Aerosol Optical Thickness inferred from TOMS and MISR data over North Africa

1
Aerosol Optical Thickness inferred from TOMS and
MISR data over North Africa
To be submitted to Atmospheric Chemistry and
Physics Discussion (ACPD) February, 2007

• Sundar A. Christopher and Pawan Gupta
• The University of Alabama in Huntsville
• Huntsville, Alabama
• James Haywood and others?
• The United Kingdom Meteorological Office
• Exeter, Devon

2
Motivation

• Although AERONET is one of the best method to
  retrieve aerosol optical thickness but satellites
  have potential of providing global coverage with
  good accuracies
• Rich Heritage (AVHRR, TOMS, MODIS, MISR, POLDER,
  OMI, CALIPSO)
• Desert Surface Albedo (0.25 to 0.40)
• Bright surface reflectance make very difficult to
  separate aerosol signal and surface signal into
  satellite observations
• MODIS does not retrieve AOT over bright targets
  such as Saharan Desert but MISR and OMI does

3
Motivation Cont

• NWP models need daily AOT values and spatial
  distribution
• But,
• MISR has narrow swath (360 km)
• OMI only observe elevated (2-3 km above the
  surface) aerosols
• Goal is to obtain AOT at 0.55 µm a product that
  is available from MISR and which compares well
  with AERONET data over desert
• Research
• Haywood et al., 2005
• Hsu et al., 1999, 2000
• AERONET AOT and TOMS AI were compared and
  regression relationship was developed
• Current research is build on these two papers

4
Study area and OMI/MISR Coverage
Time Period Two seasons Seven Years (2000-2006)
Data AERONET MISR (L2, L3) TOMS (L3) OMI (L2, L3)
5
MISR Validation with AERONET
MISR 4 channels 9 Camera 360 km
swath 17.6x17.6 km (AOT) global coverage 9
days
SEE (RMSE) 0.08
6
Monthly Mean OMI-MISR Inter-Comparison
MISR and OMI 0.5x0.5 degree monthly mean product
there are several grid points, where MISR shows
very high AOT values where as AI is low. This
could be due to monthly averaging of large number
of observations of AI whereas MISR has fewer
observations and that could corresponds to high
aerosol loading.
Dust dominant season
Mixture of dust and smoke
7
in a theoretical study, for an aerosol layer
height at 3 km, Hsu et al 1999, showed that the
AI-AOT relationship is linear up to AOT 4 for
weakly absorbing smoke aerosols up to AOT 2
for strongly absorbing smoke, and up to AOT 1
for large dust particles

• AOT-AI relationship depends on..
• Aerosol Layer height
• Type of Aerosols
• Thickness of aerosol layer (AOT)
• Data sampling
• Meteorology

Estimation of dust layer height using radiosonde
data over Saharan region suggests values between
0.5 to 3 km during winter-spring months and it
varies from 1.5 to 5.5 km during the summer
months (Hsu et al., 1999).
8
Bin Averaged OMI-MISR Inter-Comparison
Bins of 0.5 AI Jan Mar, 2001-2006 Jun-Aug,
2000-2005
9
Zonal or Regional OMI-MISR Inter-Comparison
High AOT due to smoke
10
Regression Parameters OMI-MISR Inter-Comparison
11
Validation of TOMS/OMI monthly AOT with MISR
12
Validation of TOMS/OMI monthly bin averaged AOT
with MISR
13
Spatial Distribution of MISR and OMI AOT, Jan 2006
14
Spatial Distribution of MISR and OMI AOT, July
2005
15
Frequency Distribution of difference between MISR
and TOMS AOTs
16
OMI/TOMS Validation with AERONET
17
Key Conclusions

• The MISR continues to be an excellent sensor for
  obtaining AOT over bright targets such as deserts
  as seen by the correlations between MISR and
  AERONET AOT (R 0.81)
• Although there is a large scatter due to sampling
  issues and vertical distribution of aerosols, as
  expected, there is a linear relationship between
  AI-AOT that indicates that the AI is a good
  surrogate for AOT.
• However, the AI-AOT relationship is region
  specific and is not robust over the entire study
  domain. It appears to work well in dusty
  conditions closer to the source and also when the
  AOT and AI values are high.
• The relationship is not as robust over smoke
  aerosol regions especially during low aerosol
  loadings. Extreme caution should be used when
  using AI-AOT relationship during January-March,
  between 0-10N.
• Overall, applying the AI-AOT relationship to
  predict AOT values indicates that the TOMS-AI can
  predict AOTs to within 0.2 with much lesser
  uncertainties in high dust concentration regions.

18
More than 1250 figures generated for this study
are kept here http//vortex.nsstc.uah.edu/public/
outgoing/gupta/DABEX/
19
AI is a measure of the wavelength dependent
change in Rayleigh scattered radiance from
aerosol absorption relative to pure Rayleigh
atmosphere Hsu et al., 1999, 2000.
20
Cloud Free Pixels

• Percentage of cloud free pixels for OMI-type
  pixels (13x24) is about 17 , whereas it is about
  9 for TOMS-type pixels (50x50) and only about 4
  for GOME-type pixels (40X320). If a 5 cloud
  fraction is still allowed for retrieval, these
  percentages are 27 , 21 , and 18
  ,respectively. Kerridge et al. 2001

21

• MODIS (Remer et al., 2005)
• Over Oceans
• ?t 0.05 0.15 t
• Over Land
• Dt 0.05 0.15 t
• For t 0.1 gt Dt 0.065
• ( 65 uncertainty)
• For t 1.0 gt Dt 0.20
• ( 20 uncertainty)

MISR over desert 0.08 (Martonchik et al.,
2004) 0.05 for AOT less than 0.5 and 10 for
AOT greater than 0.5 (Martonchik et al., 1998)