Page 1 of 1

BRDF correction S3 OLCI data

Posted: Fri Apr 09, 2021 11:21 am
by spardo
Hi there,

I'm processing S3 OLCI data from level 1 to level 2 using POLYMER v.4.13. The idea is to compare the POLYMER outputs with the OLCI standard product which, from the user manual, are directional water-leaving reflectances - i.e. not corrected for BRDF effects.

From the literature I see that the POLYMER v.4.13 uses the Park and Ruddick (2005) water model by default, so I'm guessing the POLYMER outputs are normalized water-leaving reflectances i.e. corrected for BRDF effects. Is that right?

Thanks in advance!

Re: BRDF correction S3 OLCI data

Posted: Fri Apr 09, 2021 3:18 pm
by fsteinmetz
Hi Silvia,

Indeed, Polymer's output water reflectances are normalized by default to a zenith-zenith geometry using Park and Ruddick (2005) - the same model that is used for Polymer's atmospheric correction.
This normalization can be disactivated by passing a value to the argument normalize.
Please bear in mind that this normalization step is directional but also spectral : before this step, Polymer works at each detector wavelength.
From polymer/params.py:

Code: Select all

# water reflectance normalization
#   * no geometry nor wavelength normalization (0)
#   * apply normalization of the water reflectance at nadir-nadir (1)
#   * apply wavelength normalization for MERIS and OLCI (2)
#   * apply both (3)
self.normalize = 3
So, in order to compare with the standard product, you may want to disactivate the directional normalization but not the spectral normalization by overriding "normalize":

Code: Select all

run_atm_corr(..., normalize=2)
I hope this helps.
Cheers,
François

Re: BRDF correction S3 OLCI data

Posted: Mon Apr 12, 2021 7:24 am
by spardo
Hi François,

Many thanks for the information, that's very helpful. I had a quick scan looking for references to BRDF in the code, but I should have used 'normalization' instead! We might leave the POLYMER dataset as it is and correct the standard product, but it's good to have the two options.


Thanks again, take care,

Silvia