Atmospheric correction: what is it, how does it work, and can I change it?

Atmospheric correction is the process by which the atmospheric component of the
top-of-atmosphere signal (the signal seen by the satellite) is removed so that
the water-leaving radiance can be estimated.

The atmospheric correction algorithm used operationally by NASA to process
ocean color data is based on the work of Gordon and Wang (1994). 
This is the same code (MSl12) provided with SeaDAS. 

The atmospheric correction is based on the assumption of zero (or at least negligible)
radiance in the NIR bands. The NIR bands are used to select the aerosol model to
use in the atmospheric correction. For waters where this assumption is not valid, the
algorithm has been modified to estimate the water-leaving radiance in these NIR bands,
and subtract it prior to the aerosol retrieval. This "NIR correction" is not perfect, but it does
greatly improve retrievals in non-Case 1 waters.

MSl12 provides the capability to alter many of the parameters used in the atmospheric
correction code.  For example, alternative aerosol selection options are provided should
the standard method prove insufficient.  See the user guide for details.

Are the default atmospheric correction methods for MODIS and SeaWiFS identical?

The atmospheric correction method for all the sensors MSl12 recognizes is identical.  Where
necessary, sensor specific corrections/algorithms are used (e.g. polarization correction for MODIS). 
All the required look up tables (Rayleigh, aerosol, etc) have been tailored to the specific sensors,
or are interpolated where possible from the LUTs for other sensors.  The process is as similar as
possible to ensure consistency between sensors.