Why is S2 MSI B8 (833nm) central WL different than product (842nm)?

Optical Toolbox
#1

Apologies if this is a silly query, kind of a newbie here.

  1. The MSI central wavelength for band 8 (NIR) is either 832.8 nm for S2A and 832.9 for S2B (https://earth.esa.int/web/sentinel/technical-guides/sentinel-2-msi/msi-instrument)
    or 835.1 for S2A and 833 for S2B (Sentinel-2 - Missions - Resolution and Swath - Sentinel Handbook - Sentinel Online), with varying bandwidths, depending on where you look. Can someone please confirm that the first source/wavelengths are the correct ones?

  2. The product central wavelength for band 8 is 842 nm (https://earth.esa.int/web/sentinel/technical-guides/sentinel-2-msi/performance). How is this difference in wavelengths explained? Is this due to spectral response functions? I can understand the difference between MSI and product central wavelengths in bands 11 and 12 for example, since there is some difference between S2A and S2B wavelengths.

Many thanks

#2

Hi @dimitris_p

That’s probably because the editor (me!) neglected to change the Performance page - which if I recall correctly we wrote based on the pre-launch literature - when he updated the other pages. My apologies. It will be updated.

Cheers

Jan

S2 MPC Operations Manager

#3

Hi @Jan,

Thanks for the reply. I guess you’re referring to the inconsistency between the 2 web pages? But I still don’t get how/why would the product central wavelength for band 8 (842 nm) be that different from the central wavelength of the same band for the MSI (833 nm). It seems to me that if the instrument, for both S2A and S2B, is capturing at 833 nm, the product for that band would be of the same wavelength? or am I missing something?

Thanks again,
Dimitris

#4

Hello Dimitris,
Are you referring to the wavelength displayed in SNAP, in Product Explorer?
That is just a general constant, you can see the real central wavelength in product metadata:

image
image.png1703×1040 237 KB

Best regards,
Oana

#5

that is interesting. How does this affect tools like the spectral unmixing which (at least according to what I thought) take the value from the band properties?

#6

Maybe @obarrilero can give further details on spectral wavelengths handling?

#7

Hi @oana_hogoiu

Yes, that’s what i was talking about. Thank you very much for the clarification. So most bands have a slightly different central wavelength than what appears in the product explorer, why is that?

Also very interested to see how, if at all, this affects spectral unmixing. For instance, if importing endmember signatures, which values will be used?

#8

Hello @dimitris_p

I believe the answer you are looking fro regarding the bands not being aligned with the central wavelength can be determined from the Spectral Response Functions (S2-SRF) spreadsheet in the User Guide Document Library.

Cheers

Jan

#9

Hello @Jan

Yes that was my initial thought in the post. But I was unable to figure out why 842nm; it turns out i missed the overview sheet from the SRF file and was trying to use all values where there was a response to make sense of the central wl.
I’m still not getting the difference with the metadata though, if 842nm is the one that results from the SRF.

Many thanks again and please excuse my ignorance and persistence.

1 Like