You are on page 1of 14

Supplementary and background information for

Miskolczi’s greenhouse papers (Part 1.)


This document gives additional information related to the computational results
and theoretical concepts of two greenhouse related research papers. The articles were
published in the Hungarian journal Időjárás in the following issues: Időjárás Vol. 108,
No. 4, 2004, and Időjárás Vol. 111 No. 1, 2007. The articles are referred in the text as M4
and M7.

1. ● The general purpose of the above papers was to study the relationships between
the radiative fluxes of the planetary atmospheres using accurate LBL simulations with
real atmospheric structures obtained from radiosonde observations. The relationships
among the SU, OLR, ST, ED, and EU fluxes together with the total global average IR flux
optical depth, τA, and transmittance, TA are the keys to realistic estimates of global
warming due to GHG perturbations.

2.● Global Change studies require realistic global average atmospheric structures.
Considerable efforts were devoted to collect a globally representative set of radiosonde
observations and construct a global average atmosphere. For the flux density simulations
we used a subset of the TIGR radiosonde archive, which preserved the local statistical
characteristics of the original data set of 1732 observations, see Fig. 1.

3.● We obtained the TIGR radiosonde database from R. Rizzi in 1996 (University of
Bologna, Italy, private communication). This data set was designed for global scale
temperature and humidity retrievals using statistical inversion. This means, that the
regional space-time averages of the temperature and water vapor profiles are
representative for the regional climate. This is necessary to get the most realistic first-
guess profiles for the subsequent physical retrievals. The thermal and humidity structures
of the selected 228 profiles are presented in Fig. 2. For reference, the Global Average
TIGR (GAT) and the USST-76 profiles are also displayed.

4.● The motivation of this research was related to difficulties in interpreting results of
global scale extremely high accuracy broadband flux computations that was done for
remote sensing purposes. Around 2002 at NASA we were after the atmospheric window
radiation, but as a side product of HARTCODE we also had the total IR flux
transmittance the downward atmospheric emittance and the absorbed surface upward
radiation AA. Examples of the 228 original HARTCODE spectral radiance outputs in the
0.0 and 80.0 degree viewing angles are given in Figs. 3 and 4.

5. ● The 228 spectral directional radiances and transmittances computed in 9 viewing


angles were the original sources of the flux density and flux transmittance computations
which lead us to the discovery of the Su=Ed/A, Su=3OLR/2, Su=2Eu empirical laws, and
later to the confirmation of the Su=OLR/f theoretical relationship.

1
Fig. 1

Fig. 2

2
Fig. 3

Fig. 4

3
6.● To gain confidence in the results of the HARTCODE LBL radiance simulations
we present comparisons with satellite, aircraft and ground based radiance measurements.
Fig. 5 shows the geographic locations of two AIRS granules with the scan-line patterns.
The original low resolution noisy directional measured AIRS radiances are shown in the
left plot of Fig. 6. For our flux density study this radiance spectrum must be cleared and
the measurement gaps must be filled with realistic radiance estimates. In this case
HARTCODE was used to remove the noisy channels and create a continuous spectrum
using the information content of the noise-free part of the original spectrum. The cleared
radiances are shown in the right plot. This process used SVD for an implicit
instantaneous retrieval of the temperature and humidity profiles (from the noise-free
channels) and a pre-computed HARTCODE radiance database (see Fig. 7) for the gap-
filling. The next two examples (Figs. 8 and 9) compare high-resolution up- and down-
looking HIS interferometer measurements with HARTCODE radiance computations.
The purpose of these simulations was to evaluate the effect of the LPMA line mixing
model on the accuracy of the simulated high resolution radiance spectra. In all three
comparisons with spectral measurements the simulation errors were less than the
accuracy of the radiance measurements. The simulated flux densities from radiosonde
observations and HARTCODE directional radiances are regarded as the accurate
representations of the real radiative fluxes of the observed atmospheric thermal and GHG
(humidity) structure.

7. ● Comparisons between simulated ED and real downward atmospheric radiation


measurements using pyrgeometers at two different sites (at NOAA at Sterling VA USA,
and another at Neumayer Station, Antarctica, Germany) were also reported, in F.
Miskolczi, 1994, (Journal of Atmospheric and Oceanic Technology, Vol. 11, No. 1.) and
in F. Miskolczi and K. Langlo, 1997, (IRS’96: Current Problems in Atmospheric
Radiation, A. Deepak Publishing, 1997).

Fig. 5

4
Fig. 6

Fig. 7

5
Fig. 8

Fig. 9

6
8. ● As a final numerical test for the HARTCODE flux algorithm we created a control
run with a hypothetical isothermal atmosphere which gives the same OLR as the GTA
structure. In the isothermal atmosphere we used the vertical distributions of the GHGs
from the GAT profile. The spectral Kirchhoff law gives a unique opportunity to test the
computed flux densities against the theoretical expectations. For such an atmosphere the
monochromatic Kirchhoff law is trivial and exact. In Fig. 10 and 11 we present the
spectral hemispheric mean transmittances and spectral Su, ED, and ST fluxes. In this
simulation we used 1cm-1 spectral resolution. The theoretical total flux density
expectations from this absolutely unrealistic atmosphere are: OLR = SU = ED / A = EU /
A, and EU / SU = ED / SU = 1 - ST / SU = A. As we can conclude, the spectral Kirchhoff
law is perfectly reproduced with the HARTCODE, see the red line in Fig. 11. Similar,
but spectral radiance simulations for an isothermal atmosphere are usually performed to
test the numerical performance of LBL radiative transfer codes, see Kratz et al. 2005, in
JQSRT Vol. 90, page 332. The HARTCODE computational accuracy for flux
transmittance is excellent (100*(SU-ED/A)/SU = 0.0000022 per cent).

Except in the τA = 0 trivial case, isothermal atmospheres can never be in radiative


equilibrium. Radiative equilibrium atmospheres simultaneously satisfy the SU=OLR/f and
SU=ED/A relationships, see the proof in page 24 of M7. In our case the flux transmittance
TA = 0.1542, the transfer function f=2/ (1-log (TA) +TA) =0.6614, and the OLR/SU = 1 =
f. The last two equations can never be fulfilled simultaneously.

9.● In Fig. 12 and 13 simulated global average flux transmittance, atmospheric


downward emittance, and observed source function profiles are presented for clear and
cloudy GAT atmospheres. In these simulations the cloud layer is represented by a perfect
black surface at a given altitude with an infinitesimal vertical extension and in thermal
equilibrium with the surrounding air. The thermal equilibrium also assumed at the ground
surface (zero altitude).

Fig. 12 shows that the atmospheric downward emittance is equal to the atmospheric
absorption of the surface upward radiation, SU=ED/A. The whole atmospheric column is
in radiative equilibrium with the surface. This is the obvious condition for the LTE, and
the existence of a stable temperature profile. At higher altitudes this figure shows, that
any emitting cloud layer is also in radiative equilibrium with the atmospheric column
above. This is the condition of the LTE in the air column above the cloud layer.

Fig. 13 shows that up to about 3 km altitude the mean atmospheric emittance is equal to
the absorbed mean surface radiation (from ground and cloud bottom), EU+ED=(SU+B)(1-
TM),where TM is the (weighted) average flux transmittance. Apparently the atmosphere
below the cloud layer is in radiative equilibrium with the emitting surfaces at the upper
and lower boundaries. Above 3 km (not shown in the figure) the relationship changes to
EU+ED= (OLRU+OLRD) (1-TM), where OLRU and OLRD are the sums of the emitted and
transmitted flux densities at the lower and upper boundaries.

7
Fig. 10

Fig. 11

8
Fig. 12

Fig. 13

9
The newly discovered atmospheric radiative properties presented in Figs. 12 and 13 are
the definition of the Atmospheric Kirchhoff law. Later it turned out that the clear sky
Kirchhoff law at zero altitude (Fig. 12) is quite general, it is valid for local instantaneous
atmospheric structures on the Earth and closely satisfied also on the Mars. This finding
was not consistent with the ‘standard’ greenhouse theory. Since AA=ED the IR
atmospheric back radiation can not be responsible for the excess surface warming.
Further on, this relationship is not consistent with the theoretical expectations of the
semi-infinite classic Eddington solution. We recognized that there is something wrong
with our knowledge and understanding of the greenhouse effect and we had to go back to
the roots of the greenhouse theory.

10. ● In 2002 the only available theoretical relationship between the IR optical depth
and the source function profile was the classic semi-infinite Eddington solution, and its
corrected versions (which tried to resolve the surface temperature discontinuity problem).
However, the related ‘linear in τ’ equations for the semi-infinite atmosphere turned out to
be mathematically incorrect. In my view Milne’s 1922 paper was a kind of ‘trap’ since at
that time he was one of the most respected scientists in the field of the theoretical
radiative transfer, and his ideas, opinions and statements were taken to be true. During
my university time I learned his equations and explanations from the astrophysics text
books and it never occurred to me that he might be wrong. After the Su=OLR/f equation
was derived we knew that nothing is to be taken at it’s face value, the scientific
knowledge evolves in a way that from time to time one has to go back - sometimes
decades - and double check complex theoretical issues. The OLR/f equation has dramatic
consequences regarding the sensitivity of the atmosphere to GHG perturbations.

11.● In M7 we referred to the Stephens and Greenwald, 1991, and Lorenz and McKay,
2003 papers just to show that researchers still use the semi-infinite approximations. Here
we should point out that, in return for a relatively small increase in the mathematical
complexity of the new equations, one may get a far more realistic view of the radiative
properties of the real atmosphere.

12.● GCM climate simulations should produce reasonable global average atmospheric
structures (temperature, absorber, and IR flux density profiles) which comply with the
known physical laws of the radiative transfer and the observed radiative properties of the
atmosphere. The success of GW predictions using only the local primitive equations of
motions, local thermodynamics, and the continuity and conservation laws seems to be
questionable. Global climate prediction is a boundary condition problem and the physical
laws governing the global average radiative fluxes at the boundaries are explicitly
needed.

13.● Global IR budget estimates must relay on flux density computations from realistic
global average atmospheres. Unlike the USST-76 atmosphere (which was used for
example in the Kiehl-Trenberth 1997 IR budget estimate), the H2O content of the GAT
humidity profile compares very well with the 60 year global average obtained from the
NOAA Earth System Research Laboratory time series, see the comparison in Fig. 14.

10
14.● The observed clear-sky radiative structure of the GAT atmosphere is presented
(up to 20 km altitude) in Fig. 15. The black dots indicate the altitudes where different flux
density terms are supposed to be equal. At these altitudes cloud cover may compensate
for any discontinuity in the vertical IR fluxes (by evaporation or condensation) and may
interact with the system in a way that assures the overall radiation balance between the
Fo and OLR. The numerical data related to Fig. 15 are included in Table. 1. The LBL
simulation results for the GAT are slightly different from the ones in the M4 paper. Here
all the fluxes are normalized for the surface, and notice, that – due to possible nonlinear
effects – the global average fluxes from 228 radiosonde observations could be different
from the fluxes obtained from the simulation using one global average profile.

Fig. 14

15.● In the next three figures (Figs. 16-18) the relationships between the flux density
components for the Earth and Mars are presented. The data points in Figs 16 and 17
(Plots A, B, C, and D) are individual LBL simulations results, one data point belongs to
one atmospheric structure (one radiosonde observation with one original simulation
results). In these figures the crosshair indicates planetary average values. Plots E and F
in Fig. 18 displays planetary average flux density ratios for the Earth and Mars.

11
Fig. 15

Table 1. Radiative structure of the GAT atmosphere, (see Fig. 14).

Altitude OLR Ed Eu B St
Ta Τ f Comment
Km Wm-2 Wm-2 Wm-2 Wm-2 Wm-2
60.0 202.1 0.090 0.083 202.1 202.0 0.9995 0.0005 1.0000 Ed=Eu~0
38.4 221.1 2.38 2.38 220.9 218.7 0.9901 0.0100 1.0000 Ed=Eu
12.4 131.0 31.5 31.6 131.0 99.4 0.7587 0.2760 0.9829 Su=OLR
12.3 131.4 32.0 32.0 131.5 99.3 0.7551 0.2809 0.9823 Ed=Eu
6.88 189.9 108.2 81.73 219.6 108.2 0.4936 0.7060 0.9092 St=Ed
5.24 209.3 148.8 104.6 256.8 104.6 0.4083 0.8957 0.8681 St=Eu
2.24 236.4 236.4 149.9 323.9 86.5 0.2670 1.3206 0.7729 Ed=OLR
0.00 251.2 323.8 192.7 379.6 58.5 0.1542 1.8693 0.6615 Su=B

12
Fig. 16

Fig. 17

13
Fig. 18

16. ● So far we showed new empirical facts based on observations and at this point we
do no claim for any theoretical consideration. In this respect the titles of the plots or the
‘names’ of the above empirical relationships or atmospheric properties are irrelevant. The
association of the observed empirical properties with known physical laws came later. In
the next supplement (Part 2.) we shall comment the theoretical interpretations of the
observed IR radiative properties of the planetary atmospheres.

14

You might also like