1. Accounting for non-sphericity of aerosol particles in photopolarimetric remote sensing of desert dust
Oleg Dubovik (UMBC / GSFC, Code 923)
Alexander Sinyuk (SSAI, Code 923)
Tatyana Lapyonok ( GSFC, Code 923)
Brent Holben ( GSFC, Code 923)
Michael Mishchenko (NASA/GISS)
Ping Yang (Texas A&M University)
Anne Vermeulen (SSAI, Code 923)
Tom Eck (UMBC/GSFC, Code 923)
Ilya Slutsker (SSAI, Code 923)
Hester Volten (Free University,Netherlands)
Ben Veihelmann (SRON Space Res., Netherlands)

2. Outlines:
Simulating non-spherical dust scattering in remote sensing retrievals
Fitting laboratory polarimetric measurements of dust light scattering
Sensitivity of polarimetric measurements to aerosol parameters
Applications to AERONET polarimetric retrievals

3. Difficulties of accounting for particle non-sphericity
Difficulties of accounting for particle non-sphericity in aerosol retrievals:
many limitations in simulating light scattering by non-spherical particles (on particle size, shape, refractive index, etc.)
2. Simulation are too slow for operational retrievals (much slower than Mie scattering by spherical particle)
3. Concept of choosing particle shape is unclear
4. Validation of models is ambigious
Main limitations of T-Matrix code (Mishchenko et al.):
- only spheroid shape (?)
size parameter ≤ ~ 60
aspect ratio ≤ 2.4
speed (for large aspect raitos) ~ 100 times slower than Mie

4. AERONET model of aerosol
Simplest model of non-spherical aerosol
How to implement operationally ???
Is this correct???
Randomly oriented
spheroids :
(Mishchenko et al., 1997)

5. Modeling Polydispersions
V(ri)
V(ri)
Kernel look-up table for fixed ri (22 points)
(1.33 ≤ n ≤ 1.6; ≤ k ≤ 0.5)

6. Single Scat. By spheroids
Single Scattering using spheroids:
Model by Mishchenko et al. 1997:
particles are randomly oriented homogeneous spheroids
w(e) - size independent aspect ratio distribution
K - kernel matrix:
0.05 ≤ r ≤ 15 (mm)
1.33 ≤ n ≤ 1.6
≤ k ≤ 0.5
0.4 ≤ e ≤ 2.4

7. Single Scattering using spheroids
spheroid kernels data base for operational modeling !!!
K - pre-computed
kernel matrices:
Input: n and k
Input: wp (Np =11),
V(ri) (Ni =22 -30)
Basic Model by Mishchenko et al. 1997:
randomly oriented homogeneous spheroids
w(e) - size independent shape distribution
Time: < one sec.
Accuracy: < 1-3 %
Range of applicability:
0.15 ≤ 2pr/l ≤ 280 (26 bins)
0.4 ≤ e ≤ 2.4 (11 bins)
1.33 ≤ n ≤ 1.6
≤ k ≤ 0.5
Output: t(l), w0(l),
F11(Q), F12(Q),F22(Q),
F33(Q),F34(Q),F44(Q)

8. Modeling of dust light scattering by mixture of spheroids
n(l)
k (l)
w(e) - size independent shape
distribution
Averaging with w(e)

9. Modeling of dust light scattering by mixture of spheroids
n(l)
k (l)
w(e) - size independent shape
distribution
Averaging with w(e)

10. Computational challenge
Computational challenge of using spheroids (phase function)
Contribution of different
sizes to scattering at 1200
Mishchenko and
Travis, 1994
Yang and Liou, 1996

11. Computational challenge
Computational challenge of using spheroids (polarization)
Contribution of different
sizes to scattering at 1200
Mishchenko and
Travis, 1994
Yang and Liou, 1996

13. Inversion of Scattering Matrices
Forward Model:
F11(l ,Q), -F12 (l ,Q)/F11 (l ,Q)
F22/F11 , F33/F11, F34/F11, F44/F11
Numerical inversion:
-Accounting for uncertainty (F11; -F12/F11 !!!)
- Setting a priori constraints
aerosol particle sizes,
refractive index,
single scattering albedo,
aspect ratio distribution

14. Fitting of Measured Scattering Matrix by spheroids model
Feldspar
0.441 mm

15. Role of total reflectance
Accounting for polarization in radiation
transmitted through the atmospheric
L1; L2 - rotation matrices
Total:
phase
matrix !!!
I =
- Stokes vector
F(Q;l) =
- Intensity
-Linear Polarization

16. Fitting of Measured Scattering Matrix by spheroids model
Feldspar
0.633 mm

17. Fitting of Measured Scattering Matrix by spheres
Feldspar
0.441 mm

18. Size and shape distributions retrieved from Scattering Matrix
Spheroids
Aspect ratio
distribution
dV(r)/dlnr

19. Sensitivity of Linear Polarization of fine mode aerosol to real part of refractive index
Log-normal monomodal dV(r)/dlnr : sv = 0.5,
m = 0.44 mm, k = 0.005

20. Sensitivity of Linear Polarization of coarse mode aerosol to real part of refractive index
Log-normal monomodal dV(r)/dlnr : sv = 0.5,
m = 0.44 mm, k = 0.005

21. Shape effect in presence of Multiple Scattering (Radiance)
Log-normal monomodal dV(r)/dlnr : rv= 2mm, sv = 0.5,
m = 0.44 mm, n = 1.45, k = 0.005

22. Shape effect in presence of Multiple Scattering (Polarization)
Log-normal monomodal dV(r)/dlnr : rv= 2mm, sv = 0.5,
m = 0.44 mm, n = 1.45, k = 0.005 t
t =1.0

23. AERONET Polarized Inversion
Forward Model:
Single Scat:
Multiple Scat:
DEUZE JL, HERMAN M, SANTER R, JQSRT, 1989
Successive Orders of Scattering Code
t(l), I(l,Q),P(l,Q)
Numerical inversion:
-Accounting for uncertainty (F11; -F12/F11 !!!)
- Setting a priori constraints
aerosol particle sizes,
refractive index,
single scattering albedo

24. Inversions of intensity and polarization measured by AERONET
Banizombu
(Africa)
Sept. 26, 2003
t(0.87)~ 0.5

25. Inversions of intensity and polarization measured by AERONET
Cape Verde
July 12,2001
t(0.87)~ 0.6

26. Inversions TESTS of intensity and polarization measured at 4 wavelengths
Solar Vilage
t(1.02)~ 0.4

27. Modeling Desert Dust Lidar Ratio
Muller, et al., 2003:
S(0.532mm)= 50~80sr
Dhabi Aerosol
S=19
S=50
S=80

28. Conclusions:
Kernel look-up tables seems to be promising for remote sensing retrievals
Spheroids may closely reproduce laboratory polarimetric measurements of dust scattering
Spheroid model is successfully employed in both intensity and polarized AERONET retrievals
Sensitivity to particle shape is a challenge for utilizing polarization for aerosol retrievals