Quantifying the Greenhouse Effect

When I began having conversations about the atmospheric Greenhouse Effect (GHE), those conversations were often burdened by vagueness.

It was said that the GHE meant “the Earth would be colder if it weren’t for the presence of Greenhouse gasses.” Unfortunately, that formulation seemed to leave the GHE in the realm of the hypothetical.

It led to fruitless arguments about the parameters of the hypothetical scenario. After all, water vapor is a major Greenhouse gas (GHG), and if that was removed, a lot of other things about the situation would change, not just the absorption and emission of longwave radiation in the atmosphere. And, that way of talking about the GHE also made it difficult to point to anything observable, in the world as it is, that represents the GHE.

So, I was relieved when I finally (all too recently) came to understand that there is a measure of the Greenhouse effect that is entirely quantifiable and measurable in the world as it is.

Quantitatively, the GHE is the difference between the radiative flux of longwave (LW) thermal radiation emitted by the Earth’s surface (denoted \SLR) and the radiative flux of outgoing LW thermal radiation emitted to space at the top of Earth’s atmosphere(denoted \OLR):

(1)   \begin{equation*} \GHE = \SLR - \OLR \end{equation*}

For present-day Earth, the values of these quantities are around \SLR = 398 W/m2, \OLR = 239 W/m2, and \GHE = 159 W/m2.

Why does it make sense to refer to this as being the quantitative definition of the Greenhouse effect? There are two reasons:

  1. If there was nothing in the atmosphere that could absorb, emit, reflect, or scatter LW radiation (the things said to be responsible for the GHE), then the amount of LW thermal radiation reaching space would have to equal the amount emitted by the surface, \OLR = \SLR. That would mean the Greenhouse effect, as I’ve defined it, would be zero, \GHE = 0. Thus, the value of \GHE is a measure of the energetic effect of atmospheric materials which interact with LW thermal radiation.
  2. \GHE, as it has been defined, directly relates to the portion of a planet’s surface temperature that is attributable to the effect of atmospheric materials which interact with LW thermal radiation.

Table of Contents

Relationship to surface temperature

Let’s look at how \SLR and \GHE relate to temperature. The Stefan-Boltzmann law (SB law) tells us that

(2)   \begin{equation*} \SLR = \emis \,\sigma \,\tsurf^4 \end{equation*}

where \emis is the emissivity (a value between zero and 1, typically close to 1, which is characteristic of the material emitting thermal radiation), \sigma is the Stefan-Boltzmann constant, and \tsurf is the temperature of the surface.

The SB law applies at each individual location. However, one can use it to obtain the global average \SLR by averaging the equation globally. That leads to a version of the SB equation which applies globally:

(3)   \begin{equation*} \SLR = (1+\tveb) \,\emis \,\sigma \, \tsurf^4 \end{equation*}

In this SB equation for global averages, \SLR is the global mean surface longwave radiative flux, \tsurf is the global mean surface temperature (\GMST), \tveb is what I call the temperature-variation emissions boost factor (greater than or equal to 0), where (1+\tveb) is defined as the ratio of the average \tsurf^4 to \GMST^4, and \emis is the weighted global mean emissivity (weighted by \tsurf^4).

So, given \SLR, \emis, and \tveb, we can calculate the average temperature.

  • \SLR is estimated to be 398 W/m2.
  • \tveb = 0.0187 ± 0.0003 for Earth based on my analysis of CERES data.
  • One data set puts Earth’s area-weighted average surface emissivity at \emis = 0.935. However, I don’t think there is universal agreement on this value.

These values yield \tsurf = 293 K / 20℃ / 68℉.

We can think of the surface temperature, \tsurf, as being equal to the sum of two parts:

(4)   \begin{equation*} \tsurf = T_\mathrm{ng} + \Delta T_\mathrm{ghe} \end{equation*}

Here, the “no GHE” temperature, T_\mathrm{ng}, is the temperature we would compute from the SB law if \SLR = \OLR as would necessarily be the case if there was nothing in the atmosphere that could intercept LW radiation, and \Delta T_\mathrm{ghe} = \tsurf- T_\mathrm{ng} is the portion of the temperature that we attribute to the existence of the Greenhouse effect.

Using the SB law for global averages, one can show that:

(5)   \begin{equation*} T_\mathrm{ng} = \sqrt[4]{\frac{\OLR}{(1+\tveb)\,\emis\,\sigma}} \end{equation*}

(6)   \begin{equation*} \Delta T_\mathrm{ghe} = \left(1 -  \sqrt[4]{1-\frac{\GHE}{\SLR}} \right) \tsurf \end{equation*}

For Earth, \Delta T_\mathrm{ghe} = 0.1197 \tsurf.

Assuming \emis = 0.935 and \tveb=0.0187, then T_\mathrm{ng} = 258 K / -15℃ / 5℉ and \Delta T_\mathrm{ghe} = 35℃ / 63℉.

So, as advertised, the quantity \GHE relates to the temperature of the surface. The Earth’s surface temperature can be thought of as involving two terms:

  • A term T_\mathrm{ng}, which is the temperature Earth’s surface would necessarily have if (a) there were no materials in Earth’s atmosphere that could interact with LW thermal radiation, and (b) the outgoing longwave radiation flux, \OLR, and emissivity, \emis, were unchanged.
  • A term \Delta T_\mathrm{ghe}, which is non-zero only because atmospheric materials that interact with LW thermal radiation make it possible for OLR to be different than SLR.

GHE as energy vs. GHE as temperature

The quantities \GHE and \Delta T_\mathrm{ghe} can be thought of as, respectively, the expression of the GHE in terms of radiative flux, and the expression of the GHE in terms of surface temperature.

Either of this quantities could legitimately be called “the Greenhouse effect.”

However, in technical discussions about climate, I believe it’s usually more useful to talk about the GHE expressed in terms of radiative flux, as define by the quantity I’ve denoted \GHE.

Note that in some of my work I might use the single-letter symbol, \ghe, to denote the quantity \GHE as I’ve defined it above.

Dimensionless Greenhouse effect

It can be useful to express the GHE in terms of a dimensionless value:

(7)   \begin{equation*} \nghe = \frac{\GHE}{\SLR} = \frac{\SLR - \OLR}{\SLR} = 1 - \frac{\OLR}{\SLR} \end{equation*}

The symbol \nghe denotes the normalized greenhouse effect. The tilde over the “g” helps avoid confusing it with the gravitational constant, g (9.8 m/s2), but for casual work it can be okay to omit the tilde and use g to denote the normalized GHE, as long as it’s clear what is being talked about. In some of my work, I’ve alternatively thought of this quantity \nghe as being the longwave effective absorptance, \LEA, since it indicates the fraction of LW surface emissions that effectively do not reach space.

\nghe is a dimensionless quantity between 0 and 1. It is zero when no atmospheric materials interact with LW thermal radiation.

For Earth at present, \nghe \approx 0.4.

Other quantities may be expressed in terms of \nghe as follows:

(8)   \begin{equation*} \GHE = \nghe \cdot \SLR \end{equation*}

(9)   \begin{equation*} \OLR = (1-\nghe)\cdot\SLR \end{equation*}

(10)   \begin{equation*} \Delta T_\mathrm{ghe} = \left(1 -  \sqrt[4]{1-\nghe} \right) \tsurf \end{equation*}

\LEA or \nghe is the effective LW absorptance of the atmosphere, not the actual LW absorptance:

  • The actual LW absorptance indicates what fraction of LW power sent upwards through the atmosphere is eventually absorbed.
  • \LEA or \nghe indicates what fraction of LW power sent upwards through the atmosphere is eventually absorbed without be replaced by an equivalent amount of LW power emitted from within the atmosphere; \LEA the net effect after both absorption and emission are taken into account.

Global and local versions of GHE

The GHE is usually talked about as a global effect related to global average radiative fluxes and global average surface temperature. All the specific values I’ve calculated for Earth have reflected this global orientation.

However, it is also valid to interpret the quantities I’ve worked with above as being local or location specific, i.e, having values which depend on the particular latitude and longitude, (\theta, \phi). In particular, \GHE, \nghe, and \Delta T_\mathrm{ghe} can be evaluated locally, provided we omit the factor \tveb which is relevant only in the context of global mean values.

Each equation I’ve presented so far remains valid if the quantities are interpreted locally rather than globally (aside from the factor \tveb being present in global equations and absent in local equations).1Note, however, that this particular description of the Greenhouse effect hasn’t included equations derived from TOA energy conservation. Those equations would be need to be modified to be valid, if interpreted locally. In particular, one would need to include a term for advection or lateral heat transport. Lateral heat transfer averages to zero globally, and so does not appear in global equations.

It may be useful to be more explicit about the relationships between local and global values. For some quantities, the global value needed to make the equations work is a simple area-weighted global average. For others, it is a weighted average. And, for still others, deriving the global value is best done by referring to another global value, rather than through working directly with the location-specific values.

I’ll use the notation \ex{X} to refer to the unweighted (i.e., area-weighted) global average of X, and \exw{X}{W} = \ex{X\cdot Y}/\ex{Y} to refer to the global average of X weighted by Y. To explicitly differentiate local and global versions of a quantity, I will write the local quantity as \xloc{X} and the global quantity as \xglob{X}. For many quantities, the global value will simply be the unweighted average of the local value. But, for some quantities, the global quantity \xglob{X} will be calculated another way.

For the following quantities, the relevant global version of the quantity is simply the unweighted global average:

  • Radiative fluxes: \SLR, \OLR
  • Greenhouse effect: \GHE
  • Surface temperature: T

The global values of these quantities are calculated as the indicated weighted average:

  • Emissivity: \xglob\emis = \exw{\xloc\emis}{T^4} (and replace \xloc\emis with \xglob\emis\cdot(1+\tveb))
  • Dimensionless GHE or LEA: \xglob\nghe = \exw{\xloc\nghe}{\xloc\SLR}

The global values of these quantities should be calculated from other global quantities, rather than trying to calculate them from their local values:

  • GHE temperature component: \Delta T_\mathrm{ghe}
  • No-GHE temperature: T_\mathrm{ng}

In the global equations, if \tsurf is not uniform, then wherever \emis appears, it should be replaced by \xglob\emis\cdot(1+\tveb), where \tveb or \TVEB is the temperature variation emissivity boost factor, \tveb \geq 0, defined as:

(11)   \begin{equation*} 1+\tveb = \frac{\ex{\tsurf^4}}{\ex{\tsurf}^4} \end{equation*}

Small changes

For small changes in temperature and emissions, one can effectively linearize the SB law. Small changes in surface thermal emissions and surface temperature are related, to a good approximation, by:

(12)   \begin{equation*} \Delta\SLR \approx \frac{4 \cdot \SLR}{\tsurf} \cdot \Delta \tsurf \end{equation*}

(13)   \begin{equation*} \Delta  \tsurf \approx \frac{ \tsurf}{4 \cdot \SLR} \cdot \Delta\SLR \end{equation*}

These equations apply to both local and global quantities.

Response to TOA radiative forcing

It is important to keep in mind that radiative forcing \Delta F is not equal to change in \SLR.

The effect of GHGs is often quantified in terms of the top-of-atmosphere (TOA) radiative forcing it produces. The TOA radiative forcing, \Delta F associated with a change in GHG concentration is the negation of the change in \OLR that would result if the GHG concentration change happened instantly and the surface wasn’t able to make any adjustments in response (so \SLR is unchanged), and atmosphere was only allowed to make a few minor adjustments. (There are lots of technicalities in the definition, and several variants of the definition.)

Momentarily and theoretically, given those constraints:

(14)   \begin{equation*} \Delta \OLR = -\Delta F \end{equation*}

This implies an equivalent change in \nghe:

(15)   \begin{equation*} \Delta \nghe = \frac{\Delta F}{\SLR} \end{equation*}

However, it’s important to notice that this state, with \Delta \OLR = -\Delta F (and with \Delta\SLR=0) is an initial state, in which there is a TOA energy imbalance and the system is not in thermal equilibrium.

When we ask about how \SLR will change, we are not asking about this state, but about the response of the system after it arrives at thermal equilibrium.

By the time that response happens, there will be no flux or change in flux anywhere in the system that has a value \Delta F. Instead, there will be an equilibrium state in which \Delta\OLR will be zero and \Delta\SLR has some nonzero value.

There is no need for \Delta\SLR to equal \Delta F any more than there is a need for the distance that one moves one end of a lever (the “input” movement) to match the distance that the other end of the lever will travel (the “output” movement).

Just as \OLR \leq \SLR with \OLR being smaller than \SLR by a factor (1-\lea), \left| \Delta F \right| \leq \left| \SLR \right| with \Delta F being smaller than \Delta\SLR by a response factor -\alpha defined so that:

(16)   \begin{equation*} \Delta T = \frac{\Delta F}{-\alpha} \end{equation*}

(17)   \begin{equation*} \Delta \SLR = \frac{4 \cdot \SLR}{-\alpha \cdot T} \cdot \Delta F \end{equation*}

Climatologists use an (arguably regrettable) standard sign convention for \alpha such that the net value of \alpha is generally negative, and a negative value of \alpha means that a positive forcing \Delta F at TOA leads to a surface temperature increase \Delta T and surface emissions increase \Delta\SLR. (For that reason, when using official notation, I usually write -\alpha, since I expect that -\alpha will be positive.)

I’ve only seen -\alpha be used in the context of global equations, not location-specific equations. Although we could define a location-specific multiplier, that would appear to be outside the scope of what I’m aware of others doing. Yet, if you’re going to write location-specific equations, be aware that the actual multiplier will surely vary by location.

The value of -\alpha depends on whether one puts any constraints on the system’s ability to respond, and on how long one waits for a response. As one waits for a response, different aspects of the climate system have time to respond.

The IPCC typically focuses on three different time frames, a short-term time frame, a 70-year mid-term time frame which is reported on via a metric called the Transient Climate Response, TCR, and a centuries long long-term time frame which is reported on via a metric called the Equilibrium Climate Sensitivity, ECS.

It’s critically important to understand that, even in the short term \Delta \SLR \neq \Delta F.

I’ve shown elsewhere[LINK] that for an instantaneous forcing \Delta F, if the system is constrained so that nothing but surface temperature is permitted to adjust as the system comes to equilibrium, then:

(18)   \begin{equation*} q = -\alpha = 4 \cdot \ex{\OLR}/\ex{\tsurf} \end{equation*}

(19)   \begin{equation*} \Delta \SLR = \frac{\ex{\SLR}}{\ex{\OLR}} \cdot \ex{\Delta F} \end{equation*}

(20)   \begin{equation*} \Delta \ex{\tsurf} = \frac{\ex{\tsurf}}{4 \cdot \ex{\OLR}} \cdot  \ex{\Delta F} \end{equation*}

I’m using location-specific versions of T, \SLR, \OLR and \Delta F in these equations, and am explicitly indicating how they are averaged.

For Earth at present, q \approx 3.3 W⋅m-2⋅K-1.

The IPCC defines a response called the Planck response, -\alphap, which corresponds to similar constraints (though there may be nuances that I have yet to digest). They report that 3.3 W⋅m-2⋅K-1 is a “crude guess” at the value of -\alphap, but estimate a most likely value of -\alphap = 3.2 W⋅m-2⋅K-1. (See IPCC Climate Change 2021: The Physical Science Basis, p. 968.)

Spectral version of GHE

One can define versions of \GHE and \nghe that characterize the GHE for specific frequencies or wavelengths of LW radiation.

Suppose E_\SLR(\nu) and E_\OLR(\nu) are the spectral irradiance curves that describe the spectra of LW emissions at the surface and TOA, respectively.

Then we could define the spectral GHE, E_\GHE(\nu), as:

(21)   \begin{equation*} E_\GHE(\nu) = E_\SLR(\nu) - E_\OLR(\nu) \end{equation*}

Similarly, we could define the spectral longwave effective absorptance (or spectral normalized GHE), \nghe(\nu), as:

(22)   \begin{equation*} \nghe(\nu) = \frac{E_\SLR(\nu) - E_\OLR(\nu)}{E_\SLR(\nu)} \end{equation*}

See Also

These results are derived formally on this page: Analysis: Planetary Temperature – a Rigorous Formula

This topic is discussed in an overlapping-but-different fashion in this blog post: 11+12 =23, or How I know the Greenhouse Effect is real (The discussion on the current page doesn’t depend in any way on heat sources, but depends only on outgoing radiation. The blog post takes a more conventional approach, focusing on the balance between incoming and outgoing energy flows.)


The definition of the GHE I’ve described is used, for example, in: