I have been checking different SeaDAS atmospheric corrections in Rio de la Plata in multiple images and am consistently seeing higher Rrs values from white aerosol extrapolation (aer_opt=0, having set far SWIR band ~2200 nm as extrapolation band) than the correspondent Rrs values retrieved in the case of no aerosol correction (aer_opt=-99).
I would have always expected this to be the other way round, Rrs(AC = white aerosol) < Rrs(AC = black aerosol), since the first one actually should subtract more signal (assuming aerosols are mainly adding) to the TOA budget. Maybe the white aerosol option does not perform Rayleigh correction?
I am currently seeing this behavior in many MODIS/VIIRS imagery of various years. Is this expectable for you?
I haven't found the way to attach files in the post, but as an example I tried this in VIIRS-NOAA20 image V2020059174800.
In both types of atmospheric correction I have set aer_wave_short=aer_wave_long=2257 (or close to this value), so I don't get masked the NIR bands.
Other values are set the same for both, as follows:
Thanks in advance!
I think you've identified a bug. It appears that the code is setting transmittance to 1 when the aer_opt=-99 option is selected. This shouldn't be, as there is still an atmosphere in the way - even if it's assumed to have no aerosols. With a transmittance of 1, you'll get higher Lw (thus higher Rrs) than you would if there were a reasonable estimate of transmittance (one that you get with aer_opt=0).. I suggest you don't use aer_opt=-99 until we have a chance to squash this bug.
OK, I may have been mixing things in my brain...there also seems to be a transmittance issue with the aer_opt=0 when aer_wave_long=2257. It seems more reasonable with 868, 1238 or 1604, although the reasonableness (in my estimation) goes down with increasing wavelength. Haven't wrapped my head around that one yet, but I'd avoid aer_opt=0 with aer_wave_long=2257 for now as well.
The bug in the aer_opt=-99 is real though :grin: , so progress toward cleaner code!
Just a quick follow up. The issue with the decreasing transmittance with increasing SWIR wavelength is likely due to the retrieved AOT. The SWIR bands do not have a vicarious calibration applied and are likely not retrieving valid AOT. Higher AOT translates to lower transmittance. So, yeah, probably best to avoid using the aer_opt=0 with swir bands.
...and the aer_opt=-99 bug is fixed. It will make it into the next SeaDAS release.