I’m having trouble understanding how the processing for the L1 to L2 in OLCI marine products works. I’ve read the User Handbook Manual and it states that the processing in made in parallel using the BAC for Case 1 waters and the AAC (C2RCC) for Case 2/complex waters. I’ve downloaded the L1 and L2 images for a certain AOI and I extracted pixel from the L2 downloaded and from the L2 obtained by applying the C2RCC (Optical - Thematic Water Processing - C2RCC - OLCI) on L1. Both extraction do not show the same values… I downloaded the image data set from the EUMETSAT website. Do they use a different algorithm for AC?
In addition to what I wrote as answer there.
In the L2 marine products created by the ground segment the same neural nets are used for the water part. But the input to those nets is different because a different atmospheric correction is used.
I’m not sure I understand. The L2 marine products are produced by the ground segment with the same neural network (from C2RCC). The AC used is not coming form the BAC in both cases? If I only apply the C2RCC processor available in SNAP I’m only applying the neural nets to the radiance available without any AC? And if I download the L2 I have the BAC + C2RCC?
Actually I was wrong.
The AC is the same in the ground segment and in SNAP/C2RCC.
But the AC used for the C2RCC in the ground segment is not the one used for water leaving reflectances in the L2 Water products. That’s the AAC.
However, still the inputs to C2RCC are different in snap and the ground segment.
There are preprocessing steps, like the smile correction, or the gaseous absorption correction which are not applied within SNAP.
I am now more confused about the difference between the chla inversion result of the ocean part of S-3 L1 using C2RCC and the water products of S-3 L2? Are they using the same neural network?