Climate Feedbacks

Figure 1 from Soden and Held (2006) showing ranges for each model for each of the key atmospheric feedbacks for the IPCC AR4 models and a comparison with an earlier survey (Colman, 2003).

All models predict the concentrations of water vapor to increase as the climate warms due to the rapid increase in saturation vapor pressure with temperature. Because water vapor is the dominant greenhouse gas, this provides a strong positive feedback in the climate system. In current models, water vapor was found to provide the largest positive feedback in all models and its strength was shown to be consistent with that expected from a roughly constant relative humidity change in water vapor mixing ratio. It should be noted that models are not constrained to conserve relative humidity and significant regional changes in relative humidity are simulated by models in response to atmospheric warming. On the global scale, however, changes in relative humidity are small.

Models do exhibit a range of values for water vapor feedback. This range is not due to departures from constant relative humidity behavior, but rather from intermodel differences in the response of the atmospheric lapse rate to surface warming. All models suggest that the troposphere warms more than the surface (at equilibrium at least – responses are more varied for a short transient period – see the CCSP report). This amplified warming of the troposphere represents a key negative feedback in models because it further increases the thermal emission of energy to space. Models with more surface warming in low latitudes tend to have larger atmospheric warming (and more negative lapse rate feedback) because the surface and free troposphere are more strongly coupled in the tropics than at higher latitudes. Because the water vapor and temperature responses are tightly coupled in the troposphere, models with a larger (negative) lapse-rate feedback also have a larger (positive) water vapor feedback. These act to offset each other. As a result, it is more reasonable to consider the sum of water vapor and lapse-rate feedbacks as a single quantity when analyzing the causes of intermodel variability in climate sensitivity. As shown in the figure, the range for the sum of these two feedbacks is considerably smaller than the range of either the water vapor or lapse rate feedbacks individually.

Not surprisingly, the surface albedo feedback due to changes in snow and ice cover was also found to be positive in all models, although its magnitude is only about 25% of that from the combined “water vapor plus lapse rate” feedback.

Consistent with previous studies, clouds were found to provide the largest source of uncertainty in current models. For the most sensitive models, cloud feedback is positive and comparable in strength to the combined “water vapor plus lapse rate” feedback. For the least sensitive models, cloud feedback is close to neutral. Many specialists and non-specialists alike are sometimes surprised to see that the model-predicted values for cloud feedback ranges from neutral to strongly positive; often believing that cloud feedback is more uniformly distributed between negative and positive values. This confusion may stem, in part, from misinterpretations of the change in the easily-diagnosed “cloud radiative forcing” in model simulations of climate change. This diagnostic, based on the comparison of clear sky and cloudy sky radiation differences (Cess et al, 1996) is related to the more-difficult-to-calculate cloud feedback, but can be negatively biased by correlated changes in water vapor and temperature (Soden et al., 2004). Thus studies that use the “cloud radiative forcing” calculation have reported a more negatively skewed ‘cloud feedback’ then seen here. However, based on these estimates and on a survey of published values of feedback calculations (Colman 2003), there do not appear to be any models for which clouds provide a substantial negative feedback on the climate.

Observational studies do have the potential to help narrow the uncertainties in these individual feedbacks – for instance from studies of the response to Mt. Pinatubo and from long time series of satellite measurements – but observational constraints of cloud feedback remain elusive.

Page 2 of 2 | Previous page

118 comments on this post.
  1. Leonard Evens:

    I doubt very much if it affects the thrust of the argument, but I think diagram might be a trifle misleading. Eyeballing it, one tends to look at the average value for each contribution and then mentally to try to combine those average values. Presumably the individual feedbacks are not independent variables. That would mean the contribution from each factor and hence the total would be dependent on the model. So ideally, might it not be clearer if the individual dots were labeled by model?

    [Response: The individual numbers are all in Table 1 in the paper. – gavin ]

  2. Roger A. Pielke Sr.:

    Brian- In your discussion of climate feedbacks, you ignored an important 2005 National Research Council report.

    National Research Council, 2005: Radiative forcing of climate change: Expanding the concept and addressing uncertainties. Committee on Radiative Forcing Effects on Climate Change, Climate Research Committee, Board on Atmospheric Sciences and Climate, Division on Earth and Life Studies, The National Academies Press, Washington, D.C., 208 pp.

    Among the conclusions in the Executive Summary of the Report is,

    “Despite all these advantages, the traditional global mean TOA radiative forcing concept has some important limitations, which have come increasingly to light over the past decade. The concept is inadequate for some forcing agents, such as absorbing aerosols and land-use changes, that may have regional climate impacts much greater than would be predicted from TOA radiative forcing. Also, it diagnoses only one measure of climate change ‘global mean surface temperature response’ while offering little information on regional climate change or precipitation. These limitations can be addressed by expanding the radiative forcing concept and through the introduction of additional forcing metrics. In particular, the concept needs to be extended to account for (1) the vertical structure of radiative forcing, (2) regional variability in radiative forcing, and (3) nonradiative forcing. A new metric to account for the vertical structure of radiative forcing is recommended below.”

    Your comments on this to Real Climate readers would be informative. Your statement, for example, that “Consistent with previous studies, clouds were found to provide the largest source of uncertainty in current models.” is not supported by the findings in the 2005 NRC report. The climate forcings are more diverse than summarized in your weblog, and the climate feedbacks, therefore, are more complex and involve more than atmospheric processes.

    Interested Real Climate readers who want a broader perspective on this subject are invited to read the weblogs at

    [Response: Roger, I think you are confusing the role of cloud effects as forcings (principally from indirect aerosol effects) from their role as feedbacks to changes in the radiative budget caused by other factors. This post (as I read it) is concerned with feeedbacks – not the multitude of different forcings. – gavin]

  3. Alastair McDonald:

    I’ve given a fair amount of thought to the problem you describe, and my conclusion is that the models are under estimating the water vapour feedback which is strongly positive, but are correct when finding the cloud feed back is close to neutral in the short term. This will mean that when the Arctic sea ice melts, the decrease in albedo will cause a rapid warming due to the positive feedback from the greenhouse effect of water vapour. It is only when the temperature has risen to such an extent that there is a switch into a new cloud regime that the climate will again be in equilibrium.

    Basically this is what happened at the end of the Younger Dryas, when the sea ice that had spread out of the Arctic Ocean as far south as Ireland suddenly disappeared. Temperatures in Greenland rose rapidly by 20C, and only stabilised when the ITCZ expanded making the Amazon wetter [Maslin & Burns, Science, 2000, ]

  4. Steffen Christensen:

    Thanks for the great post, Brian!
    I’m sure that I am misunderstanding some of the basic physics here, maybe you can clarify. You state that doubling atmospheric CO2 (presumably from 280 ppm to 560 ppm or thereabouts) adds around 4 W/m^2 to Earth’s power budget, with a naive effect of, in the steady-state, increasing the mean surface temperature by ~1 degree Celsius. You then compute the key atmospheric feedbacks as having an aggregate effect of 0.85 to 1.7 W/m^2/K or so – with a mean value of maybe 1.25 W/m^2/K or so. When I go and plug this mean value in with the 1 Kelvin rise in temperature from CO2 doubling, I get an additional effect of 1.25 W/m^2 from feedback, which with a linear temperature response, heats up Earth by an extra 0.3 K or so. We have to add the feedback from this as well and so on, so in the limit I get a temperature increase of 1.45 degrees C or so. But you state that the expected mean temperature impact of doubling CO2 is between 2.6 and 4.1 degrees Celsius increase, so I’ve done something wrong somewhere. What was it?

  5. Jeffrey Davis:

    I’ll 2nd the remark in #1. The preliminary discussion above posits a straight 1C rise in temps per doubling of CO2 without a consideration of feedbacks. The graph, then, has a value marked ALL which looks like it posits a 1C (or less) rise in temps due to all the forcings under examination. I presume that means all the interactions of positive forcings, negative forcings, and associated effects. So, the combined rise in temps looks to be 1C for CO2 alone and around 1C (or less) when the feedbacks are included in the equation. A total of 2C (or less). How does one get to the 2.6-4.1C rise predicted in the first sentence of the article?

  6. John Dodds:

    You must have been anticipating my question: What is the magnitude of the negative feedback from convection and conduction (& the associated negative Water Vapor feedback) that results when the added GHGs (& positive WV feedback) result in warming the atmosphere? Is the convection effect ALL included in the “lapse rate feedback”? But what is the magnitude of conduction (electron movement) feedback?

    It seems to me that when the anthropogenic GHGs result in atmospheric warming by keeping the absorbed & quickly released radiation energy in the air longer than it would have been without the GHGs, then the subsequent atmospheric warming results in a larger temperature differential between the ground and space. This differential will drive larger (faster) convection currents (aka stronger hurricanes) and larger vertical conduction currents (aka more vertical movement of electrons and more lightning to cause more wildfires). The question is what is the magnitude of this increased transportation of NON RADIATIVE energy or negative feedback (or decreased energy residence time in the air) and are these effects accounted for in the computer simulation programs? Is there a reference for the discussion of conduction effects anywhere?

    I visualize this ground to space energy transfer process as being a set of 5 (or more) “resistance” pathways in parallel connecting the ground & space- ie 1.Direct radiation energy to space, 2.Radiation energy thru Water Vapor absorption which is temperature dependant, 3.Radiation thru GHG (mostly CO2) absorption other than WV, 4.Convection or physical transport of hot air molecules/wind energy and 5.Conduction or transport of energy by electron movement. If you increase the “resistance” in one leg (eg add CO2 and thus raise the air temp due to the increased residence time of the GHG radiation energy in the added 200ppm/century of CO2) then the energy current will compensate somewhat & go thru the other paths (eg faster movement of 1,000,000 ppm of air) to get from ground to space in order to maintain the equilibrium energy-in equals energy-out balance. But what are the relative magnitudes? and the net result?

  7. Lynn Vincentnathan:

    So greater/lesser albedo is a feedback, but what about carbon releases from melting permafrost & ocean hydrates due to the warming? Would this be a feedback, or a forcing? Or maybe a feedback that becomes a forcing?

    Are there any models that include these factors? I imagine the uncertainties & inability to quantify with any precision would be an obstacle to that — which doesn’t mean it isn’t happening or won’t happen.

  8. Eric (skeptic):

    This is an excellent topic, the added accuracy and resolution of weather modeling along with modeling other factors is what is needed to determine the warming, determine the effects (good and bad), and determine the best solution.

    My primary comment about feedbacks is that they need to be considered in their time scale. Weather responds in hours, vegetation in months, snow cover in years (with seasonal variations) and ocean changes in decades. Those are only some and only rough.

    Regarding weather feedback, there are lots of subtleties that can make a large difference in climate results. Some examples are cloud at night (positive) versus the day (negative), higher latitude cloud (positive) versus lower, high topped clouds (positive) versus low topped, diffuse or weak convection (positive) versus concentrated and strong, and a steady jet stream (positive) versus amplified.

    Other feedbacks to be considered and modeled are rotting vegetation (positive), burning vegetation (negative), ocean stratification (positive) versus circulation (negative), soil moisture (negative), etc. many of which are tied to local weather conditions.

    Most of my conclusions are gleaned from here:

  9. Eric (skeptic):

    Lynn, I would consider carbon and methane from melting permafrost to be a feedback. A forcing would be a variable that does not depend on warmth, CO2, weather or any other similar factor.

  10. Roger A. Pielke Sr.:

    Gavin- Regarding your reply to my posting, The 2005 NRC Report discussed and illustrated feedbacks;

    “A climate forcing is an energy imbalance imposed on the climate system either externally or by human activities. Examples include changes in solar energy output, volcanic emissions, deliberate land modification, or anthropogenic emissions of greenhouse gases, aerosols, and their precursors. A climate feedback is an internal climate process that amplifies or dampens the climate response to an initial forcing. An example is the increase in atmospheric water vapor that is triggered by an initial warming due to rising carbon dioxide (CO2) concentrations, which then acts to amplify the warming through the greenhouse properties of water vapor. Climate change feedbacks are the subject of a recent report of the National Research Council (NRC, 2003).”

    The caption to Figure 1-2 in the 2005 NRC Report reads,

    “FIGURE 1-2 Conceptual framework of climate forcing, response, and feedbacks under present-day climate conditions. Examples of human activities, forcing agents, climate system components, and variables that can be involved in climate response are provided in the lists in each box.”

    As Brian wrote,

    “The exact boundary between a feedback and a forcing depends on what is considered to be part of the ‘system’ and can sometimes be a little fuzzy. This discussion addresses just the feedbacks associated with the atmospheric physical system (see this earlier article for why that is), but other, less well understood, feedbacks (changes in land vegetation, biogeochemical processes, and atmospheric chemical feedbacks – see the NRC 2003 report), while potentially important, are not part of the generally understood definition of ‘climate sensitivity’. ”

    A major finding of the 2005 NRC report is that these other feedbacks are very much a part of “climate sensitivity”. The feedbacks also cannot be adequately discussed without also discussing the climate forcings as illustrated in Figure 1-2 in the NRC Report;

    The climate system is clearly more than the “the atmospheric physical system.”. Note that the heading of Brian’s weblog is “Climate feedbacks”, not “Feedbacks with the atmospheric component of the climate system.”

    [Response: I don’t really disagree. The climate is sensitive in a much more complex way to emissions etc. than is encapsulated in the traditional “climate sensitivity” metric. I think Brian was clear that he is talking about the the feedbacks that determine the variation in the classically defined “climate sensitivity” – and within that limited definition (which was clearly outlined) these results are very interesting. We had a big long discussion on the wider issue of ‘how sensitive climate is’ (in the larger sense) in the comments on carbon cycle feedbacks post recently. It seems to be that as long as one is clear about what is being talked about, you should be fine. But talking about classical ‘climate sensitivity’ in no way implies that there is nothing else worth talking about. -gavin]

  11. wayne davidson:

    It seems complex, but may be it is not? Given that there is OLR at all height levels , and also CO2 equally distributed, I presume that there would be warming at all levels of the troposphere, causing a potential increase for water vapor feedback everywhere till the tropopause (water vapor is rare in the stratosphere due to tropopause inversion). Convective transport, lapse rate negative feedback, is a bit confusing since this convection distributes water vapor to higher altitudes, some of which precipitates, but I see the increase presence of water vapor as the main source of temperature increase everywhere in troposphere, I look at convection slightly differently than the results suggest, the over all mechanics over time would be slight, but significant feedback increase due to water vapor interspersed more uniformly during longer periods of time. This is seen in the Polar Regions, more rain during summer, and in the tropics, less rain, as saturation levels are not reached due higher temperatures

  12. Brian Soden:

    Re: #1 Thanks for pointing out the discpreancy between the “All” column and the sum of the other feedback terms. The values for “All” are acutally mis-labeled in the Figure (and erratum to be written!). They are, in fact, the “effective sensitivity” from Table 1 of Soden and Held (2006). The effective sensitivity can be computed by summing the individual feedback terms and including the Planck radiative damping (which is roughly -3.2 W/m2/K; see Table 1 of Soden and Held) and then flipping the sign. The results of this are listed in Table 1. Or, in ascii math, dT = dQ/eff_sens, where dT is the temperature response and dQ is the radiative forcing. My apologies for the confusion and thanks for pointing it out.

    [Response: I edited the figure to make this clearer. One other point should be clearer is that the ‘traditional’ climate sensitivity (deg K/(W/m2)) is simply the inverse of this number, and then for the sensitivity to 2xCO2 it is ~ 4/eff_sens. – gavin]

  13. Brian Soden:

    Re #2. Roger – The subject of radiative forcings is an equally interesting one and I certainly do not mean to suggest that uncertainties in radiative forcings are not an important part of the problem. Regarding the connection between feedbacks and forcings, I believe this reflects, in part, the limitations of defining forcings in terms of tropopause level fluxes. I personally like the “adjusted troposphere and stratosphere” forcing approach outlined by Shine et al.,2003 ( which they argue provides a more reliable metric for relating climate forcing with the associated climate feedbacks and temperature response. As far as our calculations, we used the climate forcing estimates and uncertainties cited in TAR. It should be noted, however, that to my knowledge there has never been a formal intercomparison between GCMs of the radiative forcings that drive their scenario responses. As discussed in our paper these also represent a potential source of error in our estimates of the cloud feedback.

    [Response: Actually, we just contributed to such a comparison for AR4. The preliminary range of actual (adjusted) forcings for 2xCO2 is between 3.5 and 4.2 W/m2 – comparable to the +/-10% error bars you used. -gavin]

  14. Brian Soden:

    Re: #4 – See my above comment to #1. I believe this should clear up the confusion.

  15. Jim Redden:

    Unless the answer to Lynn Vincentnathan’s question has already been answered, I too have wondered if the strong potential for increased Artic carbon release and methane, as a result of melting permafrost, has been factored into any model.

    My question: How does the GISS GCM Model E take relate the biogeochemical release of Artic CO2 and Methane?

    It’s my understanding that buried methyl hydrates also exist in areas of the permafrost, in addition to deep ocean; if permafrost continues to melt, the methane and co2 from decay are very likely going to release in the atmosphere.

    With respect to evidence of increased climate sensitivity, I recall a 2006 paper from the journal of Paleoceanography: A multiple proxy and model study of Cretaceous upper ocean temperatures and atmospheric CO2 concentrations (Bice, K.L., Birgel, B., Myers, M.A., Dahl, K.A, Hinrichs, K., & Norris, R.D.).

    Indicated by plankton ratios, were very high tropical Atlantic ocean temps–as the press reported, the water of 100 million years ago was warmer than a hot tub.

    On a general pattern for systems, as different phases are reached, relationships change. I don’t see how the Earth’s climate system, with the driving currency being the phase state change of water, being any different. At some increased energy state, a new equilibrium point is likely to dominate.

    I have heard several scientists exclaim it’s not what we know that we are concerned about, it’s the surprises worry us the most. After reading the draft of the next IPCC, I was left to wonder if in the name of agreement, the â��policymakerâ�� warnings of future warming (that we have already bought by current day emissions) in the pipeline were not forcefully enough stated. As most of us would agree, there seems to be a lot at stake.

  16. Isaac Held:

    I would also like to thank readers #1,#4, and #5 for pointing out this error in the figure. It is interesting how these things get through the review process!

    Another way of stating the results from this paper is that the feedbacks that we are moderately confident about (water vapor, lapse rate, and snow/sea ice albedo) seem to generate a sensitivity near the low end of the canonical range, with the more uncertain cloud feedbacks then providing the positive push, in these models, to generate all of the higher sensitivities. I think the picture that many of us had, speaking for myself at least, was that the first set of feedbacks brought us with moderate confidence to the middle of the canonical range, with cloud feedbacks, both positive and negative, then providing the spread about this midpoint. One evidently has to argue for a signficantly positive cloud feedback to get to the 3K sensitivity that various empirical studies seem to be pointing towards.

    We needed to make a lot of approximations in this analysis, especially for the cloud feedback term, because of the limitations of what we could do with the model results that have been archived, so it will be interesting to see if this picture holds up. If, in fact, this is an accurate diagnosis of what the models are doing, why is it that they all have positive cloud feedbacks? This is in itself a bit surprising given the diverse schemes used to predict clouds in these models.

  17. Hank Roberts:

    More good discussion here:

    I’m glad to see the nomenclature problem acknowledged — by seeing “classic ‘climate sensitivity'” used above by Gavin in the inline response to #10.

    But please don’t change the meaning of the original term, even by adding “classic” — beware the New Coke confusion.

    Nor do I recommend varying it.
    Imagine: Classic plus unexpected natural events — “baroque climate sensitivity”?
    — and anthropogenic events — “modern climate sensitivity”
    — and unexpected events caused by anthropogenic warming — “hiphop climate sensitivity”?

    I suggest leaving alone the original defined term (at equilibrium, at 2xCO2, defines climate sensitivity).

    Too confusing otherwise.

    Does climatology have no simple term for what people are really asking about — how fast things will change?
    Climate fungibility? climate fragility? Climate vector sum divergence?

    “Dangerous Climate Change” is accurate, eh? Arrrrgh.

  18. Steve Sadlov:

    Regarding moisture, especially in the tropics. I just got back from spending a few weeks in the tropics in SE Asia. Here are some qualitative observations. This is related to being at the edge of some typhoon outflow. With the relatively high sun angle still in place, prior to the approach of the disturbance it was very hot and muggy, with much heat trapped in the boundary layer especially in urban areas. With the onset of typhoon / monsoonal mode, firstly there was enough wind to stir things up and secondly the clouds and rain resulted in significant cooling overall. Without the muggy stillness, less heat was trapped in the boundary layer at night.

    Changing gears to the mid latitudes. Witness the synoptic progression in California since mid Spring. First we had the persistent Siberia Express / deep digging trough until June. It slackened a bit in June but nonetheless we had a cold and moist June. We had a “normal” Summer pattern for about a week (e.g. Pacific High, onshore breeze, warm inland). Then the “heat wave” set in, driven by three weeks of a rather non wavy jet, coupled with a tripple barrel high and subsequent offshore flow. Interestingly, such a configuration is far more typical of Fall than Summer here. Coupled with the July sun angle, it meant oppressive heat. When this broke down last week, it broke down in the extreme. Got the major onshore push and a 3K ft deep marine layer, well intruding. Now, there is a trough digging, and a cut off low progged to be in place by the weekend. It will be interesting to see what we get the rest of this month. The cut off may actually yield some pops. A cut off doing that in September would be unremarkable but in August, it would be unusual – August rain here is usually from the monsoon not a cut off.

    Key metric – opening day for Squaw Valley.

  19. pete best:

    Dear RC

    Radiation is a forcing as is CO2 and Methane. However water vapour is a feedback? Does that means that carbon and methane release from organic matter decay is a forcing as ultimately it is just gas and hence a forcing?

    Earths albedo is a feedback because it is caused by a forcing?

  20. Eric (skeptic):

    Hi Pete, forcings are external to a system and feedbacks are part of the system. Climate forcings are those factors that affect climate but aren’t affected by it. So methane from plant decay is mostly a feedback since it changes when climate changes. Methane released from volcanoes and man-made methane are forcings.

  21. Hal Aljibury:

    re: 19. This is where I understand the border between a ‘forcing’ and a ‘feedback’ becomes blurred. The original input to the system of doubling the CO2 level I think we can call a ‘forcing’. If one of the effects of doubling the CO2 level is to generate some more CO2 from forest fires, or releasing methane from decaying organics, I think we can term the secondary inputs as ‘feedbacks’. Sound good to the rest of you?

  22. Ezequiel Marti­n Camara:

    In response to 7, Lynn Vincentnathan, I am no climatologist but by what I have read, if the model knows how to calculate it then it is a feedback, if you have to tell the model then it is a forcing.

    In your example, if the model has formulas on how much CO2 is released from a certain square of permafrost with a given raise of temperature, then this raise is a feedback. If there is a table where you tell the model “add X tons of CO2 in this square for this year”, then it is a forcing. That is why Pinatubo and LA will always be forcings, unless you can model their behaviour and then you should drop climatology and get rich in volcanic eruption prediction or politics.

  23. Brian Gordon:

    I could easily see permafrost methane starting out as a feedback but becoming a forcing. Eric(s) brings up an interesting point: what about factors that are external to the climate AND are affected by it, like the frozen methane?

  24. wayne davidson:

    In this presentation, what is considered to be “top of the atmosphere”? I suspect it is the tropopause. Also Relative humidity considered almost constant is a bit misleading, since air with 50% RH at +30 C has a whole lot more water vapor molecules than with the same parcel of air with 50% RH at -30 C. I would like to know if there is any idea, perhaps with numbers beyond Giga Tons of water vapor increased on a yearly basis? Can this increase be measured using satellite technology?

    [Response:No. Top of the atmopshere is the top of the atmosphere (zero mb). The tropopause is usually considered to be a key level for the forcing and is around 100 to 200 mb depending on latitude. -gavin]

  25. Bill Sneed:

    The principal by-product of hydrogen combustion is water vapor. If the hydrogen economy becomes a reality, does this mean that water vapor will become a forcing mechanism rather than a feedback mediator? If water vapor is the most efficient greenhouse gas, what does such an economy portend?

    [Response: It’s very difficult to alter the water vapour distribution directly since the residence time is so short and the natural fluxes so strong. There is some thought that leakages from hydrogen tanks might effect atmospheric chemistry but the impacts are expected to be small. – gavin]

  26. Thomas Lee Elifritz:

    If water vapor is the most efficient greenhouse gas, what does such an economy portend?

    Cleaner air as long as the energy comes from solar driven electrocatalysis.

    The water vapor itself is trivial. Do I have to explain it to you, or can you click your way into science literacy?

  27. Roger A. Pielke Sr.:

    Gavin- Regarding your response to my Comment(#10), we agree on the need for a broader definition of climate sensitivity. Is Real Climate going to post such a weblog based, at least in part, on the 2005 NRC report? This needs to be more than just the carbon cycle or the global average surface temperature trend, but based on climate metrics that have the most direct impact on society and the environment. As discussed on the Climate Science weblog,

    “The needed focus for the study of climate change and variability is on the regional and local scales. Global and zonally-averaged climate metrics would only be important to the extent that they provide useful information on these space scales.”

  28. Steffen Christensen:

    Thanks so much for the clarifications, Brian and Isaac, and thanks for the forcings numbers, Gavin! My new revised calculation has the range of values for total sensitivity at 3.2 – .85 to 3.2 – 1.7 W/m^2/K, or 1.5 to 2.35 W/m^2/K, in the same positive sense as the other contribution components. Using 4 W/m^2 as the baseline for doubling CO2 gives me a temperature increase at equilibrium of 1.6 to 2.4 degrees C. Using Gavin’s range of estimates gives a full range of 1.5 to 3.0 degrees C, finally according with observational constraints. It still seems a little low, which nicely explains Isaac’s statement that cloud sensitivity must be at the high end of the range to get 3 K. Any chance that your model is missing a significant source of sensitivity? You wouldn’t need much to move the numbers up to the 2.6 to 4.1 K range from other work. Alternately, your estimates could be all fine, and the forcing from doubling CO2 could actually be on the low end of Gavin’s estimates. Not being a climate scientist, I can’t say which is more likely… any thoughts?

  29. Mark A. York:

    I concur with Steve Sadlow. While I’m greatful for the relief, I’ve never seen this in my decade in S. Calif.

  30. Robin Johnson:

    This may be a dumb question but I’ll ask anyway [I have no shame so says my mother] and more importantly I couldn’t seem to find an answer using web searches: What is the local (e.g. regional) effect of burning wood from forest fires, brush clearing for farming and ranching, cooking and heating on the local climate? The amount of wood/brush consumed in natural and man-made fires is quite significant it would seem. Does the moisture laden smoke close to the ground raise surface temps to any significance or have any effect other than on the carbon cycle? Or if it does, does it cancel itself out when the moisture condenses out and the aerosols remain producing temporary cooling? Or would it do something weird like warm things slightly in winter and cool things slightly in the summer?

    I’m just curious. A link to a reliable explanation would suffice if someone knows of one.

  31. Martin Lewitt:

    Steffan #27, you seem to assume that “observational constraints” are an additional estimate of climate sensitivity independent of the models. However, if you are thinking of Annan’s work for instance, models were central to his analysis of the Last Glacial Maximum and of the Maunder Minumum. Furthermore Annan’s work and the earlier work of Gregory are both based on the assumption that all radiative forcings are equivilent, and therefore past sensitivity of the climate to solar forcings can be used to estimate sensitivity to CO2 forcing. This assumption is open to question, given the unexplained strength of paleo climate correlations to apparently weak changes in solar forcing.

  32. Eric (skeptic):

    Hi Robin, this link has the results of some modeling of fires (e.g. 9927482) that suggests the aerosols produce temporary cooling. Another consideration is that these wood products won’t then rot and produce methane. I think there are probably studies of the Indonesian fires like this one that have more complete analyses. Also check the studies of forcings from specific volcanoes, the effects might be similar.

  33. Jeff Weffer:

    The solar irradiance is 1367 W/m^2 with the 11-year solar cycle varying by +/- 0.6 W/m^2 and longer-term changes varying as much +/- 3.0 W/m^2 over the past 1,000 years.

    If these changes are not reflected in the climate record, why would we expect 4 W/m^2 to make such a diference?

    [Response: To compare like with like you need to divide the solar irradiance changes by 4 and multiply by 0.7 to account for the geometry and albedo effects. So even with your (rather high) estimate of what the long term solar changes are only about 0.5 W/m2 – significantly smaller than the ~1.6 W/m2 net estimate of all anthropogenic forcings since the pre-industrial. That’s why it has been such a challenge to tease out a consistent solar response in the paleo-record. – gavin]

  34. Chris Rijk:

    Re: #30
    I was actually just trying to lookup how much difference the Milankovitch cycles cause in W/m²… Particularly, how much of an increase in the amount of energy absorbed was enough to move the earth’s climate into the current interglacial.

    Anyway: The 1367 W/m² you quote is for how much radiation hits the earth, not how much is absorbed. Eg, fresh snow reflects 90% of the sunlight that hits it. But even factoring in the Earth’s albedo, I’m not quite sure how an extra 4W absorbed leads to a 1K change directly (and more with feedback effects).

  35. Joel Shore:

    Re #33 (the question about why the forcings from the sun’s variability aren’t comparable to that from doubling CO2):

    In addition to needing to correct for the earth’s albedo, as noted in #34 (which entails multiplying the solar constant by something like 0.7), the solar constant has to be divided by 4. The reason for this is that although the output of the sun is 1367 W/m^2 at the radius of the earth’s orbit, the earth’s cross-section to this radiation is pi*r^2 while its total surface area is 4*pi*r^2. [From a physical point-of-view, you can also look at it like this: Yes, on the sunny side of the earth, there is 1367 W/m^2, but on the opposite side there is 0 W/m^2…And, even the 1367 W/m^2 on the sunny side is only true at the point where the sun is directly overhead. At other points, it has to be corrected by an obliquity factor. But noting the ratio of 4 between the earth’s surface area and its cross-section to solar radiation is a way to determine the correction factor quickly without having to do this calculation in this more complicated way.]

    So, at the end of the day, your estimates of the top-of-the-atmosphere solar forcing variation are almost a factor of 6 too high than the correct number to use.

  36. Joel Shore:

    …Well, my last comment essentially duplicates Gavin’s response…which wasn’t yet there 10 minutes ago when I started mine!

    At any rate, my comment explains in more detail the “geometry effect” that Gavin refers to, and will hopefully be useful to some people in that regard.

  37. Brian Gordon:

    Re: 26″ “If water vapor is the most efficient greenhouse gas, what does such an economy portend?”

    Reponse by Elifritz: “The water vapor itself is trivial. Do I have to explain it to you, or can you click your way into science literacy?”

    I’ll bite on this snide reponse. (I ‘click’ to this site for some enlightenment.) Why is the water vapour trivial? If, for example, we replace all current fossil fuel and nuclear energy generation with H2 (assuming it was somehow generated with zero emissions), and the rest of the world somehow achieves our lifestyle, that’s 6.5B people spewing water vapour.

    Is it trivial because it promptly rains out or disappears into bodies of water? Is it trivial because, even in the 6.5B-people-on-H2 scenario, the water vapour we would emit is dwarfed by that in the atmosphere from natural sources? (Focusing on climate, because presumably a city full of water-vapour-emitting cars would increase local humidity, molds, etc)

  38. Steffen Christensen:

    Re: Chris Rijk, #30. Converting the 4 W/m^2 absorbed to a 1 K difference can be done by a slightly modified version of the Arrhenius equation. Arrhenius figured out the temperature of a blackbody at a given distance from the sun by equating the incoming energy, bouncing energy off using albedo, and assuming that the round body acts as a black-body emitter with emissivity 1. The zeroth-order approximation of climate comes from adding in a factor to this equation accounting for heat trapping; i.e. accounting for long-wave radiation emissivity. The final equation looks like this: Temperature = ((1-albedo) * solar_luminosity / (16 * sigma * emissivity * pi * solar_distance^2))^0.25. If you use SI units, the solar luminosity of the sun is 3.844E26 watts, the solar distance is 1 AU = 1.496E11 meters and sigma is the Stefan-Boltzmann constant = 5.6704e-8 W/m^2/K^4. The bond albedo of Earth is around .29 (the authors above are using .3), and you have to run the equation backwards to get the long-wave emissivity of the Earth in pre-industrial times. If you take the mean preindustrial temperature as 288.0 K (= 14.85 degrees Celsius), sub it in the equation, the time-averaged long-wave emissivity comes out to .6220. You can then compute the average intensity of the Earth’s outgoing radiation using Intensity = sigma * emissivity * Temperature^4. This comes out to 242.6 W/m^2 for preindustrial Earth. You then suck back 4 W/m^2 from that number for the doubled CO2, dropping it to 238.6 W/m^2, and recompute the emissivity for the new Earth. This gives .6117. Plug that back in the temperature equation, and voila, new Earth has a balanced temperature of 289.2 K. Subtract the two temperatures to get 1.2 K per 4 W/m^2 extra absorbance, as required. Or you could use calculus and get essentially the same number, which is what they are doing.

  39. Leonard Evens:

    With respect to Steffen’s comments in #28:

    I don’t know what calculations he is doing. I don’t myself know how to calculate any of this, but consider Gavin’s remark following #13. He says sensitivity to doubling CO_2 is given approximately by 4 divided by effective sensitivity. Working backwards, that would yield a range of 2.6 to 4.1 K in CO_2 doubling sensitivity from a range of about 1 to 1.4 in effective sensitivity. That seems broadly consistent with the range for effective sensitivity shown in the diagram. But I presume it is actually a bit more complicated than that for reasons similar to those I raised in my comment #1.

    Part of the trouble with reading these comments, particularly those getting into technical detail is that you don’t know how seriously to take them. Some people are talking with authority based on a thorough understanding of the research literature in the subject. That would include the RC moderators. Others are doing back of the envelope calculations based on what they imagine the theory should be. It is sometimes hard to tell one from the other.

    [Response: Well, our stuff is always in turquoise…. In the Soden and Held calculation they used 4.3 W/m2 for the instantaneous forcing, and so that would give a range of 1 to 1.7 W/m2/K for their effective sensitivity. – gavin]

    [Response: Whoops. Scrub that last comment. I misunderstood what they did. Their 4.3 W/m2 is the forcing from the scenario they were using, not from a 2xCO2 calc. -gavin]

  40. Eli Rabett:

    wrt to Gavin’s comment on #24, while his definition of TOA makes sense, the effective level from which the earth radiates is about 6 km (much lower than the tropopause) and one of the principal effects of raising greenhouse gas mixing ratios is to raise this level and make it wetter.

    Here is a paper which puts the TOA at 20 km, which is below the ozone layer

    and here is another one that puts it at 30 or 120 km

    [Response: Point taken. In practice ‘0 mb’ is difficult to precisely locate and so for various purposes levels lower down (i.e 0.1 mb (~60 km)) end up being used. However, in a model context, ‘0 mb’ makes sense and so all model-related uses of TOA really do refer to the very top. – gavin]

  41. Hank Roberts:

    Yep — “zero” depends on the sensitivity of the instrument. And will change all the time! Each time the sun flares the upper atmosphere expands. The SpaAlpha (the ISS) orbits at zero millibars, I’m quite sure, but still within the atmosphere. They photograph the aurora while flying through it.

    (Photo by Astronaut Pettit, the ISS Expedition 6 Crew)

    Also why Mir needed, and Alpha needs, to be regularly higher. And on the short term it’s unpredictable — every time the sun flares the upper atmosphere expands, drag on the spacecraft increases.

    That’s tangential to Gavin’s point, but it is a complicated answer and has to be measured at the moment the research is done and defined.

  42. Jeff Weffer:

    Using these formulae, what was the average temperature of the earth when the CO2 content of the atmosphere was higher, say 1.0%.

  43. Joel Shore:

    Re #42: Since the dependence of the forcing on CO2 levels is roughly logarithmic and 1% CO2 corresponds to roughly 36X the pre-industrial values, this would correspond to roughly 5 doublings so you would take the estimates for doubling and multiply by 5…I.e., if the climate sensitivity were 3 deg K per doubling, you would get a temperature of ~15K higher.

    However, there are probably lots of reasons why such an estimate would not be very accurate…most of which probably relate to the caveat “all else being equal”. In fact, the further you go back in time, the more things are likely to not be equal. For examples, the continents would be in different places and the amount of other stuff in the atmosphere (such as aerosols) might be very different. And, since the last time when CO2 levels were believed with confidence to be higher than they are today is on the order of 20 million years ago, it is indeed likely that such differences would have been significant. (In fact, to get up to levels of roughly 0.5% for CO2, it looks like you have to go back close to 200 million years according to this graphic: )

    Also, I don’t know if climate models really do forecast the temperature rise to remain approximately linear on the forcing (and the forcing to remain approximately logarithmic on the CO2 concentration) up to such large values since a paper in Science [Daniel P. Schrag and Richard B. Alley, Science, Vol. 306, pp. 821-822 (2004)] commented that current climate models could not simulate the warmth of some previous climates such as the eocene of ~50 million years ago (especially the high-lattitude warmth in continental interiors) no matter how high one turned up the CO2 levels. [Perhaps the experts can weigh in on this?]

  44. Steve Reynolds:

    Is the quoted 4 W/m^2 forcing for doubling CO2 a global average? Where could I find how this changes with latitude or other important factors? I assume CO2 is more important at high latitudes due to less masking by water vapor.
    Also, I’ve seen statements that the CO2 forcing is logarithmic with concentration. Is some data available showing the exact dependance?

  45. Steffen Christensen:

    Re: Leonard’s comments in #39.

    Sorry if I gave the impression that I am being authoritative, it wasn’t my intent. With respect to Brian’s post #12, I reran the calculations with a couple of different sets of numbers. The ones that worked were consistent with the data in the graph and with this statement of Brian’s: “The effective sensitivity can be computed by summing the individual feedback terms and including the Planck radiative damping (which is roughly -3.2 W/m2/K; see Table 1 of Soden and Held) and then flipping the sign,”. I took that statement to mean that you take the data graphed in the effective sensitivity column, add -3.2 W/m^2/K, and flip the sign to get the summed sensitivity. The data in the graph ranges between 0.85 W/m^2/K and 1.7 W/m^2/K, which gives us an ordinary summed sensitivity range of between 1.5 W/m^2/K and 2.35 W/m^2/K, which is fully consistent with the data presented, and follows your intuition of #1 that the errors are not independent. To get the temperature increase at equilibrium, I took the 4 W/m^2 from doubling CO2 giving a 1 K rise in the absence of feedbacks noted in the original post, and added the 1.5 W/m^2/K feedback rise on top of that, using the same 4:1 conversion ratio. This raises the temperature another 0.375 K. That temperature rise, in turn, must get amplified by the same feedback effects as well, adding 0.56 W/m^2 more input. That raises the temperature another 0.141 K. This continues until the terms get so small that they don’t count any more. To three decimal places, this gives a 1.600 K temperature rise at equilibrium. Same calculation for the upper bound gives a 2.424 K temperature rise. My post in #38 comes straight out of Physics class or a first class on physics in climate, for instance .

  46. Steffen Christensen:

    Re: Martin #31, thanks, good point. I thought I gathered from other posts here that there are some observational constraints on climate sensitivity based on historical data. Certainly I can’t expect models to give an independent test of the possible values, since they’re the same models. Of course, what is in and out of the models is a free parameter, and not all models use the same outside factors.

  47. Head in a Cloud:

    Still a thorn in our side / Positive cloud feedback?
    Cloud feedback still largest source of uncertainty in GCM’s
    What did they say? Clouds a positive feedback?
    Last week, a new article by Brian Soden of the University of Miami and Isaac Held of GFDL entitled “An Assessment of Climate Feedbacks in Co…

  48. Hank Roberts:

    >25, hydrogen effect on the ozone layer

    But will even a little more damage from leaking hydrogen be a problem in the near term? The Arctic ozone, not just the Antarctic, remains very depleted (and this is a feedback issue, assuming the Arctic stratosphere too is getting cold enough for catalysis on the surface of ice clouds to happen).

    “The severe Arctic ozone reduction in the winter 2004/2005 is analyzed … The average maximum ozone loss was about 2.1 ppmv at 475 Kâ��500 K (â�¼18 kmâ��20 km). Over 60% of the ozone between 425 Kâ��475 K (â�¼16 kmâ��18 km) was destroyed. The average total column ozone loss was 119 DU, â�¼20â��30 DU larger than the largest previously observed Arctic ozone loss in the winter 1999/2000.”

    On the good side I recall the current knowledge of hydrogen behavior is good enough that changes can be studied against a balanced baseline.

    Balancing Earth’s hydrogen budget

    If I recall correctly this year (quiet Sun, low part of the 11-year cycle) is making the ozone layer look better than usual.

  49. Ike Solem:

    Well, interesting comments. Let me point out that the climate sensitivity estimates are the results of GCMs, and no napkin scribbles are going to reproduce these results – if they could, this would have all been figured out 100 years ago. What about a variability index along with a climate sensitivity index? – averages can be misleading when it comes to the dynamics of an oscillating system.

    As I recall, the central problem with clouds in models was that grid scales in models were far too big (for reasons of computing power) to explicitly model cloud formation, so various parameterizations were used. If different parameterizations produce similar results, that’s somewhat encouraging. The original notion was that cloud tops have high albedo, reflecting incoming light to space and producing a cooling effect – but where in the troposphere the clouds form was also important. It does make some sense that the feedbacks are largely positive if one thinks about shady day UV sunburns – but I was surprised. Thanks for tackling this topic, also.

    I think this really points out the need for NASA to refocus on climate monitoring programs and to abandon the whole man on Mars concept. The potential surprises in the climate system are not likely to be revealed by theoretical modelling, but rather by observations and data collection. I won’t harp on any more about the disproved notion that glaciers are ice cubes, other then to point out that it was field data, not theoretical modelling, that led to that revelation. It looks like this info is starting to influence notions of glacial cycles, as well.

    For those interested in the Milankovitch forcings and the glacial cycles, the recent issue of Science has interesting reports and an overview by Didier Paillard at on this issue – and a nice graph showing that modern CO2 concentrations are exceeding any seen in the last 3 million years.

    There is also, in this issue, an article by Landsea et. al in which the recent record of tropical hurricane intensity is described as artifactual – the gist is that old data collection techniques underestimated hurricane intensity, so the observed trend of increasing intensity is just an artifact. Yet Landsea also uses the historical record of hurricane intensity as the basis for the North Atlantic Oscillation notion, stretching back to the early years of this century – What!? In this case it’s good data, in that case it’s bad data? Looks like a bit of a logical contradiction there, as far as I can tell.

  50. Leonard Evens:

    With respect to #39 and #45:

    I’m not color blind, so I do understand when the RC moderators comment. But other people also go into elaborate technical detail. Sometimes this amounts to just throwing words around without any real understanding of the science. Other times real experts are responding, and there is a continuum in between. When RC moderators respond, it becomes clearer how sensisble the comment was, but you don’t always do that, nor should you.

    [Response: It was just a joke, not a criticism. Apologies if that didn’t come over properly. Your main point is very well taken though and we do often comment on the seemingly technical stuff that could be misleading to other readers. The obvious errors don’t need so much correction… – gavin]

    I did think I understood the process you used to draw your conclusion. I taught many generations of caluclus students about the geometric series. It seems a plausible way to argue to me, but what do I know? What I didn’t understand was why it was at variance with the estimate of 2.6 to 4.1 K. Presumably what any given model does is hard to analyze by such methods, and the Soden Held paper is an attempt to grapple with that. I think we amateurs have to accept the 2.6 to 4.1 estimate as roughly correct whether or not it seems to square with a back of the envelope estimate we can do. From what you’ve said since I think you weren’t suggesting you had discovered some basic flaw but rather you were trying to understand, like the rest of us, just what is going on.