Random noise produced by ARC SST Processor with SLSTR

Optical Toolbox
#1

When processing SLSTR products with the Arc.SST Processor (v0.1) available in SNAP (5.0, S3TBX version 5.0.2), the result seems to be corrupted by random noise. The noise pixels are not consistent when the same product is reprocessed identically several times, but they change, and there is no indication in the original dataset of the reason that could produce this noise.

Below is an image showing the results from four identical processings of the same SLSTR product and the random “moving” noise is clearly visible. The product was
S3A_SL_1_RBT____20170507T084735_20170507T085035_20170508T143057_0179_017_221_1980_LN2_O_NT_002.SEN3

I was using the SLSTR ARC N2 coeffs as Nadir-view coefficient file.

Is there some bug in the processor, or am I missing some important steps or options in my processing?

S3A_SL_1_RBT____20170507T084735_20170507T085035_20170508T143057_0179_017_221_1980_LN2_O_NT_002_ARC_SST.png1531×960 150 KB

#2

Just to give quick feedback.
I can confirm your observation. And the noise is not neglectable. I noticed that the value range increases with each run. First the maximum value for this scene is 529.37, than 535.21 and the third run gives a maximum of 559.015.

Thanks for reporting. We will investigate this.

#3

This should be fixed with the next module update of SNAP, which will be out tomorrow or next week.