I am trying to use all available optical sensor data in addition to MODIS [i.e. Landsat OLI, Sentinel 2 MSI, Sentinel 3 OLCI ] for a Chesapeake Bay water quality monitoring program.
I have been testing various l2gen options for OLI. When aer_opt is set to -2, the ch-a statistics differ substantially from cases when aer_opt is -1. I am not sure how to choose aer_opt for this processing. I have not tried any other options.
Another question, especially for OLI, MSI, and MODIS is about using SWIR for aer_short and aer_long. There are more missing pixels when SWIR bands are used - compared to default but the ch-a statistics are not too different.
Are there rules on how to decide these options? Particularly, aer_opt.
There are no rules, but some options are more (or less) applicable depending upon the specific needs of the user. That said, setting
aer_opt=-1does not enable the NIR correction (as is done with -2 ...and -3, which is effectively the same as -2 now), and should not be used over waters for which the black pixel assumption is violated. The waters of the Chesapeake definitely violate that assumption.
As for using the SWIR bands, Werdell et al., (2010, https://doi.org/10.1016/j.rse.2010.04.027) details some of the limitations of the SWIR channels on MODIS for atmospheric correction....with a particular focus on the Chesapeake. It boils down to an issue of inadequate signal to noise. The OLI and MSI instruments are focused on land applications and so the SNR is rather low and the bandpasses rather wide for ocean color applications, but they can be of use if these limitations are understood and accounted for when these data are used.
atmocor=0, since no aerosol subtraction is performed for that product....so yes, the result would be the same regardless of the aer_opt setting
The processing will be faster with
atmocor=0,but if you need any products that require Rrs, you cannot use this setting (Rrs requires aerosol subtraction)