Different coordinates between OLCI L2 product and C2RCC output


In my work I’ve come to realise that when processing a sentinel 3 L2 OLCI product with C2RCC, the input (L2 product) pixel coordinates are different to the output (C2RCC) pixel coordinates.

As an arbitary example, for a randomly selected pixel in following product (and it’s L2 and C2RCC counterparts) S3A_OL_1_EFR___20160425T221505_20160425T221705_20170928T191525_0119_003_243______MR1_R_NT_002

latitude -34.476915
longitude 172.90746
latitude -34.476915
longitude 172.90746
TP_latitude -34.478897
TP_longitude 172.9085

As you can see, the L1 and L2 products have the same coordinates, but the C2RCC coordinates are different. The distance between these two coordinates for this arbitrarily selected pixel in fact turns out to be over 200m.

So; which coordinate should I be using, and why does this difference exist?

Thanks for all the help,

Hello Oli,
You can see these differences already in the original OLCI products.


OLCI products provide geo-location in two kinds.

  1. Tie-points.
    For each row in the scene only 77 tie-points are provided. In between the locations are interpolated by SNAP.
  2. Bands
    For each pixel a geo-location is provided. This geo-location is also orthorectified, the tie-points not.

Which information is used by S3TBX can be set in the options (Tools / Options)

The selection has also influence on the speed of processing. Using the per-pixel geo-location is often slower.

So, I think in your case the tie-points have been used to process the data with C2RCC and for the original data you are looking at the bands.

1 Like

Thanks for all the help marpet :slight_smile: