Dependence of Earth’s Thermal Radiation on Five Most

Abundant Greenhouse Gases

W. A. van Wijngaarden1 and W. Happer2
arXiv:2006.03098v1 [physics.ao-ph] 4 Jun 2020

1 Department of Physics and Astronomy, York University, Canada, [email protected]

2 Department of Physics, Princeton University, USA, [email protected]

June 8, 2020

The atmospheric temperatures and concentrations of Earth’s five most important, green-
house gases, H2 O, CO2 , O3 , N2 O and CH4 control the cloud-free, thermal radiative flux
from the Earth to outer space. Over 1/3 million lines having strengths as low as 10−27 cm
of the HITRAN database were used to evaluate the dependence of the forcing on the gas
concentrations. For a hypothetical, optically thin atmosphere, where there is negligible sat-
uration of the absorption bands, or interference of one type of greenhouse gas with others,
the per-molecule forcings are of order 10−22 W for H2 O, CO2 , O3 , N2 O and CH4 . For cur-
rent atmospheric concentrations, the per-molecule forcings of the abundant greenhouse gases
H2 O and CO2 are suppressed by four orders of magnitude. The forcings of the less abundant
greenhouse gases, O3 , N2 O and CH4 , are also suppressed, but much less so. For current
concentrations, the per-molecule forcings are two to three orders of magnitude greater for
O3 , N2 O and CH4 , than those of H2 O or CO2 . Doubling the current concentrations of CO2 ,
N2 O or CH4 increases the forcings by a few per cent. These forcing results are close to
previously published values even though the calculations did not utilize either a CO2 or H2 O
continuum. The change in surface temperature due to CO2 doubling is estimated taking into
account radiative-convective equilibrium of the atmosphere as well as water feedback for the
cases of fixed absolute and relative humidities as well as the effect of using a pseudoadiabatic
lapse rate to model the troposphere temperature. Satellite spectral measurements at various
latitudes are in excellent quantitative agreement with modelled intensities.

1
1 Introduction
The temperature record from 1850 to the present shows the average surface temperature
of the Earth has increased by about one degree Celsius[1]. The Interovernmental Panel on
Climate Change (IPCC) attributes most of this temperature rise due to increasing greenhouse
gas concentrations associated with anthropogenic activity. The average concentration of CO2
in the atmosphere has increased from 280 ppm to over 400 ppm largely due to the combustion
of fossil fuels. Concentrations of N2 O and CH4 have also risen substantially since the start
of the industrial revolution [2].
Greenhouse warming of Earth’s surface and lower atmosphere is driven by radiative forc-
ing, the difference between the flux of thermal radiant energy from a black surface through a
hypothetical, transparent atmosphere, and the flux through an atmosphere with greenhouse
gases, particulates and clouds, but with the same surface temperature [3, 4]. This paper
examines the effect of greenhouse gas concentrations on thermal radiation for the case of a
clear sky. It considers the five most important naturally occurring greenhouse gases: H2 O,
CO2 , O3 , N2 O and CH4 .
The spectra of greenhouse gases consists of hundreds of thousands of individual rovibra-
tional spectral lines whose strengths and transition frequencies can be downloaded from the
HITRAN database [5, 6]. The earliest global warming estimates approximated the multitude
of lines by various absorption bands [7, 8, 9]. The most accurate forcings are found by per-
forming line by line calculations [10, 11, 12, 13, 14, 15]. Here, we use line by line calculations
to estimate the effects of doubling CO2 , N2 O and CH4 concentrations from current levels.
The forcings are strongly affected by saturation of the absorption bands and spectral overlap
with other greenhouse gases. Recently, this was found to significantly affect methane forcing
[16].
This work downloaded over 1/3 million rovibrational lines from the most recent HITRAN
database to calculate the per-molecule forcings. The concentration of each greenhouse gas
was varied from the optically thin limit where there is negligible saturation or interference
of one type of greenhouse gas with others; to current levels where the per-molecule forcings
are suppressed by up to four orders of magnitude. The “instantaneous” forcings resulting
from doubling concentrations of CO2 , N2 O and CH4 were compared to those published in
the literature. The change to the temperature profile was estimated taking into account
radiative-conective equilibrium of the atmosphere as well as water feedback. Finally, the
intensities modelled at the top of the atmosphere were compared to satellite spectral mea-
surements at various latitudes.

2 Altitude Profiles of Temperature and Greenhouse

Gases
Radiation transfer in the cloud-free atmosphere of the Earth is controlled by the tempera-
ture T = T (z) at the altitude z and the number densities, N {i} = N {i} (z) of the ith type of
molecule. Representative midlatitude altitude profiles of temperature [17], and greenhouse
gas concentrations [18], are shown in Fig. 1. Altitude profiles directly measured by radioson-
des in ascending balloons are always more complicated than those of Fig. 1, which can be

2
Figure 1: Left. A standard atmospheric temperature profile[17], T = T (z). The Earth’s
{i}
mean surface temperature i T (0) = 288.7 K . Right. Standard concentrations[18], Csd for
greenhouse molecules versus altitude z.

thought of as appropriate average profiles for the year 2020.

We divided the atmosphere into 500 altitude segments, 100 segments for each layer:
the troposphere, lower stratosphere, mid stratosphere, upper stratosphere and mesosphere.
The segment midpoints are labeled by the integers i = 1, 2, 3, ..., 500. We characterize the
initial temperature profile of the atmosphere, with six breakpoints, with temperatures θα
and altitudes ζα , where α = 0, 1, ..., 5. Between the breakpoints, the atmosphere is assumed
to have constant temperature lapse rates Lα = −dT /dz = −(θα − θα−1 )/(ζα − ζα−1 ). For a
midlatitude standard atmosphere, the breakpoint temperatures and altitudes are
   
288.7 0
 217.2   11 
   
 217.2   20 
θ= 
 K, and ζ =  32  km.
   (1)
 229.2   
 271.2   47 
187.5 86
The lapse rates are L = [ 6.5, 0, −1, −2.8, 2.1 ] K km−1 . The temperature profile determined
by (1) is shown as the solid blue line in the left panel of Fig. 1.
In the troposphere, convective transport of sensible and latent heat of water vapor, es-
pecially near the equator, is as important as radiant heat transfer. Above the troposphere is
the stratosphere, which extends from the tropopause to the stratopause, at a typical altitude

3
 {i}
i Molecule N̂sd (cm−2 )
1 H2 O 4.67 × 1022
2 CO2 8.61 × 1021
3 O3 9.22 × 1018
4 N2 O 6.61 × 1018
5 CH4 3.76 × 1019

{i}
Table 1: Column densities, N̂sd , of the 5 most abundant greenhouse gases obtained using
the standard altitudinal profiles [18] of Fig. 1.

of zsp = 47 km, as shown in Fig. 1. The temperature in the stratosphere is nearly constant
at low altitudes, but increases at higher altitudes due to the heating of ozone molecules that
absorb ultraviolet sunlight. The stratosphere is much more stable to vertical displacements
of air parcels than the troposphere and negligible moist convection occurs because of the
very low water vapor concentration.
Above the stratosphere is the mesosphere, which extends from the stratopause to the
mesopause at an altitude of about zmp = 86 km. With increasing altitudes, radiative cool-
ing, mainly by CO2 , becomes increasingly more important compared to heating by solar
ultraviolet radiation. This causes the temperature to decrease with increasing altitude in
the mesosphere.
The vertical radiation flux changes rapidly in the troposphere and stratosphere com-
pared to the mesosphere where the atmospheric density is very low. Changes in flux above
the mesopause are negligible and the mesopause is therefore referred to as “the top of the
atmosphere” (TOA), with respect to radiation transfer.
{i}
The standard concentrations for the ith greenhouse gas, Csd , based on observations[18],
are shown as functions of altitude on the right of Fig. 1. The sea level concentrations are
7, 750 ppm of H2 O, 1.8 ppm of CH4 and 0.32 ppm of N2 O. The O3 concentration peaks at
7.8 ppm at an altitude of 35 km, and the CO2 concentration was 400 ppm at all altitudes.
Integrating the concentrations over an atmospheric column having a cross sectional area of
{i}
1 cm2 yields the column number density of the ith type of molecule N̂sd which are listed in
Table 1.

3 Greenhouse Gas Lines

A line by line calculation of radiative forcing utilizes various parameters that are now briefly
discussed.

3.1 Line Intensities

Fig. 2 illustrates the greenhouse gas lines considered in this work. The Bohr frequency νul
for a radiative transition from a lower level l of energy El to an upper level u of energy Eu
of the same molecule is denoted by

4
 Eul
νul = , where Eul = Eu − El . (2)
hc
where the energy of a resonant photon is Eul , h is Planck’s constant and c is the speed of
light.
The cross section, σ {i} = σ, for the ith type of greenhouse molecule is written as the sum
of partial cross sections σul , corresponding to each Bohr frequency νul ,
X
σ= σul . (3)
ul

The partial cross section, σul , is assumed to be the product of a lineshape function, Gul =
Gul (ν, τ ), and a line intensity, Sul = Sul (T ),

σul = Gul Sul . (4)

The lineshape functions, Gul , are normalized to have unit area,

Z ∞
Gul dν = 1 (5)
0
and have units of cm. The line intensity is

ηu Wu Γul Eul
Sul = ηu πre ful Wl 1 − e−νul /νT = . (6)
4π B̃ul
In (6), Sul has the units of cm, re = e2 /(me c2 ) is the classical electron radius, where e is the
elementary charge and me is the electron mass. The isotopologue fractions are ηu . For the
most abundant isotopologues of CO2 ,


 0.9843 for 16 O 12 C 16 O
0.0110 for 16 O 13 C 16 O

ηu = (7)

 0.0040 for 16 O 12 C 18 O
0.0007 for 16 O 12 C 17 O.


The last term of (6) contains the spectral Planck intensity evaluated at the frequency νul ,

B̃ul = B̃(νul , T ). (8)

The Planck intensity is given by

2hc2 ν 3
B̃(ν, T ) = (9)
eνc h/(kB T ) − 1
The radiation frequency, ν = 1/λ is the inverse of the wavelength λ and has units of cm−1 .
The probability Wn (with n = u or n = l) to find a molecule in the rovibrational level n is

gn e−En /kB T
Wn = . (10)
Q

5
 {r}
Figure 2: Reference line intensities, Sul of (15) for H2 O, CO2 , O3 , N2 O and CH4 from
the HITRAN data base [6]. The horizontal coordinate of each point represents the Bohr
frequency νul of a transition from an upper level u to a lower level l. The vertical coordinate
of the point is the line intensity. For greater clarity we have plotted only 1/10, chosen
at random, of the extremely large number of O3 line intensities. The numbers of lines
(in parenthesis) used for this work were: H2 O (31,112), CO2 (20,569), O3 (210,295), N2 O
(43,152) and CH4 (43,696). The smooth line is the Planck spectral intensity, B̃ of (9) in
units of mW cm m−2 sr−1 for the HITRAN reference temperature, T {r} = 296 K.

Here gn is the statistical weight of the level n, the number of independent quantum states
with the same energy En . For molecules in the level n, the statistical weight can be taken
to be
gn = (2jn + 1)kn , (11)
where jn is the rotational angular momentum quantum number, and kn is the nuclear de-
generacy factor, that depends on whether the spins of the nuclei are identical or not. The
partition function, Q = Q(T ), of the molecule is
X
Q= gn e−En /kB T . (12)
n

The oscillator strength, ful , of (6) is related to the matrix elements of the electric dipole
moment M of the molecule, between the upper energy basis state |ju mu i with azimuthal

6
quantum number mu and the lower energy basis state |jl ml i, by
4πνul X
ful = hu mu |M|l ml i · hl ml |M|u mu i. (13)
3gl c re ~ m m
u l

The quantum numbers mu label the various degenerate substates of the upper level u and
the ml label the substates of the lower level l. If the levels are characterized by rotational
quantum numbers ju and jl , the quantum numbers mu and ml can be thought of as the
corresponding azimuthal quantum numbers, for example, mu = ju , ju − 1, . . . , −ju .
The rate of spontaneous emission of photons when the molecule makes transitions from
the upper level u to the lower level l is Γul , the same as the Einstein A coefficient, Γul = Aul
[19, 20]. The spontaneous emission rate is related to the oscillator strength by

8π 2 c re νul
2
ful gl
Γul = . (14)
gu
From inspection of (6) we see that the line intensity Sul = Sul (T ) at some arbitrary
{r}
temperature T is related to the intensity, Sul = Sul (T {r} ) at a reference temperature T {r}
where the partition function of (12) is related to Q{r} = Q(T {r} ) by
{r}
 −El /kB T  
1 − e−νul /νT

{r} Q e
Sul = Sul . (15)
Q e−El /kB T {r} 1 − e−νul /νT {r}
This work considered all lines in the HITRAN database of the five gases under consideration
in having intensities greater than 10−25 cm. For H2 O, lines having intensities greater than
10−27 cm were included since water vapor has an order of magnitude greater density than
any other greenhouse gas near the Earth’s surface.

3.2 Lineshapes
It is convenient to write the lineshape function as the product of a “core” profile Cul = Cul (ν)
and a wing-suppression factor, χul = χul (ν),

Gul = Cul χul . (16)

Core functions The core function is normally taken to be a Doppler broadened Lorentzian,
often called a Voigt profile[21]. If we average over a Maxwellian distribution of velocities v
for molecules of mass m at the temperature T we find that the core function has the form
Z ∞ 2
e−mv /2kB T dv
r
µul m
Cul =
π 2πkB T −∞ µ2ul + (ν − νul [1 + v/c])2
(17)

The half width at half maximum of a purely Doppler broadened line is

r
2kB T ln 2
∆νul = νul 2
≈ 0.0005 cm−1 (18)
mc

7
Here, the representative Doppler half width is for a resonance frequency νul of the 667 cm−1
band of a CO2 molecule near the cold mesopause.
The half width at half maximum, µul , of the Lorentzian function in (17) is almost entirely
due to collisions. The contribution to µul from spontaneous radiative decay is negligible at
altitudes below the mesopause. For the bending mode of CO2 , representative values[5] of µul
at atmospheric pressure p are
p
µul ≈ 0.07 cm−1 . (19)
p0
Here p0 = 1 bar, the approximate atmospheric pressure at mean sea level. The pressure
broadening coefficients depend somewhat on temperature and on the particular resonance,
ul, involved.
For (17), a small, temperature dependent pressure shift[5] must be added to the free
{0}
molecule Bohr frequency of (2), which we denote by νul , to define the resonance frequency
(0)
νul = νul + δul p/p0 . (20)

The magnitude of the pressure shift coefficient δul is of order

|δul | ≈ 0.001 cm−1 , (21)

comparable to the Doppler half width (18). The small pressure shifts have negligible influence
on radiative forcing calculations, but the pressure broadening coefficients of (19) have a large
effect. Collisions significantly broaden absorption lines in the troposphere and stratosphere.

Wing-suppression functions Lorentz profiles give far too much absorption for large
detunings |ν − νul |, so it is necessary to include wing suppression factors χ in the expression
(16) for the lineshape function [22, 23]. Lorentzian lineshapes result from assuming an
infinitely short collision duration, but in fact collisions take a few ps for completion. The
collisional interactions that lead to wing suppression are not known well enough for reliable
theoretical calculations, so we use the empirical wing suppression factor

χul (ν) = sech2 ([ν − νul ]/$). (22)

Measurements on bands of CO2 , for example by Edwards and Strow [24], suggest that the
far wings decrease approximately exponentially with detuning, |ν − νul |, as does the wing
suppression function (22). We used the width parameter $ = 2 cm−1 , corresponding to a
collision duration of a few picoseconds.

4 Radiation
In cloud-free air where scattering is negligible, radiation transport is governed by the Schwarzschild
equation [25],

∂ I˜
cos θ = −(I˜ − B̃) (23)
∂τ

8
where I˜ = I(ν,
˜ z, θ) is the spectral intensity of a pencil of radiation of frequency between ν
and ν +dν at altitude z. The pencil makes an angle θ to the vertical. In thermal equilibrium,
the spectral intensity I˜ equals the Planck intensity given by (9). The optical depth is defined
by
Z z
τ (z, ν) = dz 0 κ(z 0 , ν), (24)
0
where the net attenuation coefficient due to molecules absorbing and remitting light of
frequency ν at altitude z is given by
X
κ(z, ν) = N {i} (z)σ {i} (z, ν). (25)
i

Here N {i} (z) is the density of greenhouse gas molecule of type i and σ {i} = σ {i} (z, ν) is
its absorption cross section for radiation of frequency ν at the altitude z given by (3). The
cross section can depend strongly on altitude because temperature and pressure are functions
of altitude. Temperature controls the distribution of the molecules between translational,
rotational and vibrational states. Pressure, together with temperature, determines the width
of the molecular resonance lines.
The optical depth from the surface to the top of the radiative atmosphere, the altitude
zmp of the mesopause, is
Z zmp
τ∞ = τmp = dz 0 κ(z 0 , ν). (26)
0
As indicated by the notation (26), we have assumed that the optical depth τmp at the
mesopause altitude zmp differs negligibly from the optical depth τ∞ at infinite altitude since
there is so little opacity of the atmosphere above the mesopause.
The Schwarzschild equation (23) can be solved to find the intensity [26, 27]

Z τ
0
For ς > 0 : ˜ ς) = +ς
I(τ, ˜ ς)
dτ 0 e−ς(τ −τ ) B̃(τ 0 ) + e−ςτ I(0, (27)
0
Z τ∞
0
For ς < 0 : ˜ ς) = −ς
I(τ, dτ 0 e−ς(τ −τ ) B̃(τ 0 ) (28)
τ

where ς = sec θ. For simplicity, we assume the surface intensity is the product of Bs = B(Ts ),
the Planck intensity (9) for a temperature Ts , and an angle independent emissivity s = s (ν),

˜ ς) = s B̃s .
I(0, (29)
Over most of the Earth’s surface the thermal infrared emissivity s , is observed to be in the
interval [0.9 < s < 1] [28]. Negligible error is introduced by setting s = 1 in spectral
regions of high atmospheric opacity, with τ∞ >> 1. With only radiative heat transfer from
the surface and with negligible conduction or convection, there can be a discontinuous jump
between the surface temperature Ts and the temperature T0 of the air just above. For the
rest of this paper we assume that any temperature jump is negligibly small, and Ts = T0
and B̃s = B̃0 .

9
 For discussions of greenhouse gas forcing, we will be particularly interested in the upward
flux
Z
Z̃ = ˜
dΩ cos θ I. (30)
4π
Using (27) and (28) in (30), we find
Z τ Z τ∞
Z̃ 0 0 0
= dτ E2 (τ − τ )B̃(τ ) − dτ 0 E2 (τ 0 − τ )B̃(τ 0 ) + s B̃s E3 (τ )
2π 0 τ
τ∞
∂ B̃(τ 0 )
Z
= − dτ 0 E3 (|τ − τ 0 |) + B̃(τ∞ )E3 (τ∞ − τ ) + [s B̃s − B̃(0)]E3 (τ ). (31)
0 ∂τ 0
In this work, we have assumed that s B̃s = B̃(0). Equation (31) is the fundamental expres-
sion for the net upward flux in an atmosphere with negligible scattering and has been known
for a long time. For example, the first line can be found in the NASA reports: Equation (2a)
of Yoshikawa [29], or Equation (3.19) of Buglia [26]. Equation (31) contains exponential-
integral functions, En (τ ), that account for slant paths of radiation between different altitudes.
They are defined for integers n = 1, 2, 3, . . . by
Z ∞
En (τ ) = dς ς −n e−ςτ , (32)
1
as discussed in Appendix 1 of Chandrasekhar [7], or Section 5.1.4 of Abramowitz and Stegun
[30].
For frequencies where the atmosphere is not too optically thick, with τ∞ ≤ 10, we evaluate
Z̃ from the first line of (31). Fast Fourier transforms are used to calculate a discretized
version of the convolution with B̃. For frequencies where the atmosphere is thicker, we
use the expression on the first line of (31) to calculate Z̃ for 10 e-foldings of optical depth
down from the top of the atmosphere, τ∞ − τ < 10. We use a fast and accurate analytic
approximation of the second line to evaluate Z̃ when τ∞ − τ ≥ 10. For example, in the lower
troposphere,

Z̃ dB̃(τ ) h 4 i
=− − 2E4 (τ ) . (33)
2π dτ 3
The spectral forcing, F̃ , is defined as the difference between the spectral flux π B̃s through
a transparent atmosphere from a black surface with temperature Ts , and the spectral flux Z̃
for an atmosphere with greenhouse gases,

F̃ = π B̃s − Z̃. (34)

The frequency integrals of the flux (30) and the forcing (34) are
Z ∞
Z = dν Z̃, (35)
Z0 ∞
F = dν F̃ = σSB T04 − Z, (36)
0
(37)

10
where σSB is the Stefan Boltzmann constant.

5 Intensity and Flux

In this section we discuss model atmospheres with greenhouse gas concentrations comparable
to those of the year 2020. The spectral flux has a complicated dependence on frequency ν, the
altitude profiles of the temperature T and greenhouse gas concentrations C {i} , and latitude.
For the standard atmosphere, the optical depths of (26) can be extremely large i.e.
τ∞ >> 1 at so called “blanket” frequencies ν near the centers of the absorption lines. For
blanket frequencies a photon emitted near the surface is unlikely to escape the Earth’s atmo-
sphere because it has a high probability of being reabsorbed by a greenhouse gas molecule.
The extreme opposite of a blanket frequency is a “window” frequency, where there is little
absorption, and τ∞ ≈ 0. For window frequencies most radiation reaching space comes from
the surface, with minor contributions from greenhouse gases.
From the first line of (31), we see that for frequencies where the atmosphere is opti-
Rcally
τ∞
thick, with τ∞ >> 1, the flux at the top of the atmosphere is very nearly Z̃(τ∞ ) =
0
0
dτ E2 (τ∞ − τ 0 )B̃(τ 0 ). For an isothermal atmosphere, with constant brightness B̃, half
of the flux at the top of the atmosphere will come from altitudes above an emission optical
depth τe , defined by
Z τ∞
1 τ∞ 0
Z
0 0
dτ E2 (τ∞ − τ ) = dτ E2 (τ∞ − τ 0 ). (38)
τe 2 0
Recalling that E2 (τ ) = −dE3 (τ )/dτ and that E3 (0) = 1/2, we see that (38) implies that the
value of the emission optical depth τe is given by

E3 (τ∞ − τe ) = 1/4, or τ∞ − τe = 0.41904. (39)

Fig. 3 shows a plot of upwards flux as a function of altitude at four different frequencies.
Fig. 3(a) shows a moderate blanket frequency ν = 500.5 cm−1 , where the optical depth,
τ∞ = 2.82. The emission height ze = 2.80 km is in the lower troposphere. From inspection
of Fig. 2, we see that transitions of the pure rotation spectrum of H2 O dominate the
atmospheric opacity at the frequency, ν = 500.5 cm−1 .
Fig. 3(b) shows an extreme blanket frequency ν = 667.4 cm−1 coinciding with a peak
in the CO2 absorption cross section where the optical depth is τ∞ = 51, 688, and where the
emission height is ze = 84.8 km, just below the mesopause.
Fig. 3(c) shows an extreme window frequency, ν = 971 cm−1 , where the optical depth is
only τ∞ = 0.029. At this frequency, and in the absence of clouds, surface radiation reaches
space with negligible attenuation by greenhouse gases. The band centered on the wavelength
λ = 1/(971 cm−1 ) = 10.3 µm, is therefore called the “clean infrared window.”
Fig. 3(d) shows a blanket frequency, ν = 1016.2 cm−1 , with a moderate optical depth
τ∞ = 7.54 in the O3 band. Not surprisingly, the emission height, ze = 33.0 km, is in the
upper stratosphere, where Fig. 1 shows that the O3 concentration is maximum.
High resolution spectrometers on satellites seldom provide measurements of intensity I˜
with resolutions less than 1 cm−1 . For comparison of modeled spectral intensity or flux, it
is useful to plot filtered spectral quantities.

11
Figure 3: Spectral fluxes Z̃ for representative frequencies, ν indicated: (a) τ∞ = 2.82, ze =
2.80 km, B̃s = 134.2 i.u.; (b) τ∞ = 51688, ze = 84.8 km, B̃s = 131.8 i.u.; (c) τ∞ = 0.029, ze =
0 km, B̃s = 86.4 i.u.; (d) τ∞ = 7.54, ze = 33.0 km, B̃s = 79.3 i.u.. Here the spectral intensity
unit is 1 i.u. = 1 mW m−2 cm sr−1 .

Z ∞
hX̃i(z, ν) = dν 0 J(ν, ν 0 )X̃(z, ν 0 ). (40)
0

The filter function J(ν, ν 0 ) smooths out sharp changes with frequency. It is normalized so
that
Z ∞
dνJ(ν, ν 0 ) = 1. (41)
−∞

From (40) and (41) we see that the unfiltered spectral flux Z̃ and filtered spectral flux hZ̃i
have the same frequency integral
Z ∞ Z ∞
Z= dν Z̃ = dν hZ̃i, (42)
0 0
and represent the same total flux Z. We found it convenient to use a Gaussian filter function
with a width parameter ∆ν,
0 2 2
0 e−(ν−ν ) /2∆ν
J(ν, ν ) = √ . (43)
2π∆ν

12
Figure 4: Effects of changing concentrations of carbon dioxide, CO2 on the filtered spectral
flux hZ̃ {i} (ν, zmp , f )i of (44) at the mesopause altitude, zmp = 86 km. The width of the filter
(43) was ∆ν = 3 cm−1 . The smooth blue line is the spectral flux, Z̃ = π B̃(ν, T0 ) from a
surface at the temperature T0 = 288.7 K for a transparent atmosphere with no greenhouse
gases. The green line is hZ̃ {i} (ν, zmp , 0)i with the CO2 removed but with all the other
greenhouse gases at their standard concentrations. The black line is hZ̃ {i} (ν, zmp , 1)i with all
greenhouse gases at their standard concentrations. The red line is hZ̃ {i} (ν, zmp , 2)i for twice
the standard concentration of CO2 but with all the other greenhouse gases at their standard
concentrations. Doubling the standard concentration of CO2 (from 400 to 800 ppm) would
cause a forcing increase (the area between the black and red lines) of ∆F {i} = 3.0 W m−2 ,
as shown in Table 2.

The effects on radiative transfer of changing the column density of the ith greenhouse
{i}
gas to some multiple f of the standard value, N̂sd , can be displayed with filtered spectral
fluxes
{1} {i−1} {i} {i+1} {n}
hZ̃ {i} (ν, z, f )i = hZ̃(ν, z, N̂sd , . . . , N̂sd , f N̂sd , N̂sd , . . . , N̂sd )i. (44)
Figs. 4 and 5 show how varying the concentrations of CO2 and CH4 affect the filtered
spectral fluxes at the mesopause altitude, zmp = 86 km. Expanded views of the differences
between the flux for standard and doubled concentrations of greenhouse gases are shown in
Fig. 6, where we display

h∆F̃ {i} (zmp , 2)i = hZ̃ {i} (ν, zmp , 1)i − hZ̃ {i} (ν, zmp , 2)i. (45)

13
Figure 5: Effects of changing concentrations of methane, CH4 , on the filtered spectral flux
hZ̃ {i} (ν, zmp , f )i of (44) at the mesopause altitude, zmp = 86 km. The blue and black lines
have the same meanings as for Fig. 4. The green line is hZ̃ {i} (ν, zmp , 0)i with the CH4
removed but with all the other greenhouse gases at their standard concentrations. The red
line is hZ̃ {i} (ν, zmp , 2)i with twice the standard concentration of CH4 but with all the other
greenhouse gases at their standard concentrations. Doubling the standard concentration of
CH4 would cause a forcing increase (the area between the black and red lines) of ∆F {i} = 0.7
W m−2 , as shown in Table 2.

Integrating spectral fluxes, Z̃, like those of Fig. 4, over all frequencies in accordance
with (35) gives Z, the frequency integrated flux shown in the right panel of Fig. 7. The
calculations used the temperature profile of Fig. 1, which is shown in the left panel of Fig. 7.
A doubling of CO2 concentration results in a 3 W/m2 decrease in the top of the atmosphere
flux. This positive forcing changes the temperature profile as is discussed in Section 7.
Thermal radiative fluxes depend on latitude. They are larger near the equator where
the surface is relatively warm than near the poles, where the surface is colder and where
wintertime temperature inversions often form in the lower troposphere. For example, Fig. 8
is the analog of Fig. 7 for Antarctica. For Fig. 8 we used a five segment temperature profile
with altitude breakpoints at ζ = [0, 2.5, 8, 25, 47, 86] km. The low tropopause at 8 km and the
strong, wintertime temperature inversion, peaking at 2.5 km, are both characteristic of the
nighttime poles. The lapse rates between the break points were L = [−12.5, 2.3, 0, −1.5, 2.1]
K km−1 . The surface temperature in Antarctica was taken to be T0 = 190 K and the surface
pressure was set to be p0 = 677 hPa, because of the high elevation of the ice surface, about
2.7 km above mean sea level. Doubling CO2 causes the negative forcing in the Antarctic

14
Figure 6: Spectral forcing increments (45) for doubled concentrations of CO2 , N2 O and CH4 .
These are the magnified differences between the black and red curves of Figs. 4 and 5. For
most frequencies h∆F̃ {i} (zmp , 2)i is positive. This is because doubling the concentrations
of greenhouse gases shifts the emission heights ze of (38) to higher, colder altitudes in the
troposphere. An exception is the band of frequencies near the center of the exceptionally
strong bending-mode band of CO2 at 667 cm−1 . Here doubling CO2 moves the emission
heights to higher, warmer altitudes of the stratosphere, where molecules can more efficiently
radiate heat to space. Also shown are forcing increments for halved concentrations of H2 O
and O3 . Halving ensures that the relative humidity does not exceed 100%, and reduces the
clutter of the graph.

opposite that shown in Fig. 1. This is due to the temperature inversion which means
an increase in CO2 causes more infrared radiation to escape to space, creating a negative
greenhouse effect [31].

6 Concentration Dependence of Forcing

The frequency integrated forcing, F , of (36) depends on the altitude z and on the column
densities of the five greenhouse gases given in Table 1.

F = F (z, N̂ {1} , . . . , N̂ {5} ). (46)

We assume the temperature T and densities N {i} have the same altitude profiles as in the
midlatitude example of Fig. 1. An important special case of (46) is the forcing, Fsd , when

15
Figure 7: Left. Midlatitude standard temperature profile. Right. Altitude dependence of
frequency integrated flux Z of (35). The flux for three concentrations of CO2 are shown,
{i}
the standard concentration, Csd = 400 ppm of Fig. 1, twice and half that value. The
other greenhouse gases have the standard concentrations of Fig. 1. The vertical dashed line
is the flux σSB T04 = 394 W m−2 for a transparent atmosphere with a surface temperature
T0 = 288.7 K. The forcings Fs that follow from (36) at 0 km, 11 km and 86 km are 252, 137
and 117 W m−2 respectively.

{i}
each greenhouse gas i is present at its standard column density N̂sd of Table 1,
{1} {5}
Fsd (z) = F (z, N̂sd , . . . , N̂sd ). (47)
{i}
A second special case of (46) is the hypothetical, per molecule standard forcing, Fsd , when
the atmosphere contains only molecules of type i at their standard column density, N̂ {i} =
{i}
N̂sd , and the concentrations of the other greenhouse vanish, N̂ {j} = 0 if j 6= i,
{i} {i}
Fsd (z) = F (z, 0, . . . , 0, N̂sd , 0, . . . , 0). (48)

We define the forcing power per added molecule as

∂F
P {i} (z, N̂ {1} , . . . , N̂ {n} ) = . (49)
∂ N̂ {i}
The densities of greenhouse gases j with j 6= i are held constant in the partial derivative of
(49). If the units of F are taken to be W m−2 and the units of N̂ {i} are taken to be molecules
m−2 , then the units of P {i} will be W molecule−1 .

16
Figure 8: Quantities analogous to those of Fig. 7 at the South Pole. There is a strong
temperature inversion at an altitude of 2.5 km above the ice surface. The flux for three
{i}
concentrations of CO2 are shown, the standard concentration, Csd = 400 ppm of Fig. 1,
twice and half that value. The other greenhouse gases have the standard concentrations of
Fig. 1. The relatively warm greenhouse-gas molecules in the atmosphere above the cold
surface cause the Earth to radiate more heat to space from the poles than it could without
greenhouse gases [31].

We define a finite forcing increment for the ith type of greenhouse molecule as
{1} {i−1} {i} {i+1} {n}
∆F {i} (z, f ) = F (z, N̂sd , . . . , N̂sd , f N̂sd , N̂sd , . . . , N̂sd ) − Fsd . (50)

Differentiating (50) with respect to f we find

∂∆F {i} {i} {i}

(z, f ) = N̂sd Psd (z, f ), (51)
∂f
{i}
where Psd (z, f ) is the forcing power per additional molecule of type i when these molecules
{i}
have the column density N̂ {i} = f N̂sd and all other types of greenhouse molecules have their
standard column densities.
The forcing increments (50) for the five greenhouse gases considered in this paper are
shown as a function of f in Fig. 9. Forcing increments are also tabulated at representative
altitudes z and multiplicative factors f in Table 2. At both the top of the atmosphere
and at the tropopause, we see that the forcing increment (50) is largest for abundant water
molecules, H2 O, and is relatively small for the much more dilute greenhouse gases CH4 and

17
Figure 9: Dependence of partial forcing increments ∆F {i} of (50) on greenhouse gas multi-
{i}
plicative factor, f = N {i} /Nsd . At the standard column densities, with f = 1, the incre-
mental forcings are well into the saturation regime, with d∆F {i} (1)/df < d∆F {i} (0)/df for
all 5 gases. For the most abundant greenhouse gases, H2 O and CO2 , the saturation effects
are extreme, with per-molecule forcing powers suppressed by four orders of magnitude at
standard concentrations (f = 1) with respect to the low-concentration, optically thin limit
(f = 0). For CO2 , N2 O, and CH4 , the areas bounded by the green and black curves of Figs.
4 and 5 give the values, −∆F {i} for f = 0, and the areas bounded by the black and red
curves give ∆F {i} for f = 2, as discussed for Fig. 6. See the text and Table 2 for more
details.

N2 O. The incremental forcings are all in the saturation regime, with ∂∆F {i} /∂f diminishing
with increasing f .
In Table 2, the forcing decrements from removing H2 O, CO2 , O3 , N2 O and CH4 , −62.2,
−30.2, −8.1, −2.2 and −2.1 W m−2 , are reasonably close to those calculated by Zhong
and Haigh [32]. In their Table 1 they cite forcing decrements at the top of the atmosphere
of −70.6, −25.5, −7.0, −1.8 and −1.7 W m−2 . Zhong and Haigh seem to have taken the
concentrations of N2 O and CH4 to be independent of altitude. The altitude dependence of
Fig. 1 were used in our calculations.
Note from Table 2 that doubling or halving the column density of CO2 changes the
forcing F by almost the same amount, either at the tropopause or at the mesopause. This
dependence of forcing increments on the logarithm of the CO2 column density was first
pointed out by Arrhenius [33]. Wilson and Gea-Banacloche [34] explain how the approximate
dependence of the CO2 absorption cross section on frequency ν, σ {i} = σe e−λe |ν−νe | , leads to

18
 {i}
Fsd (z) ∆F {i} (z, 0) ∆F {i} (z, 1/2) ∆F {i} (z, 2)
i\z ztp zmp ztp zmp ztp zmp ztp zmp
H2 O 81.6 71.6 -72.6 -62.2 -10.4 -7.8 11.2 8.1
CO2 52.4 38.9 -44.6 -30.2 -5.3 -3.0 5.5 3.0
O3 6.1 10.5 -4.7 -8.1 -1.8 -2.2 2.5 2.5
N2 O 4.4 4.7 -2.2 -2.2 -0.8 -0.8 1.2 1.1
CH4 4.2 4.4 -2.1 -2.1 -0.6 -0.6 0.8 0.7
P
i 148.7 130.1 -126.2 -104.8
Fsd (z) 137 117 137 117

{i}
Table 2: Partial forcings Fsd (z) of (48) and partial forcing increments ∆F {i} (z, f ) of (50),
all in units of W m−2 , at the altitudes ztp = 11 km of the tropopause and zmp = 86 km of
the mesopause. The last row contains the forcings Fsd (z) of (47), shown in Fig. 7, when
{i}
all greenhouse molecules are present simultaneously at their standard column densities N̂sd .
P {i}
Because of the overlapping absorption bands, i Fsd (z) > Fsd (z), and − i ∆F {i} (z, 0) <
P
Fsd (z).

the logarithmic forcing law. Here σe is the maximum cross section at the center frequency,
νe = 667 cm−1 , of the bending mode band.
The forcing increments in Table 3 are comparable to those calculated by others. For
example, in column 3 we give the increments ∆F {i} (z, f ) calculated by Collins et al [13],
as estimated from their Tables 2 and 8. These are the results of averaging five separate
line by line calculations. In addition to line intensities, three of the calculations used a
continuum CO2 opacity, and all five used a continuum H2 O opacity. The physical origin
of these continua is unclear. They are added to make the calculations agree better with
observations [10, 13]. The forcings calculated in this paper, summarized in column 4, used
only lines for the HITRAN data base and no continua. Our values are fairly close to those of
Collins et al.[13], with the largest discrepancy for H2 O. The mesopause spectral intensities,
calculated with only HITRAN lines and with no continuum contributions, are in excellent
agreement with satellite measurements over the Sahara Desert, the Mediterranean Sea and
Antarctica, as discussed in Section 8.
The three mesopause flux increments ∆F {i} in the fourth column of Table 3 for doubled
concentrations of CO2 , N2 O and CH4 sum to 4.8 W m−2 . The calculated flux increment
from simultaneously doubling CO2 , N2 O and CH4 is the slightly smaller value, ∆F = 4.7 W
m−2 . Similarly, the four mesopause flux increments ∆F {i} in the fourth column of Table 3
for doubled concentrations of CO2 , N2 O and CH4 as well as a factor of f = 1.06 increase of
H2 O concentration sum to 5.5 W m−2 . The calculated flux increment from simultaneously
doubling CO2 , N2 O and CH4 , and increasing the H2 O concentrations by a factor of f = 1.06,
is the slightly smaller value 5.3 W m−2 . The “whole” is less than the sum of the parts, because
of the interference of greenhouse gases that absorb the same infrared frequencies.
Table 4 summarizes the forcing powers (49) per additional molecule in units of 10−22 W
at the tropopause altitude, ztp = 11 km and at the mesopause altitude, zmp = 86 km. The

19
 ∆F {i} (z, f ) in W m−2
Ref. [13] This Work
i f ztp zmp ztp zmp
H2 O 1.06 1.4 1.1 0.9 0.7
CO2 2 5.5 2.8 5.5 3.0
O3 1.1 0.3 0.3
N2 O 2 1.3 1.2 1.2 1.1
CH4 2 0.6 0.6 0.8 0.7

Table 3: Comparison of the forcing increments ∆F {i} (z, f ) of Collins et al.[13] in column 3,
and the results of Table 2 and (50) in column 4, at the altitude ztp = 11 km of the tropopause
and zmp = 86 km of the mesopause. For H2 O, the relative increase, f = 1.06, of the column
density is approximately that caused by a 1 K increase of the surface temperature.

surface temperature was T0 = 288.7 K, and the altitude profiles of temperature and number
density were those of Fig. 1. The first column lists the molecules we considered. The
{i}
numbers in the second column are forcing powers, Pot (z), of (54) in the optically thin limit.
{i}
The numbers of the third column are forcing powers Psd (z, 0) from (51) for an atmosphere
that previously had no molecules of type i (so N̂ {i} = 0) but all other greenhouse molecules
{j}
had standard concentrations, N̂ {j} = N̂sd if j 6= i. The forcings of the third column are less
than those of the second because of interference between absorption by different greenhouse
{i}
gases. The numbers in the fourth column are the forcing powers Psd (z, 1) from (51) when a
single molecule of type i is added to an atmosphere that previously had standard densities for
{j}
all greenhouse gases, N̂ {j} = N̂sd . Saturation of the absorption suppresses the per-molecule
forcing by about four orders of magnitude for the abundant greenhouse gases H2 O and CO2 .
Saturation causes less drastic suppression of per-molecule forcings for the less abundant O3 ,
N2 O and CH4 .
We now consider the optically thin limit, where the concentrations of greenhouse gases
are sufficiently low that the optical depths τ of (24) will be small, τ  1, for all frequencies
ν and at all altitudes z. The frequency integral of the spectral forcing (35) at altitude z can
then be written as

{i}
X
Fot (z) = N̂ {i} Pot (z) (52)
i

where the forcing power per greenhouse molecule of type i is

Z z {i}0 Z z {i}0
{i} 1 0N
 {i} 0  1 0N
Pot (z) = dz Π (T , T0 ) − Π{i} (T 0 , T 0 ) + dz Π{i} (T 0 , T 0 ). (53)
2 0 N̂ {i} 2 0 N̂ {i}
0
Here N {i} = N {i} (z 0 ), T 0 = T (z 0 ) and T0 = T (0). The mean power absorbed by a greenhouse
gas molecule of temperature T from thermal equilibrium radiation of temperature T 0 is

20
 {i} {i} {i}
Pot (z) Psd (z, 0) Psd (z, 1)
i\z ztp zmp ztp zmp ztp zmp
H2 O 1.49 1.49 1.16 1.19 3.3 × 10−4 2.5 × 10−4
CO2 2.73 3.45 2.24 2.53 9.0 × 10−4 4.9 × 10−4
O3 2.00 5.69 1.68 4.57 3.3 × 10−1 3.8 × 10−1
N2 O 1.68 2.24 0.73 0.91 2.1 × 10−1 2.0 × 10−1
CH4 0.51 0.71 0.21 0.27 2.8 × 10−2 2.6 × 10−2

Table 4: Forcing powers (49) per additional molecule in units of 10−22 W at the altitude
ztp = 11 km of the tropopause and zmp = 86 km of the mesopause. The surface temperature
was T0 = 288.7 K, and the altitude profiles of temperature and number density were those of
{i} {i}
Fig. 1. Pot (z) of (54) is for the optically-thin limit. Psd (z, 0) from (51) is for an atmosphere
that previously had no molecules of type i (so N̂ {i} = 0) but all other greenhouse molecules
{i}
had standard concentrations. Psd (z, 1) from (51) is for a single molecule of type i added to
an atmosphere that previously had standard densities for all greenhouse gases.

{i}
X
Π{i} (T, T 0 ) = 4π Sul (T )B̃(νul , T 0 ). (54)
ul

For the special case of T 0 = T we can substitute (6) into (54) to find
{i} {i}
X
Π{i} (T, T ) = Wu{i} (T )Γul Eul . (55)
ul

Since we are considering a single isotopologue, we have set ηu = 1 in (6). The three factors
{i}
in the summed terms of (55) are the probability Wu (T ) to find the molecule in the upper
{i}
state u, the radiative decay rate Γul from the upper level u to the lower level l and the
{i}
mean energy Eul of the emitted photon. This is obviously the total power radiated by a
molecule of temperature T . For a molecule of temperature T in thermal equilibrium with
radiation of the same temperature, the radiative power absorbed by the molecule is equal to
the spontaneous radiative power it emits. The forcing powers per molecule are summarized
graphically in Fig. 10.

7 Temperature and Forcing

The forcings due to instantaneous changes of greenhouse gas concentrations can be calcu-
lated quite accurately. Temperature changes induced by the forcings are less clearly defined
because various feedbacks change the temperature profile of the atmosphere. After doubling
CO2 concentrations, a new, steady state will eventually be established by these feedback
processes.
As shown in Fig. 7, we have computed the upwards flux Z = Z(Cg , Cw , T ) where Cg and
Cw are the initial concentrations of some greenhouse gas and water vapor respectively, and T
is the temperature. The flux depends on altitude z because Cg , Cw and T are each functions

21
Figure 10: A graphical display of the per-molecule forcing powers of Table 4. At standard
{i}
column densities the (red) powers, Psd (z, 1), for H2 O and CO2 are suppressed by four
orders of magnitude from their values in the optically thin limit (blue) where the powers
{i}
are Pot (z). This is due to strong saturation of the absorption bands. Saturation effects
(difference between the blue and red lines) are much less for the minor gases, O3 , N2 O and
{i}
CH4 . The green lines are the powers per molecule, Psd (z, 0), of the ith greenhouse gas in
its low-concentration limit, but when the forcing power is suppressed by other gases at their
standard densities. Interference effects (difference between the blue and green lines) are more
pronounced for N2 O and CH4 than for H2 O and CO2 . Fig. 2 shows the strongest bands of
O3 overlap little with those of other greenhouse molecules, minimizing interference effects.

of altitude. In the right panel of Fig. 7 we showed that changing the concentration Cg of
the greenhouse gas CO2 to twice its value, Cg0 = 2Cg , while holding all other atmospheric
properties the same, changed the flux from its initial value Z(Cg , Cw , T ) to a slightly smaller
value, Z(Cg0 , Cw , T ). In this hypothetical “instantaneous” process there is no change in the
atmospheric temperature T = T (z). The concentrations of all other greenhouse gases, most
notably, the concentration Cw = Cw (z) of water vapor, also remain the same. The difference
between the flux before and after addition of the greenhouse gas is called the instantaneous
forcing increment, and can be written as
∆F = −∆Z = Z(Cg , Cw , T ) − Z(Cg0 , Cw , T ). (56)
The flux increment ∆Z for doubling CO2 concentrations is too small to be seen clearly in
Fig. 7 so we have plotted an expanded version in the right panel of Fig. 11. The magnitude
of the flux increment is somewhat greater in the lower atmosphere than at higher altitudes.

22
Figure 11: Temperature and Flux Adjustments. Left. An initial temperature profile T
(continuous blue line), which is characteristic of mid latitudes, and the adjusted profile
T 0 = T + ∆T (dotted red line), from (69). Right. The continuous blue line is the altitude
profile of the instantaneous flux decrease ∆Z of (56) caused by increasing CO2 concentrations
from 400 ppm to 800 ppm, with no change in the profiles T of temperature or Cw of water-
vapor concentration. The dotted red line shows the residual flux change δZ of (59) for the
adjusted temperature profile T 0 of the left panel, together with the adjusted water-vapor
concentration Cw0 to keep relative humidity constant at all altitudes. These adjustments
restore the flux as nearly as possible to its original value for altitudes above the tropopause,
the criterion for maintaining radiative-convective equilibrium.

In steady state the atmosphere is commonly assumed to be in radiative-convective equi-

librium, described in 1967 by Manabe and Wetherald[35]. If the concentration Cg of a green-
house gas other than water vapor instantaneously changes to a new value Cg0 = Cg + ∆Cg ,
the atmosphere will no longer be in equilibrium. To restore radiative-convective equilibrium
the temperature profile will change to T 0 = T + ∆T and the water-vapor concentration will
change to Cw0 = Cw + ∆Cw .
The first criterion for radiative-convective equilibrium is no change in thermal flux at the
top of radiative atmosphere, which we take to be the mesopause altitude zmp = ζ5 ,

Zmp (Cg0 , Cw0 , T 0 ) = Zmp (Cg , Cw , T ), for z = ζ5 . (57)

This criterion assumes that net solar heating remains the same and is balanced by thermal
radiation to space.
Above the tropopause, vertical convection is negligible. Thermal radiation carries off heat
resulting from the absorption of solar ultraviolet radiation by ozone, and also any heating
by horizontal convection. In radiative equilibrium, the volume heating rate is balanced by
the thermal-radiation cooling rate, dZ/dz. So a second criterion for radiative-convective

23
equilibrium is no change in either the heating or cooling rates
dZ(Cg0 , Cw0 , T 0 ) dZ(Cg , Cw , T )
= for ζ1 < z < ζ5 . (58)
dz dz
We can integrate (58) down in altitude from the top of the atmosphere, using (57) as a
boundary condition. This gives a combined criterion for radiative equilibrium above the
tropopause,
δZ = Z(Cg0 , Cw0 , T 0 ) − Z(Cg , Cw , T ) = 0 for ζ1 < z < ζ5 . (59)
Below the tropopause, much of the heat is transported by vertical convection rather
than radiation. Manabe and Wetherald suggested that for convective equilibrium in the
troposphere, the temperature lapse rate, −∂T /∂z should equal (or not exceed) an equilibrium
value, L, that is a known function of the altitude z and of the surface temperature θ0 .
Formally, the criterion is
∂
− T (z, θ0 ) = L(z, θ0 ) for z < ζ1 . (60)
∂z
Integrating (60) we see that the tropospheric temperature is a function of altitude z and
surface temperature θ0 , given by
Z z
T (z, θ0 ) = θ0 − dz 0 L(z 0 , θ0 ). (61)
0

Tropospheric lapse rates measured by radiosondes are often quite complicated [36]. Sim-
plified functions are normally used to approximate lapse rates. Manabe and Wetherald[35]
made extensive use of the altitude-independent tropospheric lapse rate
L̄ = 6.5 K km−1 . (62)
This can be thought of as the average, over the troposphere, of a more complicated lapse
rate
1 ζ1 ∂T θ0 − θ1
Z
L̄ = − dz = . (63)
ζ1 0 ∂z ζ1
Pseudoadiabatic temperature profiles result if air with 100% relative humidity is adiabat-
ically decompressed, with liquid or solid phases of water removed as they form. But no heat
is exchanged with the environment during the decompression. Fig. 12 shows various pseu-
doadiabatic temperature profiles that can be found using the Clausius Clapeyron equation
[37]. Also shown as black, dashed lines are representative profiles with altitude-independent
lapse rates. The pseudoadiabat with a surface temperature of θ0 = −50 C has very little
water vapor and cannot be distinguished from a profile with a fully adiabatic, dry-air lapse
rate of 9.8 K/km.

7.1 Temperature adjustment

Setting z = ζ1 in (61) we see that the breakpoint temperature at the tropopause is
Z ζ1
θ1 = T (ζ1 , θ0 ) = θ0 − dz 0 L(z 0 , θ0 ). (64)
0

24
Figure 12: Temperature profiles for lapse rates that are altitude-independent (black dashed
lines) and pseudoadiabatic (colored continuous lines). The dashed black line that starts at
a surface temperature of -50 C, has an altitude-independent lapse rate of ΓW = 9.8 K/km,
the dry adiabatic lapse rate of (87). Little condensible water is available to slow the cooling
along the pseudoadiabat that starts from a surface temperature of -50 C, so the dry adiabat
and the pseudoadiabat are almost indistinguishable. The dashed black line that starts from
a surface temperature of 15.5 C (288.7 K) has an altitude-independent lapse rate of 6.5
K/km. Representative tropopause altitudes are 8 km (polar latitudes), 11 km (temperature
latitudes) and 18 km (tropical latitudes).

If the surface temperature increases by a small increment ∆θ0 , the tropopause breakpoint
temperature will increase by ∆θ1 = µ∆θ0 . From (64) we see that the tropopause temperature
magnification factor Z ζ1
∂θ1 ∂
µ= =1− dz 0 L(z 0 , θ0 ). (65)
∂θ0 ∂θ0 0
Some illustrative magnification factors µ for pseudoadiabatic lapse rates are shown in Fig.
13.
From (61) we see that a surface temperature increase ∆θ0 , will cause the tropospheric
temperature to increase by ∆T , where
∆T = µΦ1 ∆θ0 = Φ1 ∆θ1 . (66)
The temperature-adjustment basis function Φ1 (z) is
 Z z 
1 ∂T 1 ∂ 0 0
Φ1 (z) = = 1− dz L(z , θ0 ) , for z ≤ ζ1 . (67)
µ ∂θ0 µ ∂θ0 0
For altitudes above the tropopause we extend the definition (67) to

(ζ2 − z)/(ζ2 − ζ1 ) if ζ1 < z < ζ2 ,
Φ1 (z) = (68)
0 if ζ2 < z.

25
Figure 13: Magnification factor µ versus Surface Temperature for three tropopause altitudes
as indicated. The left kink occurs at the surface freezing temperature while the right kink
occurs when the tropopause temperature θ1 = 0 C.

Guided by (66) we will assume that the temperature adjustment ∆T that best restores
radiative-convective equilibrium from the surface to the top of the atmosphere is parameter-
ized by the breakpoint temperature adjustments ∆θλ , for λ = 1, 2, 3, 4, 5 . We represent the
temperature adjustment by the expansion on temperature-adjustment basis functions Φλ .
5
X
∆T = ∆θλ Φλ . (69)
λ=1

We already defined Φ1 with (67) and (68). For λ = 2, 3, 4 we write the temperature-
adjustment basis functions as

 (ζλ+1 − z)/(ζλ+1 − ζλ ) if ζλ < z < ζλ+1 ,
Φλ = (z − ζλ−1 )/(ζλ − ζλ−1 ) if ζλ−1 < z < ζλ , (70)
0 otherwise.


For the highest altitudes we take the temperature-adjustment basis function to be


(z − ζ4 )/(ζ5 − ζ4 ) if ζ4 < z,
Φ5 = (71)
0 otherwise.
Representative temperature-adjustment profiles, Φλ and flux-adjustment profiles,Vλ for an
altitude-independent lapse rate, L = 6.5 K km−1 are shown in Fig. 14.

7.2 Flux adjustments

The breakpoint-temperature adjustments, ∆θλ , can be used to find the flux that restores
radiative-convective equilibrium as

26
Figure 14: The temperature basis functions Φα are shown as dashed red lines with scales at
the tops of the panels. Altitudes at the maxima of the temperature perturbations are given
in km next to the short red lines. Multiples of the flux basis functions Vα in units of W
m−2 K−1 , and with constant relative humidity water vapor feedback, are shown as solid blue
lines, with scales at the bottoms of the panels. The dashed blue line is for flux perturbation
V1 without water vapor feedback (with Cw0 = Cw ). The flux perturbations V2 , . . . , V5 are the
same with or without water vapor feedback.

X
Z(Cg0 , Cw0 , T 0 ) = Z(Cg0 , Cw , T ) + ∆θλ Vλ . (72)
λ

The flux-adjustment basis vector for λ = 1 is

Z(Cg0 , Cw0 , T + ∆θ1 Φ1 ) − Z(Cg0 , Cw , T )

V1 = lim . (73)
∆θ1 →0 ∆θ1
Manabe and Wetherald [35] pointed out that one plausible way for atmospheric water vapor
to respond to a temperature increase would be for the relative humidity to remain constant
at all altitudes. Observational support for this hypothesis is ambiguous. For example, over
70 years of radiosonde observations indicate relative humidity has decreased slightly in the
upper atmosphere [38].
We assume that changes in water-vapor concentration have negligible effect at altitudes
above the troposophere, so that the flux-adjustment basis vectors for λ = 2, 3, 4, 5 are defined

27
by
Z(Cg0 , Cw , T + ∆θλ Φλ ) − Z(Cg0 , Cw , T )
Vλ = lim . (74)
∆θλ →0 ∆θλ
Then the adjusted forcing of (59) is
X
δZ = ∆Z + ∆θλ Vλ . (75)
λ

7.3 Optimum temperature adjustments

Between breakpoint altitudes ζλ and ζλ+1 we have taken 100 evenly spaced altitude segments.
Using (72) in (59) for each of the 400 altitude samples z above the tropopause would give
400 linear equations in the five unknowns, ∆θ1 , ∆θ2 , . . . ∆θ5 . This grossly overdetermines
the ∆θλ . However, we can find values of ∆θλ that give the best approximate solution to (59)
by minimizing
X500
Q= Wi (δZi )2 . (76)
i=1

where δZi = δZ(zi ). The adjustments ∆θλ are not very sensitive to the weights Wi , and we
used 
∆zi if zi ≥ ζ1 ,
Wi = (77)
0 if zi < ζ1 .
The altitude interval size is ∆zi = zi+1 − zi .
The temperature adjustments ∆θλ , that minimize (76) are the simultaneous solutions of
the five linear equations (λ = 1, 2, 3, 4, 5)
∂Q X
=2 Vλi Wi δZi = 0, (78)
∂∆θλ i

where Vλi = Vλ (zi ) was defined by (73) and (74). We can write (75) as the 5 × 5 matrix
equation X
Aκλ ∆θλ = ∆Sκ . (79)
λ

The adjustment matrix of (79) is

X
Aκλ = Wi Vκi Vλi . (80)
i

The source vector is X

∆Sκ = − ∆Zi Wi Vκi , (81)
i

where ∆Zi = ∆Z(zi ) was defined by (56). We assume that the adjustment matrix A of (80)
has an inverse A−1 . Then we can multiply both sides of (79) by A−1 to find
X
∆θλ = A−1
λκ ∆Sκ . (82)
κ

28
 The temperature adjustment (82) will slightly affect the breakpoint altitudes. The pres-
sure will not change for the air segment initially at the altitude zi since it is due to the weight
per unit area of all higher air segments. The ideal gas law implies that the height increment
of the ith interval becomes
T 0 (zi )
∆zi0 = ∆zi . (83)
T (zi )
The adjusted breakpoint altitudes at the top of the λth layer, with λ = 1, 2, 3, 4, 5, will be
100λ
X
∆ζλ = ∆zi0 . (84)
i=1

The surface altitude adjustment obviously is ∆ζ0 = 0.

7.4 Representative Calculations

7.4.1 Anderson H2 O profile for L = 6.5 K km−1
For the temperature profile and standard greenhouse-gas concentrations of Fig. 1, doubling
the CO2 concentration changes the flux Z as shown in the right panel of Fig. 11. The
adjustments given by (82) and (84) are as follows.

Temperature adjustments but no water-vapor adjustment.

   
1.4 0
 1.4   0.06 
   
 −2.0   0.05 
∆θ = 

 K, and ∆ζ =  −0.19  km, for ∆Cw = 0.
   (85)
 −7.2   
 −7.9   −0.65 
−2.0 −1.10

The breakpoint temperature and altitude adjustments show the lower atmosphere warms and
expands slightly after doubling the CO2 concentration while the upper atmosphere cools and
contracts.

Both temperature and constant relative humidity water-vapor adjustments.

   
2.3 0
 2.3   0.10 
   
 −2.8   0.09 
∆θ = 
 −7.0 
 K, and ∆ζ =   km. (86)
 

 −0.17 

 −8.6   −0.64 
3.8 −0.98

The temperature and altitude adjustments of (86) are shown in the left panel of Fig. 11.
Doubling CO2 concentrations with water vapor feedback increases the surface temperature
warming to ∆θ0 = 2.3 K from ∆θ0 = 1.4 K, or by a factor of 1.6.

29
7.4.2 Manabe H2 O profile for L = 6.5 K km−1
The sensitivity of the results to the water vapor profile were checked by considering the
altitude dependence of relative humidity Φ(z), used by Manabe and Wetherald [35].
 
p/ps − 0.02
Φ = Φs (87)
1 − 0.02
Here, p is the pressure at altitude z while ps denotes the surface pressure. Manabe and
Wetherald set the surface relative humidity Φs to 77%. Equation (87) fails at high alti-
tudes where it gives negative relative humidity values. The minimum H2 O mixing ratio was
therefore set to 3 × 10−6 gm per gm of air which corresponds to a concentration of 4.8 ppm.
Equation (87) gives a higher surface H2 O concentration of 13,396 ppm than 7,750 ppm found
using the H2 O profile observed by Anderson [18] shown in Fig. 1. The H2 O concentration
obtained using equation (87) decreases faster with altitude than that observed by Anderson
and corresponds to a column density about 20% higher than that given in Table 1.
The adjustments given by (82) and (84) are as follows.

Temperature adjustments but no water-vapor adjustment.

   
1.4 0
 1.4   0.06 
   
 −1.8   0.05 
∆θ =  −6.7  K, and ∆ζ =  −0.17  km, for ∆Cw = 0.
   (88)
   
 −8.9   −0.64 
0.5 −1.33

Both temperature and constant relative humidity water-vapor adjustments.

   
2.2 0

 2.2 

 0.10 
 
 −2.7   0.09 
∆θ = 
 −6.8  K, and ∆ζ =  −0.17  km.
   (89)
   
 −10.0   −0.67 
2.8 −1.23
These results for both cases without and with water vapor feedback differ very little from
(85) and (86) that were obtained using the Anderson H2 O profile.

7.4.3 Anderson H2 O profile with Pseudoadiabatic Lapse Rate

The effect of a temperature profile determined using a pseudoadiabatic lapse rate illustrated
in Fig. 12 was considered. For a surface at 288.7 K, the temperature decreases to 211.2
K at 11 km altitude. This is slightly lower than the tropopause breakpoint temperature
θ2 = 217.2 K given by (1). The higher altitude temperatures were determined using the
same breakpoint temperatures as given by (1) and all of the breakpoint altitudes remained
unchanged.
The adjustments given by (82) and (84) are as follows.

30
Temperature adjustments but no water-vapor adjustment.
   
1.0 0
 3.0   0.13 
   
 −2.0   0.15 
∆θ = 
 K, and ∆ζ =  −0.11  km, for ∆Cw = 0.
   (90)
 −7.7   
 −8.1   −0.58 
2.1 −1.05

Both temperature and constant relative humidity water-vapor adjustments.

   
2.2 0

 6.2 

 0.28 
 
 −4.2   0.32 
∆θ = 

 K, and ∆ζ =  0.01  km.
   (91)
 −7.3   
 −10.2   −0.52 
8.9 −0.53

In (90) and (91), the surface warming was found using the magnification factor µ = 3.0 at a
midlatitude tropopause altitude of 11 km.

7.5 Climate Sensitivity

A comparison of our result for the climate sensitivity defined as the surface warming due
to doubling the CO2 concentration from 400 to 800 ppm, to other work is given in Table
5. These calculations considered the case of a clear sky one dimensional atmosphere in
radiative-convective equilibrium. All groups get nearly the identical value for the case of
fixed absolute humidity for a constant lapse rate of 6.5 K/km in the troposphere. Additional
significant surface warming occurs for the case of fixed relative humidity. Our result of 2.2
K is substantially lower than the value obtained by the pioneering work of Manabe and
Wetherald [35] who obviously did not have access to the current line by line information. In
a later publication [39] the authors explained that their 1967 result for the surface warming
decreased by about 20% when they replaced their radiation transfer scheme by that used by
Rodgers and Walshaw [27] which they felt was superior. A 20% reduction of their climate
sensitivity result of 2.9 K gives the value of 2.3 K closer to that of the other groups. Hunt
and Wells [40] refined the earlier model used by Manabe and Hunt [41] to encompass 18
altitudinal levels up to 37.5 km. Kluft et al [42] calculated radiative fluxes using the Rapid
Radiative Transfer Model which is used for global climate models [43]. Their temperature
adjustments show stratospheric cooling of about 10 K similar to our results but greater
surface warming. All groups obtain similar surface warming for the case of fixed relative
humidity using a pseudoadiabatic lapse rate in the troposphere. Some variation of the results
is to be expected since the calculations used different water vapor concentration profiles as
well as temperature profiles that differ slightly near the tropopause.

31
 Model Configuration Manabe et al Hunt et al Kluft et al This Work
[35][39] [40] [42]

Fixed absolute humidity, 1.4 (1.4) 1.3 1.4

constant lapse rate (6.5 K km−1 )
Fixed relative humidity, 2.9 (2.2) 2.2 2.7 2.3
constant lapse rate (6.5 K km−1 )
Fixed relative humidity, 2.0 1.8 2.1 2.2
pseudoadiabatic lapse rate

Table 5: Climate sensitivity in Kelvins for different model configurations. The bracketed
numbers next to the results found by Manabe and Wetherald [35] are the results of our
calculation using their relative humidity profile given by (87). Our result for the case of
fixed relative humidity with a pseudoadiabatic lapse rate in the troposphere was found using
the temperature profile shown in Fig. (12) which is further discussed in the text.

8 Comparison of Model Intensities to Satellite Obser-

vations
An important test is to compare calculations to observations. Fig 15 shows vertical spectral
˜
intensities, I(0), measured with a Michaelson interferometer from a satellite over the Sahara
Desert, the Mediterranean Sea and Antarctica [44]. The figure also shows values of the
˜ calculated with (27).
vertical intensity, I,
For the Mediterranean, we used the five segment temperature profile of Fig. 1. For the
Sahara and Antarctica analogous profiles with different parameters were used. The altitude
breakpoints for the Sahara were at ζ = [0, 18, 20, 32, 47, 86] km. The high tropopause at 18
km is characteristic of near equatorial latitudes. For Antarctica, the altitude breakpoints
were ζ = [0, 2.5, 8, 25, 47, 86]. The low tropopause at 8 km is characteristic of the nighttime
poles, as is the strong, wintertime temperature inversion, peaking at 2.5 km. The lapse
rates between the breakpoints were L = [6.5, 0, −1, −3.8, 2.145] K km−1 for the Sahara and
L = [−12.5, 2.33, 0, −1.5, 2.145] K km−1 for Antarctica. The surface temperature in the
Sahara was taken to be T0 = 320 K (very hot) and the surface temperature in Antarctica
was taken to be T0 = 190 K (very cold). The surface pressure in the Sahara and the
Mediterranean was taken to be p0 = 1013 hPa and the surface pressure in Antarctic was
taken to be p0 = 677 hPa, low because of the high elevation of the ice surface, about 2.7 km
above mean sea level.
For convenience, we modeled the dependence of the water vapor concentrations C {i} on
the height z above the surface as
{i}
C {i} = C0 e−z/zw , (92)
{i}
with a latitude independent scale height zw = 5 km and with surface concentrations C0 =

32
 ˜
Figure 15: Vertical intensities I(0) at the top of the atmosphere observed with a Michael-
son interferometer in a satellite [44], and modeled with (27): over the Sahara desert, the
Mediterranean and Antarctica. The intensity unit is 1 i.u. = 1 mW m−2 cm sr−1 . Radiative
forcing is negative over wintertime Antarctica since the relatively warm greenhouse gases in
the troposphere, mostly CO2 , O3 and H2 O, radiate more to space than the cold ice surface,
at a temperature of T = 190 K, could radiate through a transparent atmosphere.

{i} {i}
31, 000 ppm for the Sahara, C0 = 12, 000 ppm for the Mediterranean, and C0 = 2, 000
ppm Antarctica. For the year 1970 when the satellite measurements were made, we used
surface concentrations, in ppm, for CO2 , N2 O and CH4 of 326, 0.294 and 1.4, with the same
relative altitude profile as those in Fig 1. The altitude profile of Fig. 1 for O3 was used for
the Sahara, the Mediterranean and Antarctica.
As can be seen from Fig. 15 the modeled spectral intensities can hardly be distinguished
from the observed values. We conclude that our modeled spectral fluxes would also be close
to observed fluxes, if a reliable way to measure spectral fluxes were invented.

9 Conclusions
This work examined the transmission of infrared radiation through a cloud-free atmosphere
from the Earth’s surface to outer space. A line by line calculation used over 1/3 million lines
of the five most important naturally occurring greenhouse gases, H2 O, CO2 , O3 , N2 O and
CH4 . This included considerably more weaker rovibrational line strengths, for H2 O as small
as 10−27 cm, than other studies. The calculation of forcings took into account the observed

33
altitudinal concentrations of the various gases as well as several temperature profiles.
The upward spectral flux, Z̃, “breaks out” at an emission height ze , given by (38).
Emission heights can be near the top of the atmosphere for frequencies in the middle of
strong absorption lines. For frequencies with little absorption, the emission heights can be
close to, or at the surface as shown in Fig. 3.
The most striking fact about radiation transfer in Earth’s atmosphere is summarized by
Figs. 4 and 5. Doubling the current concentrations of the greenhouse gases CO2 , N2 O and
CH4 increases the forcings by a few percent for cloud-free parts of the atmosphere. Table 3
shows the forcings at both the top of the atmosphere and at the tropopause are comparable
to those found by other groups.
Radiative forcing depends strongly on latitude, as shown in Figs. 7 and 8. Near the
wintertime poles, with very little water vapor in the atmosphere, CO2 dominates the radia-
tive forcing. The radiation to space from H2 O, CO2 and O3 in the relatively warm upper
atmosphere can exceed the radiation from the cold surface of the ice sheet and the TOA
forcing can be negative.
Fig. 9 as well as Tables 2 and 4 show that at current concentrations, the forcings from all
greenhouse gases are saturated. The saturations of the abundant greenhouse gases H2 O and
CO2 are so extreme that the per-molecule forcing is attenuated by four orders of magnitude
with respect to the optically thin values. Saturation also suppresses the forcing power per
molecule for the less abundant greenhouse gases, O3 , N2 O and CH4 , from their optically thin
values, but far less than for H2 O and CO2 .
Table 2 and Fig. 10 show the overlap of absorption bands of greenhouse gases causes their
forcings to be only roughly additive. One greenhouse gas interferes with, and diminishes, the
forcings of all others. But the self-interference of a greenhouse gas with itself, or saturation, is
a much larger effect than interference between different gases. Table 4 shows that for optically
thin conditions, the forcing power per molecule is about the same for all greenhouse gases,
a few times 10−22 W per molecule.
Doubling the CO2 concentration will cause a temperature decrease of the upper atmo-
sphere of about 10 K as shown in Fig. 11 to restore hypothetical radiative-convective equi-
librium. For the case of fixed absolute humidity, the surface warms by 1.4 K which agrees
very well with other work as shown in Table 5. The surface warming increases significantly
for the case of water feedback assuming fixed relative humidity. Our result of 2.3 K is within
0.1 K of values obtained by two other groups as well as a separate calculation where we used
the Manabe water vapor profile given by (87). For the case of fixed relative humidity and a
pseudoadiabatic lapse rate in the troposphere, we obtain a climate sensitivity of 2.2 K. The
corresponding climate sensitivities determined by other groups differ by about 10% which
can be expected using slightly differing temperature and water vapor profiles. The issue of
water feedback would undoubtedly be greatly clarified if additional observations of water
vapor concentration as a function of altitude were available.
Fig. 15 shows that the integral transform (27) used to calculate TOA intensities I˜ with
HITRAN line intensities and with no CO2 nor H2 O continuum absorption gives results in
very close agreement with spectral intensities observed from satellites over climate zones as
different as the Sahara Desert, the Mediterranean Sea and Antarctica. One can therefore have
confidence in the calculations of spectral fluxes. The negligible effect of the H2 O continuum
on the top of the atmosphere radiative flux has also been found by Zhong and Haigh [45]. It

34
would be interesting to examine comparable data for the tropics where atmospheric moisture
is highest to determine the effect of a H2 O continuum. One would need to be careful that
any “observed continuum” not be confused with a layer of cloud like haze which can be
prevalent at high humidities. In conclusion, the combination of one dimensional radiative-
convective models and observations such as TOA intensities are invaluable for furthering
our understanding of how increasing greenhouse gas concentrations will affect the Earth’s
climate.

Acknowledgements
We are grateful for constructive suggestions by many colleagues. Special thanks are due to
Tom Sheahen for initial encouragement and to G. Iouli who helped access the HITRAN data
base. The Canadian Natural Science and Engineering Research Council provided financial
support of one of us.

References
[1] G. Myhre et al, Anthropogenic and Natural Radiative Forcing, Climate Change 2013:
The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge University Press,
Cambridge, United Kingdom (2013).

[2] Global Greenhouse Gas Reference Network, NOAA Earth System Research Labora-
tory/Global Monitoring Division, www.esrl.noaa.gov/gmc/ccgg, (2019).

[3] S. E. Schwartz, Resource Letter GECC-1: The Greenhouse Effect and Climate Change:
Earth’s Natural Greenhouse Effect, Am. J. Phys. 86, (8), 565-576, (2018).

[4] S. E. Schwartz, Resource Letter GECC-2: The Greenhouse Effect and Climate Change:
The Intensified Greenhouse Effect, Am. J. Phys. 86, (9), 645-656, (2018).

[5] L. S. Rothman, R. L. Hawkins, R. B. Watson and R. R. Gemache, Energy Levels,

Intensities and Linewidths of Atmospheric Carbon Dioxide Bands, JQSRT, 48, 537
(1992).

[6] I. E. Gordon, L. S. Rothman et al., The HITRAN2016 Molecular Spectrospic Database,

JQSRT 203, 3-69 (2017).

[7] S. Chandrasekhar, Radiative Transfer, Dover, New York (1960).

[8] S. Manabe and F. Möller, On the Radiation Equilibrium and Heat Balance of the At-
mosphere, Monthly Weather Review 89, No. 12, 503 (1961).

[9] R. M. Goody and Y. L. Yung, Atmospheric Radiation: Theoretical Basis Oxford Uni-
versity Press, Oxford (1995).

35
[10] D. P. Edwards, A General Line-by-Line Atmospheric Transmittance and Radiance
Model, NCAR Technical Note, NCAR/TN-367+STR (1992).

[11] S. A. Clough, M. J. Iacono and J. L. Moncet, Line-by-Line Calculations of Atmospheric

Fluxes and Cooling Rates: Applications to Water Vapor, J. Geophys. Res. 97, 15761
(1992).

[12] G. Myhre, E. J. Highwood, K. P. Shine and F. Stordal, New Estimates of Radiative

Forcing due to Well Mixed Greenhouse Gases, Geophys. Res. Lett. 25, 2715 (1998).

[13] W. D. Collins et al. Radiative Forcing by Well Mixed Greenhouse Gases: Estimates
from Climate Models in the Intergovernmental Panel on Climate Change (IPCC) Fourth
Assessment Report (AR4), J. Geophys. Res. 111, D14317 (2006).

[14] H. Harde, Radiative and Heat Transfer in the Atmosphere: A Comprehensive Approach
on a Molecular Basis, Int. J. Atm. Sci. 503727, (2013).

[15] F. Schreier, S. G. Garcia, P. Hedelt, M. Hess, J. Mendrok, M. Vasquez and J. Xu,

GARLIC – A General Purpose Atmospheric Radiation Transfer Line-by-Line Infrared-
Microwave Code: Implementation and Evaluation, JQSRT 137, 29 (2014).

[16] M. Etminan, G. Myhre, E. J. Highwood and K. P. Shine, Radiative Forcing of Carbon

Dioxide, Methane and Nitrous Oxide: A Significant Revision of the Methane Radiative
Forcing, Geophys. Res. Lett. 43, 12614 (2016).

[17] The U.S. Standard Atmosphere, NASA Report TM-X-74335 (1976).

[18] G. P. Anderson, S. A. Clough, F. X. Kneizys, J. H. Chetwynd and E. P. Shettle, AFGL

Atmospheric Constituent Profiles (0-120 km), AFGL-TR-86-0110 Air Force Geophysics
Laboratory, Hanscom Air Force Base, Massachusetts, (1986).

[19] A. Einstein, Zur Quantentheorie der Strahlung, Physik. Zeitschrift 18, 121 (1917).

[20] P. A. M. Dirac, The Principles of Quantum Mechanics, Oxford University Press, New
York (1930).

[21] W. Voigt, On the Intensity Distribution Within Lines of a Gaseous Spectrum, Sitzungs-
ber. Math.-Phys. Kl. Bayer. Akad. Wis., München, p. 603 (1912).

[22] M. D. Schwarzkopf and S. B. Fels, Improvements of the Algorithm for Computing CO2
Transmissivities and Cooling Rates, J. Geophys. Res. 90,10541-10550, (1985).

[23] A. A. Lacis and V. Oinas, A Description of the Correlated k Distribution Method for
Modeling Nongray Gaseous Absorption, Thermal Emission, and Multiple Scattering in
Vertically Inhomogeneous Atmospheres, J. Geophys. Res. 96, 9027-9063, (1991).

[24] D. P. Edwards and L. L. Strow, Spectral Lineshape Considerations for limb Temperature
Sounders, J. Geophys. Res. 96, 20859 (1991).

36
[25] K. Schwarzschild, Über Diffusion und Absorption in der Sonnenatmosphäre, Nachr. K.
Gesell. Wiss. Math.-Phys. Klasse 195, 41 (1906).

[26] J. J. Buglia, Introduction to the Theory of Atmospheric Radiative Transfer, NASA

Publication 1156 (1986).

[27] C. D. Rodgers and C. D. Walshaw, The computation of infra-red cooling rate in planetary
atmospheres, Quart. J. R. Met. Soc. 92, 67 (1966).

[28] A. C. Wilber, D. P. Kratz and S. K. Gupta, Surface Emissivity Maps for Use in Satellite
Retrievals of Longwave Radiation, NASA/TP-1999-209362 (1999).

[29] K. K. Yoshikawa, An Iterative Solution of an Integral Equation for Radiative Transfer

using Variational Technique, NASA Technical Report TN-D-7292, A-4774 (1973).

[30] M. Abramowitz and I. Stegun, Handbook of Mathematical Functions, Dover Publica-

tions, New York (1965).

[31] H. Smithüsen, J. Notholt, G. König-Langlo, P. Lemke and T. Jung, How Increasing CO2
leads to an Increased Negative Greenhouse Effect in Antarctica, Geophys. Res. Lett. 42,
10422 (2015).

[32] W. Zhong and J. D. Haigh, The Greenhouse Effect and Carbon Dioxide, Weather 68,
100 (2013).

[33] S. Arrhenius, Worlds in the Making; the Evolution of the Universe, translated by Dr.
H. Borns, Harper, New York, London (1908).

[34] D. J. Wilson and J. Gea-Banacloche, Simple Model to Estimate the Contribution of

Atmospheric CO2 to the Earth’s Greenhouse Effect, Am. J. Phys. 80 306 (2012).

[35] S. Manabe and R. T. Wetherald, Thermal Equilibrium of the Atmosphere with a Given
Distribution of Relative Humidity, J. Atmos. Sci. 24, 241 (1967).

[36] J. Nash, Measurement of the Upper-Air Pressure, Temperature and Humidity, Instru-
ments and Observing Methods, Report No. 121. World Meteor. Org. (2015).

[37] R. Clausius, Mechanisache Wärmetheorie, Friedrich Vieweg und Sohn, Braunschweig,

Vol. 1, Section VI-11. (1876).

[38] G. Paltridge, A. Arking and M. Pook, Trends in middle- and upper-level tropospheric
humidity from NCEP reanalysis data, Theor. Appl. Climatol. 98, 351 (2009).

[39] S. Manabe and R. T. Wetherald, The Effects of Doubling the CO2 Concentration on the
Climate of a General Circulation Model, J. Atmos. Sci. 32, 3 (1975).

[40] B. G. Hunt and N. C. Wells, An Assessment of the Possible Future Climatic Impact of
Carbon Dioxide Increases Based on a Coupled One-Dimensional Atmospheric-Oceanic
Model, J. Geophys. Res. 84, 787 (1979).

37
[41] S. Manabe and B. G. Hunt, Experiments with a Stratospheric General Circulation Model,
Radiative and Dynamic Aspects, Mon. Weather Rev. 96, 477 (1968).

[42] L. Kluft, S. Dacie, H. Schmidt and B. Stevens, Re-Examining the First Climate Models:
Climate Sensitivity of a Modern Radiative-Conective Equilibrium Model, J. Climate 32,
8111 (2019).

[43] E. J. Mlawer, S. J. Taubman, P. D. Brown, M. J. Iacono and S. A. Clough, Radiative

Transfer for Inhomogeneous Atmospheres: RRTM, a Validated Correlated-k Model for
the Longwave, J. Geophys. Res. 102, 16,663 (1997).

[44] R. A. Hanel and B. J. Conrath, Thermal Emission Spectra of the Earth and Atmosphere
from the Nimbus 4 Michaelson Interferometer Experiment, Nature 228, 143 (1970).

[45] W. Zhong and J. D. Haigh, Improved Broadband Emissivity Parameterization for Water
Vapor Cooling Rate Calculations, J. Atmos. Sci. 52 124 (1995).

38