The IOCCG Atmospheric Correction Working Group Status Report The Eighth IOCCG Committee Meeting Department of Animal Biology and Genetics University of Florence, Florence, Italy February 24-26, 2003 Menghua Wang

1
The IOCCG Atmospheric Correction Working Group
Status ReportThe Eighth IOCCG Committee
Meeting Department of Animal Biology and
Genetics University of Florence, Florence,
ItalyFebruary 24-26, 2003Menghua Wang
Contributors MERIS D. Antoine, A. Morel, B.
GentiliOCTS/GLI H. Fukushima, R.
FrouinPOLDER P. Deschamps, J-M. NicolasMODIS H.
GordonSeaWiFS M. Wang
2
Goal of the Atmospheric Correction Working Group

• The atmospheric correction working group activity
  was proposed by R. Frouin at the 5th IOCCG
  committee meeting in Hobart, Tasmania, and
  endorsed by committee and representatives of
  various space agencies participated at the
  meeting.
• The main objective of the working group is to
• quantify the performance of the various exiting
  atmospheric correction algorithms used in the
  various ocean color satellite sensors
• the derived products from various ocean color
  missions (projects) can be meaningfully compared
  and possibly merged.
• how can derived ocean color products from one
  sensor be best compared with those from others?

3
Membership

• The Working Group is composed of
• Antoine, Morel MERIS
• Dechamps POLDER
• Fukushima, Frouin OCTS/GLI
• Gordon MODIS
• Wang SeaWiFS
• Others are welcome to participate. A general
  requirement for people to join the Working Group
  is that they can contribute a well documented
  algorithm and participate some of tests.

4
Atmospheric Correction Algorithms

• The performance of the following atmospheric
  correction algorithms are intended to be tested
  and compared
• SeaWiFS/MODIS algorithm (Gordon and Wang, 1994)
• POLDER algorithm (POLDER document, Feb. 1999)
• OCTS/GLI algorithm (Fukushima et al, 1998)
• MERIS algorithm (Antoine Morel, 1999)
• Testing of the above 4 operational algorithms is
  the necessary requirement for the objective of
  the Working Group.
• Results from other algorithms for some special
  cases, e.g., Spectral Matching algorithm for
  absorbing aerosols, are also useful.

5
Parameters

• The derived parameters to be compared and tested
  are
• the normalized water-leaving reflectances at the
  visible wavelength bands
• two-band ratio values of the derived normalized
  water-leaving reflectances, i.e., 443/555 and
  490/555 and
• the atmospheric parameter--the derived aerosol
  optical thickness at 865 nm.

6
Sensor Spectral Characterizations

All comparison algorithms are operated (some have
been modified for this purpose) using the same
spectral bands of 443, 490, 555, 765, and 865 nm.
7
The TOA Reflectance (testing data) Generation
The TOA reflectances were generated based on the
following

• rw is the water-leaving reflectance from model
  (Case-1) or measurements (Case-2).
• rr is the Rayleigh reflectance.
• ?A ra rra is the aerosol and
  Rayleigh-aerosol contributions.
• t is the atmospheric diffuse transmittance.
• the sun glint and whitecap contributions are
  ignored.
• gas absorption is ignored.

8
Testing Data Sets

• Simulated Data Sets
• For the open ocean cases
• a polarized RTE (Monte Carlo method) was used for
  simulations with 15 million photons for each
  vector RTE run (within 0.5 at blue)
• TOA reflectances for spectral bands at 412, 443,
  490, 510, 555, 670, 708, 765, 779, and 865 nm
  (total 10 spectral bands) were generated
• a two-layer plane-parallel atmospheric model (78
  of molecules at the top layer)
• aerosols (Maritime with RH80, M80) located at
  the bottom layer mixed with 22 of molecules
  (Rayleigh scattering)
• aerosol optical thicknesses at 865 nm 0.05, 0.1,
  and 0.2
• a Fresnel reflecting ocean surface with pigment
  concentrations of 0.03, 0.1, 0.3, and 1.0 (mg/m3)
  from Gordon et al. (1988) model
• no gas absorption, no whitecap contributions
• the solar zenith angles 0o, 45o, 60o, 65o, 70o,
  and 78o sensor viewing angles 5o, 25o, 45o,
  55o, and 65o and relative azimuth angle of 90o.

9
Therefore, 15 million photons were used for each
vector RTE simulation
10
Uncertainty is usually within 0.5 at the blue
11
Testing Data Sets (cont.)

• Some cases for sensitivity studies (simulated
  data sets)
• absorbing aerosols Urban aerosols with two type
  vertical distributions, i.e., two-layer and
  uniformly mixed one-layer cases
• case 2 wateralthough algorithms are mostly
  intended for case 1 water, a quantitative
  estimation of atmospheric correction error over
  case 2 water is needed.
• Data from SeaWiFS measurements (this is still
  open.)
• open ocean cases (with various locations and
  seasons)
• coastal region ocean waters
• some trouble cases, e.g., nLwlt0, dust
  contamination, etc.
• For testing and comparison, SeaWiFS data sets
  are usually co-located with in situ measurements.
  It was agreed that SeaDAS will be used.

12
Diffuse Transmittance Issue

• It was realized that there were two fundamentally
  different approaches in computing the atmospheric
  diffuse transmittance and effect the atmospheric
  correction
• the SeaWiFS/MODIS algorithm assumes that the
  water-leaving radiance just BENEATH the sea
  surface is uniform.
• the POLDER algorithm (University of Lille)
  assumes that the water-leaving radiance just
  ABOVE the sea surface is uniform.
• in addition, the POLDER team includes a factor of
  the multiple surface reflection contribution,
  i.e., 1/1-Srwn .
• However, the t difference is usually within 2,
  while difference from the multiple surface
  reflection factor is within 1. Therefore, a
  simple correction to the POLDER results was
  proposed and agreed within the group. The
  correction has been applied to the POLDER results.

13
(No Transcript)
14
Atmospheric Contributions Maritime Aerosol
(2-layer)
15
Atmospheric Contributions Absorbing Aerosol
(2-layer)
16
(No Transcript)
17
NOTE Significant different contribution in
magnitude from these two type waters !!
18
Maritime Aerosol (2-layer) Cases
19
Maritime Aerosol (2-layer) Cases
20
Absorbing Aerosol (2-layer) Cases
21
Absorbing Aerosol (1-layer) Cases
22
NIR reflectances are not enough to retrieve
absorbing aerosol properties
23
(No Transcript)
24

• ALL COMPARISON RESULTS ARE PRELIMINARY!

25

• WORK IS IN PROGRESS .

26
IOCCG Report Outline

• Introduction
• Atmospheric correction working group objectives,
  members, procedures, etc.
• Overview of the atmospheric correction for ocean
  color sensors
• Algorithm Description
• MERIS
• POLDER
• OCTS/GLI
• SeaWiFS/MODIS
• Others, e.g., spectral-match algorithm for
  absorbing aerosols, etc.
• Simulated Data Set
• Brief description of the vector Monte-Carlo RTE
  for the data set
• Uncertainty of the data set, e.g., noise,
  accuracy, etc.
• Atmospheric model, e.g., two-layer, one-layer,
  aerosols M80, U80, surface, etc.
• Ocean data set Case-1 and Case-2
• Diffuse transmittance assumptions, computations,
  and two approaches
• Generating TOA data from atmosphere and ocean
  data set

27
IOCCG Report Outline (cont.)

• Comparison Results
• Open ocean (Case-1) with Maritime aerosols
• Case-1 water with absorbing (Urban) aerosols
• Case-2 water with Maritime aerosols
• Case-2 water with absorbing (Urban) aerosols
• Vertical effects for the absorbing aerosols
• Discussions
• Errors from various algorithms radiance, ratio,
  aerosol thickness
• Influence of errors in the ratio values (the
  normalized water-leaving radiance) to the
  bio-optical algorithm, e.g., the chlorophyll
  retrievals
• Cases for absorbing aerosols, Case-2 waters, etc.
• Vicarious calibration
• Others
• Recommendations and Conclusions
• Future Work
• Algorithm comparison with real satellite measured
  data, e.g., SeaWiFS data

28
Status/Time Schedule

• Setting up working group (done).
• Draft a proposal for discussing in the 1st
  working group meeting in May 16-18, 2000 (done).
• Revise working plan based on discussions (done).
• Generate the testing data sets 3-4 months
  (done).
• The 2nd working group meeting was held on
  1/18/2002 (done).
• Diffuse transmittance issue was resolved 5
  months (done).
• Algorithm testing and results analyses (on
  going).
• Write up an IOCCG report (on going).
• Workshop for the working group (planned).
• A journal paper (planned).
• Algorithm comparison with real satellite data
  (e.g., SeaWiFS, data)??? (open).