Skip to content

Fix full_width_half_max of Sentinel-2 bands - #220

Open
m-mohr wants to merge 1 commit into
stactools-packages:mainfrom
m-mohr:fix-band-fwhm
Open

m-mohr wants to merge 1 commit into
stactools-packages:mainfrom
m-mohr:fix-band-fwhm

Conversation

@m-mohr

@m-mohr m-mohr commented Oct 8, 2026

Copy link
Copy Markdown
Contributor

The full_width_half_max values in SENTINEL_BANDS are 30–150 % larger than the official Sentinel-2 bandwidths, e.g. B02 0.098 µm instead of 0.065 µm, B10 0.075 µm instead of 0.030 µm, B12 0.242 µm instead of 0.181 µm. They are roughly the full extent of the spectral response curve rather than its width at half maximum.

This PR replaces them with the mean of the S2A, S2B and S2C bandwidths, rounded to the nearest nm:

Band Old (µm) S2A / S2B / S2C (nm) New (µm)
B01 0.027 21 / 20 / 20 0.020
B02 0.098 65 / 65 / 65 0.065
B03 0.045 35 / 35 / 35 0.035
B04 0.038 31 / 31 / 31 0.031
B05 0.019 15 / 14 / 16 0.015
B06 0.018 13 / 14 / 16 0.014
B07 0.028 19 / 20 / 20 0.020
B08 0.145 118 / 115 / 114 0.116
B8A 0.033 20 / 21 / 20 0.020
B09 0.026 20 / 19 / 20 0.020
B10 0.075 29 / 29 / 33 0.030
B11 0.143 89 / 93 / 90 0.091
B12 0.242 180 / 181 / 181 0.181

Sources:

  • SentiWiki, Sentinel-2 mission, spectral band tables 3 and 4: https://sentiwiki.copernicus.eu/web/s2-mission
  • ESA Sentinel-2 Spectral Response Functions, COPE-GSEG-EOPG-TN-15-0007 v4.0 (2024), linked from the same page. Its "Bandwidth" sheet and the width at half maximum computed from the response curves agree with the new values within 1 nm (B08 depends on the definition because of its uneven response; the table above uses ESA's bandwidth).

Centre wavelengths are unchanged. Test fixtures and examples are updated with the new values only.

@tylanderson

Copy link
Copy Markdown
Contributor

@m-mohr thanks for raising this! I had just been looking at some of the new documentation that is out by ESA. Given the bandwidth varies by sensor calculating the mean is an interesting idea. One thing it might result in is these means becoming out of date once more platforms (e.g. Sentinel-2D) launch.

We could alternatively implement other options, e.g. the most lenient band width (largest), publish the documented FWHM sensor by sensor, or if there is a "design" FWHM for S2 overall publish that instead.

Curious your thoughts

@tylanderson

Copy link
Copy Markdown
Contributor

Also to be clear I'd personally be more inclined to argue for a "design"/"nominal" FWHM per band (if that exists) or a per-platform "measured" FWHM per band.

Mostly to ease the the amount of maintenance required / reprocessing has to be done when 2D launches

@m-mohr

m-mohr commented Oct 14, 2026

Copy link
Copy Markdown
Contributor Author

This was actually my first attempt at fixing it. In the other project that I was working on and where I actually found the issue, we settled on values per satellite derived from the official spectral response functions. Haven't had the time to update this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants