters here…
I suppose that the models overestimate the tropospheric (sulphate) aerosol impact, but are more or less on target for volcanic aerosols (see point 2)…
5. Of course, the calculated effect of aerosols is diluted by the overall effect of all changes, but even then, there must be a difference measurable between areas where there is no (or less) change in aerosols vs. areas where there are large changes, as all other influences have (more or less) the same effects on both areas. But there is no difference observed in Europe, and in South India, it goes the wrong way out…
If you double or halve the number of apples in your fruit mix, that will give some change in taste…
6. I have the impression that global dimming has more to do with water vapour than with aerosols. Philipona did measure a decrease of incoming light with reducing aerosols in Eastern Europe. Unfortunately, there were no data if the decrease was from increasing water vapour alone, or from a mix of increasing insolation from decreasing aerosols, overwhelmed by a larger decrease due to increasing water vapour. I have asked Philipona if this could be checked (if certain wavelengths were used where water vapour is an active absorber), but received no answer.
7. First of all, I made a mistake by interpretation, that Bellouin ea. estimated the total fine fraction above land as only anthropogenic. That was based on Fig. 1 and Table 1 of his work at NOAA and the text was not clear on that point. The Letter to Nature made it clear that the researchers made an attempt to distinguish between natural and anthropogenic fine aerosols over land by using model estimates of aerosol composition. Although this reduces the impact of my comment, if the models underestimate the amounts of natural fine fraction aerosols (which is indicated in the Heald ea. paper: a factor 2-100 in the free troposphere), then the amount (and thus the impact) of anthropogenic aerosols is overestimated (satellites measure the sum of aerosol quantities)…
You are right about the area effect, that was a mistake. It is indeed small.
First (point 1) you argue, that formation and effects of sulphate aerosols (anthropogenic in troposphere and volcanic in stratosphere) are the same, and later you suppose (point 4), that models get it right in the stratosphere but not in the troposphere. If the processes (and feedbacks) are really the same, it is very unlikely that models get it right in the stratosphere, but wrong in the troposphere by a about a factor of 10… (this is a contradiction you have to explain, not the modellers, because it’s you who suppose they’re wrong).
point 5: The other effects besides aerosols are far from being homogeneous (maybe some forcings, but not the effects). Just have a look at the GCM map including all factors. On the regional scale, e.g. internal decadal variability is also important (NAO, etc). GCMs are not very significant on the local scale, if you want to compare local points it would be better to rely on regional models. And the temperature data you investigate should be homogenized.
Urs, the primary effect (forcing by changes in reflection) is for both aerosols the same, but the feedbacks are quite different. Any change in temperature gradient (equator to poles) in the stratosphere has more impact than a similar change in the troposphere and part of the extra reflection in the troposphere is absorbed again at higher levels. Just to name only a few possible differences.
But that is not the basic question. If we may suppose that sulphate aerosols have the same effect, independent of height and despite the feedbacks, then the effect of anthropogenic sulphate aerosols is much lower than currently calculated in the models (that is a matter of straight-forward physics for amounts and accumulation). Thus I like to see an answer of the modellers, what the origin of the difference in sensitivity (of the primary effect) is.
For Europe, I have plotted the NAO index too, which has an effect both on less poluted areas and more polluted areas. In this case, there is a (NAO induced) jump in 1976 with a slightly different outcome for near-ocean stations. But the trends further are near identical (as average of three rural stations). Even without looking at any model, there should be a difference visible downwind of the largest sources (the Philipona paper shows impressive diverging regional temperature trends between West and East Europe, due to (NAO induced)regional changes in water vapour)…