Retrieval of Aerosol Properties from Limb Scattering Measurements

1
Retrieval of Aerosol Properties from Limb
Scattering Measurements

• Robert Loughman
• (Center for Amospheric Sciences, Hampton
  University)
• Didier Rault
• (Climate Science Branch, NASA Langley Research
  Center)
• Terry Deshler
• (Dept. of Atmospheric Science, University of
  Wyoming)

2
Outline

• Motivation for this work
• Aerosol properties to retrieve
• Algorithm description
• Sensitivity to aerosol size distribution
• Comparison to correlative extinction coefficient
  data
• Conclusions and future work

3
Reasons to Study Stratospheric Aerosol

• Direct effect of stratospheric aerosol on global
  climate (radiative forcing)
• Indirect effects of stratospheric aerosols
• heterogenous chemical reactions
• cirrus cloud formation is influenced by upper
  tropospheric / lower stratospheric aerosols
• Improved understanding of stratospheric aerosol
  processes is needed
• Aerosol effect on LS retrievals of ozone
  (particularly if a future volcano increases the
  stratospheric aerosol loading)

4
Background Stratospheric Aerosol(?)

• In recent years, the observed stratospheric
  aerosol has dipped below previous estimates of
  the natural background stratospheric aerosol
  (Thomason, 2002), clearly implying that our basic
  understanding of the processes controlling
  stratospheric aerosols is limited
• From the Integrated Global Atmospheric Chemistry
  Observations System Theme Report (2004)
• Given that the sulfate budget of the upper
  troposphere and lower stratosphere is not well
  understood, a coherent long-term record of the
  stratospheric aerosol will be essential. With
  SAGE-III the immediate future is well covered.
• but SAGE II R.I.P. SAGE III R.I.P.
  POAM III R.I.P.
• The future of solar occultation measurements is
  very hazy and LS instruments may be asked to
  take their place in the stratospheric aerosol
  monitoring role, ready or not.

5
For ozone retrievals, aerosols are a problem
From Loughman et al. (2005)
Even for background stratospheric aerosol
loading, the LS ozone profile retrieval error can
? 20 if aerosols are neglected in the ozone
retrieval algorithm (depending on scattering
angle).
Meeting the OMPS 3 ozone profile precision
threshold requires a high-quality LS aerosol
profile retrieval.
6
Relevant aerosol properties

• Stratospheric aerosol number density na(z)
• Stratospheric aerosol size distribution (ASD)
  f(z)
• Stratospheric aerosol complex index of
  refraction f(?, z)
• Stratospheric aerosol shape f(z)
• For LS measurements, this is always likely to be
  an underdetermined retrieval problem, since all
  properties can vary with altitude. These
  properties combine to determine the aerosol
  scattering coefficient ?a(?), aerosol scattering
  cross-section ?a(?) and aerosol phase function
  Pa(?, ?) that appear in the radiative transfer
  equation

7
SAGE III LS Aerosol Extinction Retrieval Method

• Use calculated ozone profile effective surface
  reflectivity as static inputs
• Compare Id / In to Ic / In to solve for aerosol
  extinction at each wavelength independently
• Id measured radiance data
• Ic calculated radiance with latest aerosol in
  the atmosphere (updated at each iteration)
• In calculated radiance, without aerosol in the
  atmosphere
• Use assumed aerosol shape (spherical), size
  distribution, and optical properties to solve for
  the extinction profile
• Then assess consistency of derived extinction
  profiles with assumed aerosol optical properties

8
LS Radiance Profile shows an aerosol signature

• Infer Effective Lambertian Albedo at altitudes
  35-45 km
• Then solve for Aerosol Extinction Coefficient at
  altitudes between

cloud top and 35 km,
at 6 ? 520, 600, 675, 750, 870, 1020 nm
From Rault and Taha (2005)
9
Recent Algorithm Improvements

• Optimized the retrieval process (negligible
  change in retrieval results, but up to 3X
  improvement in run time)
• Introduced use of single-scattering kernels for ?
  650 nm
• Degraded spectral resolution (relative to the O3
  and NO2 retrievals)
• Experimented with the default aerosol
  characteristics used
• Index of refraction (1.448, 0) at all
  wavelengths
• Default ASD OMPS LN (log-normal, r0 0.0725 µm,
  s 1.86)

10
Perturbed ASDs
Perturbed ASDs LN (log-normal, r0 0.0695
µm, s 1.86) MG (modified gamma) Deshler
(bi-modal log-normal) J (modified
Junge) Also HG (Henyey-Greenstein, g 0.7)
11
Impact of ASD on Aerosol Phase Function
12
Impact on Retrieved Extinction Coefficient
?0 71º Retrieved ?a changes with the assumed
ASD. The altitude dependence of the change is
much smaller than the wavelength dependence
13
Can we infer ASD information from retrieved
extinction coefficients?
Q If the wavelength dependence of the retrieved
?a does not match the wavelength dependence of ?a
for the assumed ASD, can we make an intelligent
adjustment of the ASD and repeat the retrieval to
improve ?a?
14
? (aerosol phase function) ?? (retrieved
extinction coefficient)?
This linear relationship between ??a/?a and
?Pa/Pa might form the basis for an intelligent
adjustment of the ASD
15
Comparison to SAGE II solar occultation
extinction retrievals
521 nm Bias 10-15 Std Dev 25 1020
nm Bias 20-25 Std Dev 45
16
Comparison to SAGE III solar occultation
extinction retrievals
521 nm Bias 10 Std Dev 25-75 751
nm Bias 10-20 Std Dev 50-75
17
Multi-view measurement over La Reunion, June 28,
2004
 
Scan 4
Scan 3
Scan 2
Scan 1

• SAGE III made LS measurements over La Reunion
  Island for 3 successive orbits, originally to
  test the repeatability of the ozone retrieval
  algorithm
• For LS aerosol retrievals, these measurements
  also sample different parts of the aerosol phase
  function
• Are the retrieved ?a also consistent from event
  to event?

• From Rault (2005)

18
La Reunion results

• Standard deviation 10-20 at shorter
  wavelengths, 15-50 at longer wavelengths
• Retrieval at longer wavelengths is expected to be
  poorer (lower signal, more uncertainty in stray
  light removal, poorer reflectivity retrieval).
  Improvement at these wavelengths is a high
  priority for future work.

19
Multi-view measurement over Laramie, February
2006

• In February 2006, SAGE III made a series of LS
  measurements in multi-view measurement mode over
  Laramie, Wyoming
• On Feb. 13, Dr. Terry Deshlers research group
  launched a balloon over Laramie to measure
  size-resolved aerosol concentration data through
  direct sampling, up to 28 km
• We eagerly anticipate testing our data for
  consistency with this rare set of ground-truth
  aerosol properties

 
  .  
20
Laramie analysis procedure

• Retrieve ?a from SAGE III LS data, using the
  Deshler bi-modal log-normal ASD (derived from the
  balloon data of Deshler et al.)
• Divide the retrieved ?a by ?a at each wavelength,
  to obtain an effective number density neff
• A perfect retrieval would yield the same neff(z)
  profile for all wavelengths for a given scan, and
  (given the close proximity of adjacent scans in
  space and time) should also be consistent from
  one scan to the next

 
  .  
21
Preliminary results for Laramie

• Deshler number density data is lower at most
  wavelengths and altitudes
• The standard deviation of neff is fairly
  constant at 40 for a wide range of altitudes
  and wavelengths

 
  .  

• The retrieved neff decreases monotonically with
  wavelength, until increasing again for 1020 nm

22
Conclusions

• LS measurements have shown promise in delivering
  accurate and precise ozone profile information,
  but do not yet meet OMPS precision specifcations
• To provide the desired ozone retrieval
  performance, LS aerosol retrievals must be
  developed and tested
• A need also exists for global stratospheric
  aerosol monitoring that LS should be prepared to
  fill
• The LS aerosol retrieval method is not mature,
  but analysis of SAGE III LS measurements shows
  initial agreement with correlative data at the
  10-20 level, with standard deviation 20-40

23
Current/Future Activities

• Collaborate with A. Bourassa and D. Degenstein to
  compare OSIRIS and SAGE III LS aerosol retrievals
  during the INTEX campaign
• High density of SAGE III and OSIRIS LS
  coincidences
• Variety of observing conditions
• Continue to test the SAGE III LS aerosol
  retrieval with other independent correlative data
• Work with the Timofeyev et al. statistical model
  to produce the likeliest combinations of aerosol
  properties
• Experiment with a combined aerosol extinction /
  size distribution retrieval
• Collaborate with NOAA and NASA scientists to
  bring the OMPS LS retrieval algorithms (of ozone
  and aerosol) to the state of the art

24
Acknowledgements

• David Flittner, for lots of help with the
  radiative transfer modeling (especially the
  aerosol kernel calculation)
• The SAGE III instrument team, for patiently and
  carefully implementing the LS measurement mode,
  in its numerous permutations
• The SAGE II, SAGE III, and University of Wyoming
  measurement teams, for maintaining and sharing
  their high-quality data sets
• The NASA/NOAA/IPO team, for supporting LS research

25
THE END
26
BACKUP SLIDES
27
Ozone Environmental Data Records (EDRs)
Properties and Performance
Table 1. Total Column Ozone EDR Performance.
Measurement Parameter Specification
Horizontal Cell Size 50 KM _at_nadir
Range 50 DU to 650 DU
Accuracy 15 DU or better
Precision 3 DU 0.5
Long-term Stability 1 over 7 years
Table 2. Ozone Profile EDR Performance.
Measurement Parameter Specification
Vertical Cell Size 3 KM
Vertical Coverage Tropopause to 60 KM
Horizontal Cell Size 250 KM Range
0.1 to 15 ppmv Accuracy
Below 15 KM Greater of 20 or 0.1 ppmv
Above 15 KM Greater of 10 or 0.1 ppmv
Precision Below 15 KM Greater of 10 or 0.1
ppmv 15 to 50 KM Greater of 3 or
0.05 ppmv 50 to 60 KM Greater of
10 or 0.1 ppmv Long-term Stability 2
over 7 years
28
Challenges for LS Ozone Retrievals

• Altitude Registration / Pointing
• Stray Light / Measurement Noise
• 3-dimensional Ozone Field
• Stratospheric Aerosols
• Reflectivity Variations
• PSCs and PMCs
• Validation

29
even when the loading is small
Even for a low (bkgd) aerosol loading, a
sensitivity study for the LS ozone retrieval
algorithm shows that aerosol contamination is the
second largest term in the ozone retrieval error
budget
?
From Loughman et al. (2005)
30
For Volcanic Stratospheric Aerosol
Thanks to Makiko Sato
http//www.giss.nasa.gov/data/strataer/
31
Characteristics of the perturbed ASDs LN
(log-normal, from dAlmedia, 1991) r0 0.0695
µm, s 1.86 MG (modified gamma, from WRCP,
1986) r0 1 µm, a 1, ? 1, b 18 Deshler
(bi-modal log-normal) No1 0.999045, r01
0.564 µm, s1 1.521, No2 9.55d-4, r02 0.273
µm, s2 1.218 J (Junge, more suitable for
troposphere) rm 0.1 µm, ? 3