1. Radiative transfer in the thermal infrared and the surface source term
Session 2 - Impact of thermal infrared surface emissivity uncertainty on trace gas retrieval
Introduction
Radiative transfer in the thermal infrared and the surface source term
How is emissivity taken into account in the trace gas retrieval algorithms
Example of IASI retrieval algorithm (FORLI)
Spectral emissivity in the CO retrieval spectral range
Example of Zhou et al. climatology (from IASI on IASI sampling)
Impact of emissivity on the CO retrievals –preliminary results-

2. Preliminary remark Thermal infrared Shortwave infrared IASI - FORLI
MOPITT V5
MOPITT V6 uses TIR and SWIR channels
Thermal infrared
Shortwave infrared

3. 1. Radiative transfer in the thermal infrared and the surface source termz q
The general equation
Radiance at the beginning of the light path
Total transmittance over the light path
source term
from the medium (thermal emission, scattering…)
Weighting function
Radiance at the end of the light path
radiance from the medium weighted by absorption through upper layers
Initial radiance transmitted through the entire path

4. (255 K on average for the troposphere)
1. Radiative transfer in the thermal infrared and the surface source term
Nadir
In the nadir THERMAL infrared, both term are equally important and cannot be neglected 
Air temperature
(255 K on average for the troposphere)
surface temperature
(288 K on global average)

5. Grey-body surface emission
1. Radiative transfer in the thermal infrared and the surface source term
Nadir thermal infrared – more details –
Looking at 180° (no angle)
TOA
surface source term?
zTOA
= surface spectral emissivity
= effective reflectivity
Grey-body surface emission
Reflected solar radiation (negligible below ~2200 cm-1)
Reflected downward atmospheric radiation

6. IASI radiances (W / cm2 sr cm-1)
1. Radiative transfer in the thermal infrared and the surface source term
Concretely
IASI radiances (W / cm2 sr cm-1)
Total atmospheric
transmittance
Atmospheric source term
(emission + scattering)
Transmittance from z’ to z
Surface source term
Dominant term =
Emissivity × Blackbody
Surface thermal emission
Reflected
solar radiance
Reflected downward
radiance from atmosphere
Becomes significant when e < 0.95
Becomes significant above 2200 cm-1 (daytime)

7. IASI radiances (W / cm2 sr cm-1)
1. Radiative transfer in the thermal infrared and the surface source term
Concretely
IASI radiances (W / cm2 sr cm-1)
Total atmospheric
transmittance
Atmospheric source term
(emission + scattering)
Transmittance from z’ to z
Surface source term
>> Brightness -equivalent blackbody-
temperature (K)
Tskin if e =1

8. !!! Emissivity !!! IASI radiances (W / cm2 sr cm-1)
1. Radiative transfer in the thermal infrared and the surface source term
Concretely
IASI radiances (W / cm2 sr cm-1)
Total atmospheric
transmittance
Atmospheric source term
(emission + scattering)
Transmittance from z’ to z
Surface source term
>> Equivalent blackbody
(brightness) temperature
!!! Emissivity !!!
Sand emissivity from Zhou et al. climatology

9. 2. How is emissivity taken into account in the retrieval algorithms ?
Concretely
Spectral emissivity modifies the surface source term in two ways:
it decreases the surface thermal radiance which would otherwise be described by pure a blackbody function
It allows for the reflection of the downwelling atmospheric radiation
While the emissivity is pretty close to unity and relatively constant over the entire infrared spectral range for some surfaces (oceans), see, it can be characterized by sharp spectral variations for several land surfaces, especially between 1000 and 1200 cm-1 and above 2100 cm-1
Wavenumber-dependence of TIR land emissivity

10. Grey-body surface emission
2. How is emissivity taken into account in the retrieval algorithms ?
Example for IASI (FORLIv )
= surface spectral emissivity
= effective reflectivity
Grey-body surface emission
Reflected solar radiation (negligible below ~2200 cm-1)
Reflected downward atmospheric radiation
Planck blackbody function at the temperature Ts with a spectral emissivity .
The skin temperature is retrieved together with the CO profile, using the same spectral fitting window.
For continental surfaces the spectral emissivity relies on the climatology of [Zhou et al. 2011]. In cases of missing values in the Zhou et al. climatology, the MODIS climatology of Wan [2008] is used.
A constant sea surface emissivity (possibly varying with wind speed) is used
calculation of the mean radiance associated to the total downward flux reaching the surface, integrated upon all the geometries. This is done
considering a Lambertian surface.
The third term, accounts for the reflected solar radiance in the direction of the sounding beam. It is calculated using a
Planck blackbody function at 5700 K, without including spectral lines,
a reflective surface combining Lambertian and specular reflections.

11. Monthly global variability of emissivity at 2150cm-1
3. Spectral emissivity in the CO retrieval spectral range
spatial and temporal variability of TIR land emissivity
Zhou et al. climatology
Monthly global variability of emissivity at 2150cm-1
Jan
Feb
March
Apr
May
June
July
Aug
Sept
Oct
Nov
Dec

12. 3. Spectral emissivity in the CO retrieval spectral range
Zhou et al. climatology
variability of emissivity between 1800 and 2760 cm-1 above given surfaces
Sahara
Western US
CO retrieval spectral range for IASI FORLI
cm-1
Greenland
Europe
Figures M. Van Damme
Dashed lines: min and max
Plain line: mean
Shadow: standard deviation
spatial and temporal variability of TIR land emissivity
Wavenumber-dependence of TIR land emissivity

13. 3. Spectral emissivity in the CO retrieval spectral range
Zhou et al. climatology
variability of emissivity between 1800 and 2760 cm-1 above given surfaces
Sahara
Western US
CO retrieval spectral range: cm-1
Greenland
Europe
Figures M. Van Damme

14. 4. What is the impact of emissivity on the IASI CO retrievals
4. What is the impact of emissivity on the IASI CO retrievals? –preliminary results-
First FORLI version with MODIS emissivity database (12 channels only in the thermal IR)
FORLI version v with first Zhou et al. emissivity database (all IASI channels but monthly averages.
Figure by Maya George

15. Residual bias (W/(cm2.sr.cm-1))
4. What is the impact of emissivity on the IASI CO retrievals? –preliminary results–
Larger RMS and biases above hot surfaces, including deserts
Impact on total retrieval error is, however, limited
JUNE 2008
IASI morning overpass
Retrieval error (%)
CO total column (molec/cm²)
Residual bias (W/(cm2.sr.cm-1))
RMS (W/(cm2.sr.cm-1))
Figures M. Van Damme

16. Residual bias (W/(cm2.sr.cm-1))
4. What is the impact of emissivity on the IASI CO retrievals? –preliminary results–
Larger RMS and biases above hot surfaces, including deserts
Impact on total retrieval error is, however, limited
Diurnal variability of e?
JUNE 2008
IASI evening overpass
Retrieval error (%)
CO total column (molec/cm²)
RMS (W/(cm2.sr.cm-1))
Residual bias (W/(cm2.sr.cm-1))
Figures M. Van Damme

17. Total retrieval errors
4. What is the impact of emissivity on the IASI CO retrievals? –preliminary results–
emissivity
emissivity
emissivity
IASI morning overpass
JUNE 2008
Total retrieval errors
RMS (W / cm2 sr cm-1)
Bias (W / cm2 sr cm-1)
emissivity
emissivity
emissivity
IASI evening overpass
JUNE 2008
Figures M. Van Damme
Some correlations between decreasing emissivity and larger errors/biases/RMS
Some (but weak) differences in correlation patterns day and night
Other impacts still to be verified (temperature vs. emissivty?)