Dear Colleague,
I am forwarding this message from Chuck McLain re. SeaWiFS re-processing.
You may have received it before via another route. If so, please disregard
this copy.
Regards,
Robin Williams
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
X-Sender: chuck@seawifs.gsfc.nasa.gov
Date: Mon, 22 Nov 1999 11:22:06 -0600
To: serafino, leptoukh, ocean, acker
From: Chuck McClain
Subject: Plans for Reprocessing #3
Dear SeaWiFS User Community:
I am posting this message to all of you to update you on the SeaWiFS
Project's plans to reprocess the entire data set. Annual reprocessings are
planned in order to incrementally improve the archive products and to
redefine the archive product suite as algorithms mature and more useful
data products are recommended to the Project. The last reprocessing (#2)
was in August-September 1998. Since then, the SeaWiFS and SIMBIOS staffs
have been working diligently to address problems that remain in the data
products. The primary concerns are negative water-leaving radiances
(mostly in the 412 nm and 443 nm bands) and unreasonably high chlorophyll-a
values in coastal areas. The Project's efforts have been complemented by
independent development activities by the MODIS Oceans Team, the SIMBIOS
Science Team, and other investigators. The Project hosted two algorithm
miniworkshops during last spring and summer to promote collaboration among
the various groups and met a third time during the SIMBIOS Science Team
meeting in September. Much of the work that has resulted from these
meetings is posted at
http://seawifs.gsfc.nasa.gov/~sbailey/wkshp_main.html. Since the SIMBIOS
Science Team meeting, the Project has been focused on defining what
specific enhancements are to be incorporated into this reprocessing (#3).
The Project has settled on a set of recommended improvements which are
described below. More detailed descriptions will be posted on the
wkshp_main website in a manner similar to the descriptions posted for
reprocessing #2. Also, the Project will be documenting most of this work
in the SeaWiFS TM series. We hope to initiate the reprocessing by the end
of the calendar year. However, much testing and evaluation of the final
processing code must be completed first as the entire suite of level-1, -2,
and -3 codes have been rewritten. Please review the information provided
below and submit comments to me if you wish.
Chuck McClain
SeaWiFS Project Manager
Reprocessing #3 Algorithm Modifications: Where necessary, short
descriptions are provided.
Calibration
1. Time dependent corrections for Bands 1, 2, 5, & 6
Band 7 & 8 corrections initiated at Reprocessing #2
2. Refined vicarious methods for bands 1-6 using MOBY
3. Refined vicarious adjustment of band 7
4. Bilinear gain "knee" offset adjustments: This is a result of
discontinuous radiance frequency distributions in the vicinity of the
bilinear gain knee. Adjusting the offsets alleviates this problem and
improves cloud masking and absorbing aerosol detection.
Level-2 Processing
5. Replace Oceanic 90 model with Coastal 70 model: The Oceanic 90 model
was incorporated at reprocessing #2, but has been found to intersect other
models in some situations leading to discontinuities in the derived products.
6. Whitecap correction. The Gordon & Wang (1994) correction is reduced by
60% based on recent data collected by Howard Gordon and Ken Voss. An 8 m/s
limit on wind speed is imposed to assure that over-correction does not
occur. The Frouin et al. (1996, revised) spectral dependence algorithm is
applied.
Gordon, H. R., and M. Wang, Influence of oceanic whitecaps on atmospheric
correction of ocean-color sensors, Appl. Opt., 33, 7754-7763, 1994.
Frouin, R., M. Schwindling, and P.-Y. Deschamps, Spectral reflectance of
sea foam in the visible and near-infrared: In situ measurements and remote
sensing implications, J. Geophys. Res., 101, 14,361-14,371, 1996.
7. Epsilon extrapolation for values outside aerosol tables: In situations
where the estimated epsilon (765,865) value falls outside the range defined
by the aerosol models, an analytical function is used for computing the
epsilons for the other wavelengths is used, rather than defaulting to the
nearest model values.
8. Siegel et al. (1999) NIR reflectance correction using bands 7 & 8.
Several groups have been working on the aerosol correction problem over
water masses where the NIR water-leaving radiance is not zero. The
miniworkshop participants reviewed a number of approaches, all of which are
in the developmental stage. At this point, the Project will use the
simplest algorithm, Siegel et al. (1999), which is based on Case 1 data
and theory. The correction does reduce the occurence of negative Lw values
and provides much reduced estimates of chlorophyll-a concentration in
coastal areas.
Siegel, D. A., M. Wang, S. Maritorena, and W. Robinson, Atmospheric
correction of satellite ocean color imagery: The black pixel assumption,
submitted to Appl. Opt., 1999.
9. Explicit sun glint correction outside glint mask. Outside the present
glint mask, artiifacts of glint are still present in the aerosol optical
thicknesses. The glint radiance is presently estimated in the processing
code for the glint mask. The algorithm has been refined the algorithm so
as to improve the aerosol optical thickness values. The correction does
not affect the water-leaving radiances or the chlorophyll-a values
significantly.
10. Absorbing aerosol flag. Absorbing aerosols are presently interpreted
as high chlorophyll and can result in negative water-leaving radiances when
the reflectance is low enough not to trigger the cloud and knee radiance
masks. An absorbing aerosol flag will be derived from the absorbing
aerosol index (described below) and will be an exclusion criteria for
level-3 binning.
11. Ozone interpolation modification. The present ancillary data scheme
assumes a fixed time of day for the ozone data and interpolates between
consecutive days to estimate the ozone value at a given pixel. For TOMS
ozone, the orbit is similar enough to SeaWiFS to avoid this time
interpolation and a new scheme has been implemented.
12. Out-of-band corrections applied to water-leaving radiances. It is
well-known that the SeaWiFS instrument has significant out-of-band
response. The atmospheric correction accounts for this, but no correction
is being applied to the Lw values. The error is typically greatest at 555
nm. The Project has incorporated a correction scheme into the reprocessing
code.
13. Mueller (1999) algorithm for K(490) using Lwn(490)/Lwn(555) ratio.
The existing K(490) algorithm uses Lw(443) which has problems in turbid and
bloom situations where Lw(443) is small.
14. Additional masks & flags
a. Change solar zenith angle flag to 65 degrees. Clear water Lwn's start
deviating from nominal values at higher zenith angles.
b. Partition atmospheric correction failure mask into 3 masks
c. Partition chlorophyll algorithm failure/valid range flag into 2 flags
d. Add absorbing aerosol flag
e. Add stray light correction flag
f. Add sun glint correction flag
g. Add 1 km depth flag
15. Drop insertion of L1 values into L2 fields. Inserting level-1 data
into the areas identified as clouds/ice, sun glint, etc. resulted in many
complaints and complicates the SeaDAS display program.
Level-3 Processing: no algorithm changes. The absorbing aerosol flag will
be used as an exclusion criteria.
Finally, aside from algorithm improvements, the Project is proposing to
modify the archive product suite. The Project has polled the SeaWiFS
Executive Council and the miniworkshop participates on various suggestions
and has taken their comments into account. These changes are meant to
provide more geophysically meaningful products and to set the stage for an
expanded set of archive products to be generated during reprocessing #4,
whenever that event may occur. The idea is to gradually improve and expand
the archive product set in a judicious manner so as to not overwhelm the
DAAC. The recommendation for this reprocessing is to drop three products
and add two others. The specific changes are the following:
1. Remove "CZCS-pigment". This product can be easily computed from the
chlorophyll-a product as described in O'Reilly et al. (1998) and in SeaWiFS
TM Vol. 43 (page 7).
O'Reilly, J. E., S. Maritorena, B. G. Mitchell, D. A. Siegel, K. L. Carder,
S. A. Garver, M. Kahru, and C. McClain, Ocean color chlorophyll algorithms
for SeaWiFS, J. Geophys. Res., 103(C11), 24,937-24,953, 1998.
2. Remove "La(670)" and "La(865)". The aerosol radiances themselves do
not seem to be the best products for characterizing global distributions of
aerosol properties and concentrations. The present suite of products
includes tau_a(865) and epsilon(765,865) and the Angstrom exponent is being
recommended as a new product.
3. Add Angstrom exponent (510,865). The Angstrom exponent provides
additional information on the aerosol phase function (King et al., 1999)
and the relative magnitude of aerosol radiances at different wavelengths.
The 510 nm selection is an attempt to provide a value that is near the
middle of the SeaWiFS visible bands, i.e., most representative of the
Angstrom exponent, and is compatible with values available from the AERONET
for validation purposes.
King, M. D., Y. J. Kaufman, D. Tanre, and T. Nakajima, Remote sensing of
tropospheric aerosols from space: past, present, and future, Bull. Am.
Meteorol. Soc., 80(11), 2229-2259, 1999.
4. Add "Lwn(670)". Many investigators working in Case 2 waters have
requested this product.
After reprocessing #3 is completed, the Project will begin posting
"evaluation products" which have been suggested to the Project as
potentially valuable to the Earth science community. These products can be
oceanic, atmospheric, or terrestrial and can be generated in processing
streams separate from the present suite of archive products. Processing in
separate streams allows for the use of optimized coding, algorithms, and
quality control parameters, i.e., masks and flags. The products will
available from the Project on a rolling update basis, i.e., systematically
replaced, and an on-line browse capability will be implemented. If, by the
time of the subsequent reprocessing, the user community deems a product of
sufficient quality and usefulness, the Project will include it in the
archive product suite. At this time, the following products have been
identified and will be generated. Others may be added as long as there is
sufficient justification and a viable algorithm are provided.
Initial Evaluation Products:
Ocean:
1. Coccolithophore distributions. These will be generated by compositing
the coccolithophore flag. Time scales for compositing are TBD.
2. Trichodesmium distributions: An flag algorithm to be provided by Ajit
Subramanium will be implemented and the distributions will be generated by
compositing this flag. Time scales for compositing are TBD.
3. Alternative chlorophyll-a products: An algorithm (TBD) which utilizes
443 nm has been requested by several investigators. Also, the OC4(O'Reilly
et al.) empirical and Carder MODIS semianalytic algorithms are candidates.
Land:
4. Normalized Difference Vegetation Index (NDVI). The Project has been
producing this product routine since fall of 1997 and providing it to Jim
Tucker (Code 923, NASA/GSFC) for evaluation. Recently, this activity has
been formalized with the approval of NASA HQ so that land vegetation
products will be archived at the DAAC. The time schedule for production
and archival is being worked with Tucker's group and the DAAC.
Atmosphere:
5. Daily Average Photosynthetically Available Radiation (PAR). Robert
Frouin has provided an initial algorithm and processing code to the
Project. The processing code has been developed and tested within the
Project by Bryan Franz.
6. Monthly Percent Cloud Cover. The percent cloud cover will be derived
using the cloud mask. An algorithm has been implemented by Wayne Robinson
of the SeaWiFS Project.
7. Cloud Optical Thickness: An algorithm based on Nakajima and King (1990)
has been implemented by Menghua Wang of the SIMBIOS Project.
Nakajima, T., and M. D. King, "Determination of the optical thickness and
effective particle radius of clouds from reflected solar radiation
measurements, I, Theory", J. Atmos. Sci., 47, 1878-1893, 1990.
8. Absorbing Aerosol Index: This is a product similar to that derived from
NOAA AVHRR and TOMS. An algorithm has been implemented by Christina Hsu of
the SeaWiFS Project.
***********************
Dr. Robin G. Williams
Code 902
NASA Goddard
Greenbelt
MD 20771
Tel: 301-614-5264
Fax: 301-614-5268
***********************