30 
which can provide continuous long-term data at 
a single location and as a function of depth in the 
water column. Vertical strings of bio-optical sen- 
sors, deployed strategically in the major ocean 
provinces, are required to provide the comple- 
mentary surface information necessary to fully 
exploit the scientific return from the SeaWiFS sat- 
ellite data. Unattended buoy systems would pro- 
vide the long-term bio-optical data from various 
ocean provinces and would be used (1) to inves- 
tigate both the long-term and vertical-distribution 
behavior of bio-optical algorithms, thus increasing 
the accuracy of those models, linking the dis- 
solved and suspended biological material to the 
subsequent optical properties, (2) as a platform 
for the direct long-term determination of the verti- 
cal distributions of pigment biomass, primary pro- 
duction, and ocean carbon flux, and (3) as a 
component of the multiplatform stratified sampling 
strategies for optimizing remotely sensed esti- 
mates of pigment biomass, primary production, 
and carbon flux. 
Validation of the SeaWiFS data product will 
require at least one dedicated cruise of at least 
30 days in addition to a "calibration" cruise short- 
ly after launch. Similar cruises during the opera- 
tion of the sensor will confirm its on-going perfor- 
mance and stability. The cruises should be de- 
voted to measuring pigment concentration con- 
tinuously along the ship's course as well as up- 
welling spectral irradiance, downwelling spectral 
irradiance, pigment concentration, etc., at posi- 
tions within the field of view of the sensor. This 
data will also provide a basis for developing bet- 
ter bio-optical algorithms. 
Remote sensing of ocean color with the four 
CZCS visible channels is well understood. It is 
based on firm physical principles of radiative 
transfer and environmental optics as well as on a 
considerable body of experimental data linking 
biological constituents in ocean waters with their 
corresponding optical properties. Additional 
spectral bands are proposed for SeaWiFS that 
should greatly improve the accuracy of chloro- 
phyll estimates, particularly in areas with high 
chlorophyll concentrations and with suspended 
sediments or dissolved organic matter (Case-2 
waters). There is also significant room for im- 
provement in the algorithms used for estimating 
primary production from ocean-color imagery. In 
particular, a robust and general algorithm for the 
purpose of handling imagery from various regions 
and over long time periods is needed, such as 
would be required in order to make basin-wide 
productivity estimates from color imagery. Re- 
cently, several biological optical models that 
could fill this need have been proposed. A syste- 
matic effort is required to evaluate, test, and re- 
fine these models using sample data from a wide 
variety of ocean environments. The objective is 
to develop a generally agreed-upon algorithm by 
the time SeaWiFS is operational, so that we may 
proceed to produce global ocean productivity 
maps in support of GOFS and other studies 
planned for the early 1990s. 
To achieve the accuracy in pigment- 
concentration estimates associated with the 
CZCS, the existing CZCS atmospheric-correction 
algorithm can be directly adapted to SeaWiFS 
data. The bands at 665, 765 and 865 nm, dis- 
cussed in Section 4, where the ocean approxi- 
mates a blackbody, will be used to determine the 
aerosol radiance and its spectral variation. How- 
ever, some studies will be necessary to establish 
the most accurate method of extrapolating the 
spectral variation in these bands into the visible 
region. 
To take advantage of the 10-bit sensitivity of 
the SeaWiFS sensor, compared to the 8-bit sensi- 
tivity of the CZCS, a more careful analysis of at- 
mospheric correction will be required. The as- 
sumption that Rayleigh and aerosol contributions 
to the radiance at the sensor can be separated 
will not be valid at the full 10-bit resolution of the 
sensor, and a more complex algorithm will be 
needed. Development of such algorithms has 
been underway for 4 to 5 years, their feasibility 
has been demonstrated, and they will be in place 
by launch. These improved algorithms should 
enable correction of atmospheric effects almost 
to the 58.25° edge of scan of the baseline Sea- 
WiFS instrument. 
