Step IV:Masks and Flags
Now that the chlorophyll concentrations have been visually presented, it's
time to take a look at some of the various conditions that are present in the
data.A "condition" only means that a particular pixel in the data file has
certain characteristics (it's not a disease). A simple case is where the
pixel is located on land.In that case, the land "mask" is applied.When
Level 1A data is processed to Level 2, any pixels covered by the land mask are
not processed, which saves time.
Other masked conditions are the presence of clouds (which are detected by a
high radiance value) and "radiance above the knee".The latter refers to the
way that SeaWiFS is capable of imaging both land and water. Water is much,
much darker than land, optically speaking, so SeaWiFS has two gain settings
that allow it to discriminate varying radiance levels over water and land.The
gain setting for water allows the instrument electronics to process light
levels linearly up to a certain threshold (the "knee"), at which point a higher
linear gain is used.This feature prevents saturation of the sensor over
particularly bright areas, and is referred to as a bilinear gain. When
an oceanic area is so bright that it exceeds the radiance threshold for the
lower gain setting, the "radiance above the knee" mask appears.
SeaWiFS data processing also analyzes the data for several conditions that
can make data analysis uncertain. For any pixel with one or more of these
conditions, a "flag" specific to that condition is given to that pixel.These
conditions include a large angle between the pixel and the satellite (which
occurs at the edges of the scanning swath), different types of atmospheric
aerosols, failure of the analytical algorithms (i.e., the calculation doesn't
return a meaningful value), or the presence of turbid or sediment-laden water.
Two types of phytoplankton are also flagged:coccolithophore blooms and
Trichodesmium, a species that can fix nitrogen from the atmosphere.Any
pixel can have several different flags.
The coccolithophore bloom algorithm was created based on both surface and
satellite observations of coccolithophore blooms.It uses nLw values to
determine if the characteristic optical signature of a coccolithophore bloom is
present.It isn't perfect: areas that are not coccolithophore blooms can be
flagged, and areas that very likely are coccolithophore blooms might be missed.
Even though that must be kept in mind, the algorithm works quite well.The
current coccolithophore bloom algorithm is described in Volume 9 of the
SeaWiFS Postlaunch Technical Memorandum Series.
The image displayed below shows the nLw(490) data and features three masks
(brown for land, white for clouds, and green for radiance above the knee) and
three flags (blue for coccolithophore blooms, yellow-orange for low nLw(555),
and red for absorbing aerosols). Look for each of these colors:
Most of the masks are found where they probably should be assigned (which is
desirable).The only place that the absorbing aerosol flag appears is attached
to the tail of the diamond-shaped cloudy area below the Rio de la Plata
estuary.In the true-color image, this area appeared as hazy cloud, and not as
bright as other clouds nearby.So it is possible that this is a smoke
aerosol. The low nLw(555) flag only appears on the edges of clouds.This
condition is likely due to the fact that it takes the sensor and electronics a
few milliseconds to adjust when the sensor's scan makes a transition at the
edge of a bright cloudto dark water.
Now for a closeup view of the area around the Falkland Islands:
image is the 490 nm radiance data without flags, and the right image is the 490
nm data with the flags added.In the left image, the black patches interior to
the bright areas were suspected to be the "radiance above the knee" mask;they
are green in the flagged image, so this is correct.The blue coccolithophore
flag has been assigned to a significant area of the bloom.This flag data
helps to verify the suspicion that the bright blue-green waters seen in the
true-color image are in fact due to a coccolithophore bloom.
However, also note that in the full-size image, the turbid waters of Bahia
Blanca have also been flagged as a coccolithophore bloom, and the "radiance
above the knee" mask appears inland of the coccolithophore flag.In this case
it helps to view the true-color image and see that this is an area of turbid
water, which indicates that this is less likely to be a coccolithophore bloom.
A few observations and comments, along with other relevant links, will be found on the final page.
<< Back | Next >>
Return to Index