# Explaining how the water vapor greenhouse effect works

## What the science says...

Select a level... |
Basic
| Intermediate | |||

Increased CO2 makes more water vapor, a greenhouse gas which amplifies warming |

## Climate Myth...

Water vapor is the most powerful greenhouse gas

“Water vapour is the most important greenhouse gas. This is part of the difficulty with the public and the media in understanding that 95% of greenhouse gases are water vapour. The public understand it, in that if you get a fall evening or spring evening and the sky is clear the heat will escape and the temperature will drop and you get frost. If there is a cloud cover, the heat is trapped by water vapour as a greenhouse gas and the temperature stays quite warm. If you go to In Salah in southern Algeria, they recorded at one point a daytime or noon high of 52 degrees Celsius – by midnight that night it was -3.6 degree Celsius. […] That was caused because there is no, or very little, water vapour in the atmosphere and it is a demonstration of water vapour as the most important greenhouse gas.” (Tim Ball)

When skeptics use this argument, they are trying to imply that an increase in CO2 isn't a major problem. If CO2 isn't as powerful as water vapor, which there's already a lot of, adding a little more CO2 couldn't be that bad, right? What this argument misses is the fact that water vapor creates what scientists call a 'positive feedback loop' in the atmosphere — making any temperature changes larger than they would be otherwise.

How does this work? The amount of water vapor in the atmosphere exists in direct relation to the temperature. If you increase the temperature, more water evaporates and becomes vapor, and vice versa. So when something else causes a temperature increase (such as extra CO2 from fossil fuels), more water evaporates. Then, since water vapor is a greenhouse gas, this additional water vapor causes the temperature to go up even further—a positive feedback.

How much does water vapor amplify CO2 warming? Studies show that water vapor feedback roughly doubles the amount of warming caused by CO2. So if there is a 1°C change caused by CO2, the water vapor will cause the temperature to go up another 1°C. When other feedback loops are included, the total warming from a potential 1°C change caused by CO2 is, in reality, as much as 3°C.

The other factor to consider is that water is evaporated from the land and sea and falls as rain or snow all the time. Thus the amount held in the atmosphere as water vapour varies greatly in just hours and days as result of the prevailing weather in any location. So even though water vapour is the greatest greenhouse gas, it is relatively short-lived. On the other hand, CO2 is removed from the air by natural geological-scale processes and these take a long time to work. Consequently CO2 stays in our atmosphere for years and even centuries. A small additional amount has a much more long-term effect.

So skeptics are right in saying that water vapor *is* the dominant greenhouse gas. What they don't mention is that the water vapor feedback loop actually makes temperature changes caused by CO2 even bigger.

Basic rebuttal written by James Frank

**Update July 2015**:

Here is a related lecture-video from Denial101x - Making Sense of Climate Science Denial

Last updated on 5 July 2015 by pattimer. View Archives

Falkenherzat 22:01 PM on 5 October, 2012Bob Lacatenaat 23:35 PM on 5 October, 2012Philippe Chantreauat 03:08 AM on 6 October, 2012Falkenherzat 18:35 PM on 8 October, 2012Falkenherzat 18:37 PM on 8 October, 2012AlanSEat 05:43 AM on 10 January, 2013Bob Loblawat 05:59 AM on 10 January, 2013topof the atmosphere (i.e. the exchange with space). It is this latter value that creates the imbalance that leads to global warming - reduced losses to space require a warmer system to restore balance with absorbed solar. The failure to distinguish between surface fluxes and top-of-atmosphere fluxes is a common error.KRat 05:59 AM on 10 January, 2013ΔForcing = 5.35 W/m^2 (ln [CO2_new/CO2_previous])or ~3.7 W/m^2 per doubling of CO2. Sothese are indeed two different numbers for two different situations. The ΔF of 0ppm -> 390ppm does not equal the ΔF of 250pp -> 1000ppm, in particular since at low concentrations of a greenhouse gas (near zero) it will have a linear rather than logarithmic effect.Response:[DB] Fixed link.Tom Curtisat 07:13 AM on 10 January, 2013changein the total strength of the greenhouse effect as atmospheric concentration increases from 278 to 1000 ppmv. I notice in reading the graph from Science of Doom, you use the highest estimate of radiative forcing (IPCC 1990). Subsequent to the publication of the IPCC first and second reports, Myrhe showed that models where then overestimating the forcing from doubling of CO2. The correct value is 6.85 +/- 0.68 W/m^2 for an increase in CO2 from preindustrial levels to 1000 ppmv. Further, since the intermediate article above was last updated, Schmidt et al, (2010) (PDF) have come up with a more accurate estimate of the all sky CO2 contribution to the GHE of 29.5 W/m^2 for 1980 concentrations. That will have increased by 0.8 W/m^2 since then, to 30.3 W/m^2. A further increase of CO2 concentration to 1000 ppmv will increase the CO2 contribution by 5.0 W/m^2 relative to 2010 levels, taking the total CO2 contribution up to 35.3 W/m^2. It should be noted that most of the non-CO2 contribution to the total greenhouse effect comes from water and clouds, which would largely disappear from the atmosphere given a lack of CO2, and hence constitute a feedback rather than a forcing. I should also add that while the formula given by KR above is accurate for calculating change in forcing for levels of CO2 found in the atmosphere over the last 600 million years, but becomes inaccurate for very low levels of CO2 and so cannot be used to calculate the total contribution.Tom Curtisat 07:29 AM on 10 January, 2013Bob Loblawat 07:44 AM on 10 January, 2013"carbon dioxide contributes 32 W/m2 (Kiehl 1997). These proportions are confirmed by measurements of infrared radiation returning to the Earth's surface (Evans 2006).", and got messed up by the "earth's surface" part. Of course, with no atmosphere at all, there isn't any IR towards the earth's surface, but that still doesn't apply to an atmosphere with no CO2... which would have downward-direct IR at the surface. I was hasty, and messed up. The point that you need to be careful about surface vs. TOA fluxes is valid, but I'm not explaining the actual numbers that were asked about.AlanSEat 05:50 AM on 19 January, 2013Tom Curtisat 11:31 AM on 19 January, 20131) Greenhouse effect:The explanation of the greenhouse effect at sierrapotomac.org is of poor quality, and will only confuse you if you are trying to understand it correctly. In particular, it describes the greenhouse effect as causing a greater absorption of heat. In fact, an increase in atmospheric CO2 would result in a greater increase in heat only while the Earth was not a radiative equilibrium. Once the Earth reached radiative equilibrium, there would be no further net gain in heat, but the greenhouse effect would still be enhanced relative to the condition with less CO2. It also says,"The heating of the earth due to the radiant heat of the sun is called the greenhouse effect", which is egregiously wrong. The radiant energy received from the Sun at the Earth's current albedo is 239 W/m^2. In the absence of a Greenhouse effect, the temperature of the Earth's surface would rise till it emitted 239 W/m^2, ie, approximately 255 degrees Kelvin (-18 C). A low IR emissivity would raise the temperature slightly; while a less than even temperature distribution across the surface would lower it. In practice, the second is the stronger effect so that the mean global surface temperature would be less than 255 K. As it happens, the Earth's global mean surface temperature is approximately 288 K (15 C). The higher temperature results in a much higher outward IR flux at the surface than the energy received from the Sun, and indeed, much higher than the outward flux at the top of the atmosphere. The difference between the outward flux at the surface and that at the top of the atmosphere is the atmospheric greenhouse effect. Trenberth et al (2010) give that difference as 156 W/m^2 (see diagram below), while Schmidt et al (2010) gives it as 155 W/m^2. It is very important that the IR radiation at the TOA is less than that at the surface only because greenhouse gases absorb outgoing IR radiation; and because those gases are cooler than the surface, so that when they emit IR radiation it has a reduced flux. It is also important to recognize that the TOA flux can only be smaller because energy absorbed by the atmosphere is also transferred to the surface. Without the energy transfer to the surface both the reduced flux at TOA, and the flux at the surface greater than solar radiation absorbed would violate conservation of energy. As it happens, the energy equations do balance (see diagram above). The energy transfer from atmosphere to surface is in the form of back radiation, but could be in another form and you would still have a greenhouse effect.2) Atmospheric Window:You appear to be confused by the "atmospheric window". "Atmospheric windows" are frequencies within the electromagnetic spectrum in which there is almost no atmospheric absorption, so that radiation in the window can go directly from the surface to space (or vise versa). As can be seen in the modtran image below, there is an IR atmospheric window between wavenumbers 800 cm^-1 and 1000 cm^-1, and another smaller window around 1100 cm^-1. As it happens, about 40 W/m^2 escapes from the surface to space through these "atmospheric windows", but they should not be confused with the total radiation to space (236 W/m^2, most of which comes from the atmosphere) or with the atmospheric greenhouse effect. As noted in section one, the atmospheric greenhouse effect is the difference between the upward IR radiation from the surface and the upward IR radiation to space from the Top of the Atmosphere. Doubling CO2 concentration creates a radiative forcing of 3.7 W/m^2, ie, it reduces the IR radiation to space by 3.7 W/m^2. The reduced upward flux at the TOA creates an energy imbalance which warms the Earth until the imbalance ceases to exist. Ignoring all feed backs, that requires a warming of approx 1.1 C at the surface to accomplish; or in other words an increase in the upwards IR flux at the surface by about 6 W/m^2. Because equilibrium is reached (ignoring feedbacks), the upward IR flux at the TOA will have returned to 239 W/m^2, so aradiative forcingof 3.7 W/m^2 will have caused a total change in thenet atmospheric greenhouse effectof 6 W/m^2. In practice, the net change will be approximately two to three times that amount, with much of the increase attributable to the water vapour feed back. This may seem confusing, but only if you mistake radiative forcing, ie, the net change in TOA radiative flux before temperature adjustments with the net atmospheric greenhouse effect, ie, the difference between surface and TOA upward IR flux. Unfortunately, that is a mistake I made in my previous post @134. I apologize for any confusion I have caused as a result.3) Increased CO2:Increasing CO2 in the atmosphere widens the large valley in outgoing radiation between 600 and 700 cm^-1 (see modtran graph above). To maintain equilibrium, the total area of the graph, which represents the total upward flux, must remain constant. Because the area is reduced near 600 and 700 cm^-1, it must be increased elsewhere, including in the atmospheric windows. As the upward flux in the atmospheric windows comes from the surface, this means the surface temperature must increase. Consequently your assumption that most of the warming would occur in the atmosphere (ignoring feed backs) is mistaken.Jose_Xat 09:33 AM on 10 March, 2013Tom Curtis #138, I wanted to add (correct it if wrong) that the forcing value is a theoretical value. It's as if the entire change were done instantly. In reality, we experience CO2 increases, not from 1x (of some arbitrary value) to 2x overnight, but as small increments over time, so the imbalance is likely never to get very large at all. The 1x is a reference point and the 3.7 forcing value is a theoretical imbalance that would exist if the 2x happened right away ("with surface and tropospheric temperatures and state held fixed at the unperturbed values").

Tom Curtisat 10:23 AM on 10 March, 2013Jose_X @139, that is mostly correct.

As noted @138 (point 2) , the CO2 forcing is the total net change in TOA upward IR flux before (or excluding) temperature adjustments. Clearly if we add the CO2 incrementally, there will be temperature adjustments so that the TOA energy imbalance will be significantly less than the CO2 forcing. But, the forcing is the same at 2xCO2 regardless of whether it is added incrementally or as a single slug.

That does not differ from your explanation except in terms of what is meant by "theoretical variable". Roughly (as it is a while since I studied this), a theoretical variable in physics is a value which is an element of an empirical theory that is not directly measurable itself, but whose values have direct implications for measurable variables. In this sense, forcing is "theoretical". You appear to treat "theoretical" as a synonym for "hypothetical" which would be incorrect.

Jose_Xat 08:43 AM on 11 March, 2013Tom Curtis, thanks for the response, but I was thinking of "theoretical" just as you explained, meaning that if we could actually do 2x over night then we would observe the full forcing value.

Tom Curtisat 09:17 AM on 11 March, 2013Jose_X @141, fair enough. In that case, however, you should have written "the 3.7 forcing value is a theoretical imbalance that would exist

only be directly measurableif the 2x happened right away" or possibly, "the TOA energy imbalance would only equal the forcing if the CO2 was doubled instantaneously".Even that is not perfectly accurate, in that short term fluctuations in temperature result in short term departures from radiative equilibrium, but they are small relative to the 3.7 W/m^2 forcing from doubling CO2.

Jose_Xat 03:30 AM on 13 March, 2013Tom Curtis, thanks for the response, but I was thinking of "theoretical" just as you explained, meaning that if we could actually do 2x over night then we would observe the full forcing value.

Tom Curtis #142. Rethinking the meaning of the words now after your 142 comment, I see that you were right and I did mean hypothetical. I didn't mean that the value couldn't exist under any circumstance, but rather that it could but would not unless the CO2 were eg done "over night". I concluded this from the statement by the IPCC (quoted on Wikipedia) that a reference value is used to define the forcing ("with surface and tropospheric temperatures and state held fixed at the unperturbed values") so that perhaps a perturbation "forcing" is not an entity (theoretical or measurable) that is manifested in reality if there isn't an imbalance at TOA of that magnitude at any given point in time.

Let me ask something about the meaning of the word "forcing". Would "yes" be the answer to the question, "is the forcing at TOA from CO2 a value greater than 5 W/m^2?" [ie, is it clearly greater than the 2x forcing quoted by the IPCC because it would include contributions from all CO2 in the atmosphere today?]

KRat 03:55 AM on 13 March, 2013Jose_X - It's a bit difficult to understand what you are discussing here. Are you asking for the current sum value of CO2 forcing? According to NASA numbers CO2 represents ~20% of the current greenhouse effect, water vapor and clouds ~75%, and other gases (methane, CFCs, ozone, etc) another 5%.

33C of greenhouse effect, or 6.6C directly attributable to CO2 by those numbers, means about 22-23 or so W/m^2 of total CO2 forcing at this time.

On the other hand, "

forcing" is generally referred to as the change from previous conditions, usually 1750AD, in which case as of 2005 CO2 accounts for ~1.66 (1.49-1.83) W/m^{2}as a delta. Add in all the other factors +/-, and the total net anthropogenic forcing change since 1750 is about 1.6 (0.6-2.4) W/m^{2}.[Source]

But again, I'm not entirely certain what you are asking about.

Jose_Xat 04:52 AM on 13 March, 2013I was not looking at "forcing" as something like a gravitational force value associated with a body on the earth (which tends to exist all the time) but more like (net) heat, a quantity defined as the summation of other (sign-dependent) quantities and which might frequently approach or be zero. I guess that was wrong.

So, to confuse everyone still further, I guess forcing is somewhat analogous to pressure in that it tends to be nonzero and significant whether the system is approximately in equilibrium or not and under a context where "pressure" would refer to partial pressure of a species or, more specifically, to the additional contribution to partial pressure that would be provided as the species moves from one reference concentration to a different target point separated from the first by a stadardized multiple (of 2).

Jose_Xat 08:56 AM on 13 March, 2013Are the following observations correct?

And does someone have an answer for the two questions at point 3.

1. The forcing value depends on the reference sun intensity. If the sun was providing approximately a stable 50 W/m^2 TOA, then the CO2 forcing calculations would yield different numbers.

2. For the lambda (climate sensitivity proportionality coefficient) between forcing at TOA and (yearly global mean) temperature on the surface to be approximately constant, we should use a reference point for forcing such that the delta temperatures involved are at most a few K. [This criteria would include a "near" constant solar intensity requirement that has easily been met by sun+albedo across spans of a few centuries.] This would allow, for example, the relevant Stefan Boltzmann relationships to remain in an approximately linear region, eg, where each degree change in temp would correspond approximately to a constant change in W/m^2 (currently approaching 5.4 W/m^2 per C at ground level).

Also, I was guessing that, although the lambda and forcing values might be determined from computer programs that may potentially take a long time to run in order to achieve a high level of accuracy, that using simple models (esp linear) afterward with those derived values allows researchers to ask and answer many questions involving lots of combinations of forcing agents and quantities without having to run involved simulations for every such combination of values. The simple models also help us understand the main factors in climate change and together with all models used help keep a sanity check on each other.

3. Calculating the contributions from CO2 to the greenhouse effect (using data from KR's recent comment):

287.5 - 255.5 = 33 (relevant temperature values to nearest 0.5 C)

33 / 6.6 = 5 (aka, 20% contribution by CO2)

387.38 - 241.63 = 145.75 (Stefan Boltzmann W/m^2 at surface corresponding to 33 C increase and e=1)

145.75 / 5 = 29.15 (contribution by CO2 to total W/m^2 at surface)

So there have been 29.15 W/m^2 of warming due to CO2 for its 6.6 C contribution to the total greenhouse effect of 33 C.

Question 1: How do we calculate the total forcing by CO2 that led to this 29 W/mSq?

Put differently, if we revert to an atmosphere with almost no ghg effect, meaning that the TOA and surface are roughly at 255K, how many times will we have to apply a forcing of 1 W/m^2 (TOA) from CO2 additions to get to our current atmosphere? Can someone roughly (including hand waving) show the steps in unit increments that achieve this and so that we can see how the ground temp or ground irradiance change at each step? [For the purposes of addressing this question, on may optionally assume H2O is liquid all throughout the 33 C range.]

Question 2: Whatever approximation method is used to answer Question 1, will we get roughly the same answer if we go in 5 W/m^2 increments?

My confusion is that forcings are defined at TOA yet TOA always reverts to about 240 W/m^2. And there isn't to me an obvious way to derive that total forcing value (eg, from CO2.. or even from all ghg) simply from knowing how much change was experienced on the ground (the greenhouse effect). I can see how that question might be answered possibly by calculating in delta/differential steps and taking a limit in order to try and get a unique result (although I don't know how to make that calculation). Alternatively, it might not be possible to get sequence/limit convergence to a unique value on a "delta" analysis. I don't know, and before I spend more time thinking about this problem, I'm curious if someone has access to an online reference where I can get the answer or if they can provide the answer themselves.

Resolving these questions makes it easier for me to address more definitively some skeptical questions I have been getting related to "forcings" math (and in understanding this for its own sake).

KRat 11:00 AM on 13 March, 2013Jose_X - Some misconceptions here, if I might point them out (hopefully correctly).

"How do we calculate the total forcing by CO2 that led to this 29 W/m^2?"If you start from zero CO2, the forcing per increase in CO2 concentration starts linear, and becomes logarithmic as various bands reach saturation (with increases coming from band widening, rather than peak increases), so the relationship is not consistent over concentrations. Methane (IIRC) is still in the linear region, CO2 is not. The accurate answer comes from line-by-line radiative codes such as MODTRAN using the HITRAN spectral database - essentially numeric integration. You have to do the math. There is no simple equation.

"My confusion is that forcings are defined at TOA yet TOA always reverts to about 240 W/m^2"Yes, it does, as 240 W/m^2 is what is incoming from the sun. When enough time elapses for GHG forcing changes to come to equilibrium,

energy out = energy inat TOA, although at a different surface temperature depending on those changes. The only reason for that equilibrium number to change would be changes in incoming energy, perhaps from albedo/land use or significant cloud percentage/distribution. I do expect that the melt of the Arctic ice cap, for example, will raise that equilibrium number somewhat by decreasing summer albedo.Tom Curtisat 11:11 AM on 13 March, 2013Jose_X @146:

1) Because the CO2 modulates the outgoing LW radiation, it depends on the strength of that radiation, and hence on surface temperature. That does mean it depends on the strength of the incoming solar radiation, but only indirectly. In contrast, changes in aerosol opitical depth, which change albedo, are directly dependent on the strength of the incoming solar radiation for the strength of the forcing.

2) Across the range of temperatures and conditions experienced in the Phanerozoic (approx -6 to +8 C relative to present values) λ has been fairly constant, with changes in continental configurations having a much larger effect on changes in λ than variation withing the temperature range. In broader terms, λ becomes significantly larger as ice sheets approach 30

^{o}Latitude (North or South) due to the much enhanced albedo, and as temperatures rise significantly beyond 8 C above current values (due to enhanced water vapour feedback).Other than noting that your second equation in section (3) should be 33/5 = 6.6, I will return to it when I have a bit more time.

Jose_Xat 11:26 AM on 13 March, 2013Tom Curtis, actually the equation should probably have been 6.6/33 = .2 (aka, 1/5) and then rather than divide the 5 into the 145.75 I would have multiplied the .2 and 145.75. Point is that 6.6 is 20% of the whole change as expressed in C so I wanted to find that same 20% of the whole change in W/msq.

Jose_Xat 11:34 AM on 13 March, 2013KR, calculating the forcing for CO2 early on in the process (near 0% CO2) and continuing, eg, using at each step the integration process you mentioned, is what I imagined might be done. But I am curious if that procedure would lead to the same total forcing result no matter in how many steps we calculate the increments.

As an extreme example, if we calculate the forcing of one additional molecule of CO2 at a time (assuming the standard model we use now applies, eg, ignoring quantum or other effects that might be present at very low concentrations or what not, and assuming that computation could finish some day), what total forcing would we get and would this value be the same if after an initial start we switch over to steps where we double the prior value.

Why do I care? I would like to know if the forcing calculations adhere to linear superposition like linear operators do with vector (tensor?) quantities (ie, partitioning into component parts arbitrarily, operate on these separately, and then combined back additively into a unique whole). The definition of forcing makes me a little nervous about that.

Does someone have software that can simulate several reasonable partitions for the current CO2 to test if each partitioning path leads to the same answer? [An example might be to calculate 1 W/msq of forcing at each step vs 2 W/msq at each step. In fact, there is no need to do this until the current atmosphere. We can just compare 1 W/sqm for 40 steps vs 2 W/sqm for 20 steps.]

[In these replies I giving, m^2 is same as mSq is same as sqm ...etc. I get lazy with the keystrokes sometimes.]