Fast Radiative Transfer Codes For Infrared Imaging Radiometry

1
Fast Radiative Transfer CodesForInfrared
Imaging Radiometry

Philippe Dubuisson ELICO, Université Littoral
Côte dOpale, Wimereux Jacques Pelon, Olivier
Chomette and Anne Garnier Institut Pierre Simon
Laplace, Paris, France
CALIPSO Algorithm Workshop, Williamsburg,
November 2003
2
Study Objectives
IIR algorithm Cloud effective emissivity ?k
retrieval for each band k

background reference radiance (RTM or LIDAR)
Measured radiance (IIR)
Opaque Cloud brightness temperature of a
blackbody source at Zc (RTM or LIDAR)
3
Study Objectives
- For the IIR algorithm To calculate Rk,bg and
R(TZc) for a blackbody source Radiative Transfer
Code assuming opaque media using the top of the
cloud main inputs temperature, pressure and
water vapor profiles - For more accurate
simulations or sensitivity studies Several
levels of accuracy Important objective for the
IIR mission semi-transparent clouds Influence
of scattering on the infrared radiation

4
Reference Code
Line-By-Line Code DISORT LBLDOM (Dubuisson
et al., 1996, JQSRT) Plane parallel layers High
spectral resolution HITRAN for line
absorption (Rothman et al., 2003, JQSRT) CKD2.4
absorption continuum for H2O (Clough et al,
1989) Aerosol and cloud layers Spectral
integration for gaseous absorption ? very long
computing times
Comparisons on the transmission with the MODTRAN
code for the IIR bands

5
Gaseous absorption
H2O
Transmission T and effects ?T on the brightness
temperature for a MidLatitudeSummer atmosphere

O3
CO2, CH4, N2O

• Correction of transmission using
• Standard models (O3 and CO2) and
• Meteorological data for H2O

IIR
6
Correlated k-distribution
Approximation for H2O absorption exponential
series for the transmission (Lacis et Oinas,
1991, JGR)

Fast method to account for absorption and
multiple scattering Absorption coefficient
ki(Po,To) computed with the Line-By-Line Using
the IIR spectral response Vertical Dependence
scaling approximation
Simple correction for O3 and CO2
7
Radiance calculations
Radiances are computed for each term of the
exponential series. Radiances are computed for
the rectangular function for which the radiance
of a blackbody is equal to those with the real
spectral response 8.7 0.4 µm 10.6 0.3
µm 12 0.5 µm Brightness temperatures
interpolations from Look-Up-Tables built from the
IIR spectral response (accuracy of about 0.01 K)

Organigramme
8
Radiative Transfer Models

9
Comparisons for absorbing media
Comparisons with the reference code for the three
IIR bands. Brightness temperatures for purely
absorbing atmospheres. Representative dataset of
176 atmospheric profiles from radiosondes (with
P, T, H2O and O3 profiles) 0.1 lt H2O lt 6.4
g/cm2 600 lt Ps lt 1030 hPa Accuracy of about 0.3
K on the brightness temperature adequate with
the expected IIR accuracy

10
Comparisons for absorbing-scattering media
Inter-comparisons of codes For purely
absorbing media LBLDOM / FASRAD, FASAA,
FASDOM RMS error lt 0.3 K For
absorbing/scattering media FASDOM /
LBLDOM comparisons for low or high
clouds Maximal error 0.4 K RMS error 0.2
K for cloudy atmospheres Influence of
scattering comparisons between FASDOM / FASRAD
or FASAA analyzed for liquid or ice
semi-transparent clouds

11
Comparisons for Liquid Clouds
g
Water Liquid droplet with Mie theory Size
distribution gamma-modified Effective radius 4,
8, 16 and 32 µm Optical thickness at 12 µm from 0
to 30 3 atmospheric models Tropical atmosphere
with 1- low cloud between 2-3 km (805-715
hPa) 2- high cloud between 5-6 km (559-492
hPa) 3- low high cloud

?ext
?0
wavelength
12
Influence of scattering on the brightness
temperature
?T FASDOM (FASRAD or FASAA)
FASRAD opaque clouds
Low cloud with Reff 8 µm Convergence for
optical thicknesses gt 20 Large deviations ( gt 8K)
for FASRAD assuming opaque clouds for clouds with
COTlt5 FASRAD Opaque clouds is not considered in
the following comparisons for semi-transparent
clouds.
FASRAD, FASAAA
13
Influence of scattering on the brightness
temperature
Low cloud with Reff 8 µm

8.7 µm
12 µm
10.6 µm
14
Influence of scattering on the brightness
temperature
Acceptable Error Interval
15
Influence of scattering on the brightness
temperature
Deviations on the brightness temperature T
between the radiative transfer codes with respect
to reference code FASDOM DT (FASDOM
FASRAD) DT (FASDOM FASAA) As a function of
- the Cloud Optical Thickness for the 3
atmospheric models as a function of - the
effective radius.
16
Low cloud (2-3 km)
FASDOM - FASRAD
FASDOM - FASAA
10.6 µm
12 µm
8.7 µm
17
Low High Cloud
FASDOM - FASRAD
FASDOM - FASAA

10.6 µm
12 µm
8.7 µm
18
High cloud (5-6 km)
FASDOM -FASRAD
FASDOM - FASAA
12 µm
8.7 µm
10.6 µm
19
Conclusions on Tb for Liquid Water Clouds 1/2
1. Thick clouds ( 20 lt dc ) FASRAD opaque
clouds and FASAA small influence of scattering
a significant bias only at 8.7 µm (about 2K for
Re4 µm) 2. Semi-transparent clouds ( 0 lt dc lt
20) FASRAD (purely absorbing media ?o
0) Brightness temperatures are biased low
radiation from the surface is underestimated Devi
ations are larger for large particles Large
errors for all IIR bands, especially at 8.7 µm
due to scattering FASAA (Absorption
Approximation ?o ? 0) Brightness temperatures
are biased high backscattering of radiation is
not accounted for Deviations are larger for
small particles Deviations are always less than
1K at 12 µm Deviations are always less than 2K
at 10.6 µm Deviations can reach - 6K at 8.7 µm
(dc 2)
20
Conclusions on Tb for Liquid Water Clouds 2/2
FASRAD Fast calculations Accurate
calculations only for dense clouds FASAA
(Absorption Approximation ?o ? 0) Fast
calculations Accurate calculations for both
dense and thin clouds at 10.6 and 12
µm. Correction factor for scattering effects
at 8.7 µm possible using Look-Up-Tables to
avoid systematic deviations for the brightness
temperature. mainly as a function of effective
radius (pre-defined), optical thickness (lidar,
WFC) and cloud altitude (lidar).
21
Influence of Scattering on Brightness
Temperature Differences (BTD)
Cloud Optical Properties from a split-window
technique BTD for couples of wavelengths 8.7
10.6 µm 8.7 12 µm 10.6 12 µm BTD as a
function of the brightness temperature at 12
µm And as a function of the effective radius
22
Split-window technique BTD analysis for High
Cloud
FASDOM
FASAA

FASRAD
8.7 10.6 µm
8.7 12 µm
10.6 - 12 µm
23
Split-window technique differences on BTDs for HC
BTD (FASDOM) - BTD(FASRAD)
OK for Regt30 µm
OK for most Re
BTD (FASDOM) - BTD(FASAA)
OK for most Re
OK for Regt30 µm
8.7 12 µm
10.6 - 12 µm
24
Influence of scattering for the split-window
technique for high level water clouds (5-6 km)
With the FASAA code 8.7 10.6 µm weak
signal 8.7 12 µm strong influence of
scattering up to 6 K for small particles (Reff
4 µm) lt 1.5 K for Reff 16 µm 10.6 12 µm
weak influence of scattering lt 1 K Effect of
scattering on the retrieval of the effective
radius Reff error 4 µm using 8.7 12
µm error 1 µm using 10.6 12 µm Base
analysis on 10.6-12 µm analysis at low optical
depth for water clouds with small Re (same for
ice spheres)
25
Influence of scattering for Ice clouds
4 ice clouds models are considered in single
layers 1) Ice cloud layer at 20 oC
(effective diameter 33 µm) 2) Ice cloud layer
at 40 oC (effective diameter 37 µm) 3) Ice
cloud layer at 60 oC (effective diameter 12
µm) 4) Ice cloud layer at 60 oC (effective
diameter 33 µm) Microphysical Properties for
Cirrus Models Baum et al., JGR, 2001 (Cloud
properties using MODIS Airborne Simulator) With a
Tropical Atmosphere Model.
26
Deviations on the Brightness Temperature
FASDOM - FASRAD
FASDOM - FASAA
10.6 µm
12 µm
8.7 µm
27
Deviations on the Brightness Temperature
FASDOM - FASAA
8.7 µm
12 µm
28
Split-Window Technique for Cirrus Clouds
T - 20oC De 33 µm
FASDOM FASAA

T - 60oC De 12 µm
8.7 10.6 µm
8.7 12 µm
10.6 12 µm
29
Split-Window Technique for Cirrus Clouds
Cirrus case 3 at T - 60oC (De 33 µm)
FASDOM FASAA

Cirrus case 4 at T - 60oC (De 12 µm)
8.7 12 µm
10.6 12 µm
30
Conclusions for ice clouds
Large errors on Tb with the FASRAD code up to 30
K !! Errors on Tb with the FASAA code lt 6 K
at 8.7 µm lt 2 K at 10.6 and 12 µm Effect of
scattering on the BTDs using the split-window
technique ? weak influence using 10.6 12 µm
?BTD lt 2 K ?BTD can reach 6 K using 8.7 12
µm, Comparable to the ones obtained for water
clouds at low COT but appear to be less dependent
on De ? Correction for scattering due to cloud
properties optical thickness (lidar) single
scattering albedo (cloud model) To be continued
and finalized for Build 4