Step II: Level 2 Data Processing
Processing Level 1A SeaWiFS data to Level 2 is accomplished using SeaDAS, the SeaWiFS Data Analysis System. SeaDAS processing, which duplicates the data processing performed by the SeaWiFS Project, does a remarkable number of things when it processes a scene to Level 2. Using input data -- either contained in the SeaWiFS data itself, such as the radiances at 765 and 865 nm, which are used for atmospheric correction, or additional meteorological and atmospheric ozone data -- the influence of the atmosphere is removed for each pixel, and the radiances are converted to "normalized water-leaving radiances" (nLw). nLw means that the radiance values are equivalent to what would be measured right on the ocean surface with the sun directly overhead. When oceanographers on research cruises make ocean optical measurements, they attempt to duplicate these conditions as closely as possible.
Data processing also analyzes the data for a number of different conditions. If any these conditions are detected, the pixel is assigned a "flag" or a "mask". Flags and masks will be discussed in Part IV.
After the nLw values are calculated, other algorithms can be applied to the data. The most popular SeaWiFS algorithm is the chlorophyll algorithm, which calculates the concentration of chlorophyll in milligrams per cubic meter of water (mg m-3). The updated SeaWiFS chlorophyll algorithm is called the "Ocean Chlorophyll 4 (OC4)" algorithm, because it uses four SeaWiFS bands. This algorithm is described in the recently published Volume 11 of the SeaWiFS Postlaunch Technical Memorandum Series.
Dr. Jay O'Reilly provided a summary of how the chlorophyll algorithm works:
"OC4 is a 'maximum band ratio' algorithm where the maximum of three band ratios (443/555, 490/555, and 510/555) is used to predict chlorophyll concentration.
"Over most of the deep ocean, chlorophyll concentrations are below 0.3 mg m-3, and water-leaving radiance in the 443nm band exceeds the radiance in the 490nm and 510nm bands. At chlorophyll concentrations above 0.3 mg m-3 and below 1.5 mg m-3 (values typically found on continental shelves), water-leaving radiance in the 490nm band is usually greater than the values for the 443 and 510nm bands. Finally, at chlorophyll concentrations above approximately 1.5 (mgm-3), frequently found near shore, water-leaving radiance in the 510nm band exceeds that measured in the 443nm and 490nm bands. In fact, in both chlorophyll-rich waters and phytoplankton blooms, the estimate of water-leaving radiance for the 443nm band (after atmospheric correction) may be noisy and too low to make accurate chlorophyll estimates. The OC4 algorithm takes advantage of the natural shift in the dominant radiance band, and by using the brightest band (443,490,510) in the band ratio, the algorithm is able to estimate chlorophyll concentrations with a high level of accuracy over the wide range that exists in the global ocean."
The results of data processing and the application of the OC4 algorithm are shown below. This is a "grayscale" image, where the chlorophyll concentration values are assigned different levels of gray, ranging from black to white. The coastline for all land masses has been added (in red).
It is now obvious that there is considerably more oceanic activity here than first meets the eye (compare this image to the true-color image on the first page). This oceanic region is a very dynamic region where two strong ocean currents interact (a previous Science Focus! page on convergence zones explains what's happening here in more detail). The interaction of these currents gives rise to the complex circulation patterns and variable chlorophyll concentrations that are apparent in this image.
Note that in the region near Bahia Blanca where smoke may be present, data processing has removed some of the interfering haze. However, the thicker areas are interpreted as cloud (shown here as black).
The phytoplankton bloom area near the Falkland Islands doesn't appear very prominent in this image. This is due to the fact that the chlorophyll algorithms are based on the absorption of light by the chlorophyll pigment in phytoplankton cells. Because coccoliths are highly reflective, they reflect much more light than they absorb even though they contain chlorophyll.
Now compare the chlorophyll concentration image above to the grayscale image shown below. This image displays nLw(490), the normalized water-leaving radiance at 490 nm.
The suspected coccolithophore bloom near the Falklands is much brighter here, because sunlight at a wavelength of 490 nm is reflected strongly.One of the characteristics of coccolithophore blooms is that they are almost pure white: they reflect all wavelengths of visible light very efficiently.The small areas of black that appear on the brightest areas of the bloom are a SeaWiFS data "mask" that will be explained in Part IV.
To demonstrate the difference between nLw and chlorophyll concentration, two close-up views of a feature that looks somewhat like a lobster claw are shown below. This feature appears just above the line of clouds that separates the two clear water areas of the image.
The image on the left is the chlorophyll data, and the image on the right is the nLw(490) radiance data.The lobster claw feature has higher chlorophyll concentrations than the water surrounding it, so it absorbs more light at 490 nm than the waters around it.It therefore appears as a darker feature in the radiance data.
The next step will be to add color to the chlorophyll data, to get a better idea of where the chlorophyll concentrations are high and where they are low.
<< Back | Next >>
Return to Index