Refelctance for the Dark Target product is reported in reflectance “units” using the following forumal
R = (L / F0) * (pi / mu0),
where L is the radiance at the top of the atmosphere, F0 is the solar irradiance constant (for the given wavelength * distance to sun, etc), and mu0 is cosine of solar zenith angle.
However, different instrument teams (MODIS, VIIRS, SeaWIFS) have chosen to report Level 1B “reflectance” in different forms.
In the case of MODIS, the reflectance aquired from the Level 1B product (L1B) is
R = (L / F0) * PI, so if you want true “reflectance” you have to divide by mu0.
At the same time, different retrieval teams (e.g. DT, DB) have chosen to do retrievals in different space.
For Dark Target, we do our retrieval in terms of true reflectance units, so our reported “Mean_Reflectance_Land” SDSs have the 1 / mu0 applied.
The 6S atmospheric correction can be applied directly to these values.
Deep Blue works in L/F0 space. So, if you want “reflectance”, you have to apply the (pi/ mu0) to the “Deep_Blue_Spectral_TOA_Reflectance_Land”
In summary, if you want to do atmospheric correction from the data in the Level 2 file,
1) For DT: Use what is reported.
2) For DB: Multiply by (PI/mu0).