[gdal-dev] Extending GDAL Color Interpretation enumeration for infra-red bands (+ band wavelength) ?

[Rahkonen Jukka]

Hi,

In terrestrial applications (thermal cameras) I think I have only seen MWIR and LWIR, and due to the atmospheric windows the first typically means about 3,5-5um and the latter 8-14um. Such instruments usually support only either MWIR or LWIR. I think it would be good to separate MWIR and LWIR.

-Jukka Rahkonen-

Lähettäjä: gdal-dev <gdal-dev-bounces at lists.osgeo.org> Puolesta Daniel Evans via gdal-dev
Lähetetty: lauantai 31. elokuuta 2024 21.54
Vastaanottaja: Even Rouault <even.rouault at spatialys.com>
Kopio: gdal-dev at lists.osgeo.org
Aihe: Re: [gdal-dev] Extending GDAL Color Interpretation enumeration for infra-red bands (+ band wavelength) ?

Hi Even,

Firstly, I'll state my bias as someone involved in the production of thermal MWIR EO data (3.5-5um) - unsurprisingly, I'd like some classification that fits my data :-)

> classifications...

The NIR - SWIR - MWIR - LWIR classification scheme is by far the most common I've encountered in earth observation. I've never encountered the CIE scheme or the ISO 20473 scheme in EO - the ISO definition of Mid(wave) Infrared extending all the way out to 30um is definitely not typical usage.

Thermal Infrared (TIR) can be used as a blanket term for MWIR and LWIR. I don't know if GDAL users would find it useful to split the two, or just have a broad TIR classification; MWIR data products are still relatively rare.

I'm not immediately aware of any commonly used Far Infrared (>15um) EO data products.

On the two Github links you shared, some thoughts are:
- Both lack any entry covering the MWIR (~3-5um)
- The "Thermal" definition on the second link is LWIR only, not the more normal MWIR+LWIR definition
- The second link in particular is heavily aimed at describing Sentinel and Landsat - not unfair, given that they are the most widely used IR datasets! However, the subclassifications in the second link - "NIR", "NIR 2", "SWIR 1", "SWIR 2", "Thermal 1", "Thermal 2" - are not generic terms in my experience. The terms in the first link (swir16, swir22, lwir11, lwir12) are more familiar if GDAL were to go to that level of detail, but I think you'd still need generic "swir"/"lwir" names to cover data that don't match the Sentinel/Landsat ecosystem.


> FWHM and Central Wavelength

I think this is the most sensible description of a band.

I would be in favour of your suggestion of making them generic across all bands via the IMAGERY metadata, not just IR. The increasing number of multispectral/hyperspectral EO missions mean that simple RGB classifications for optical data without further description are not always appropriate. The stac-extensions repo already shows the addition of extra terms like "coastal", "green05", and "yellow" to further define optical bands.

No strong opinions personally on whether that wider definition uses nm or um for the unit, but perhaps others have a strong opinion on "red" being "650nm" vs. "0.65um"?

Cheers,
Daniel

On Sat, 31 Aug 2024 at 15:22, Even Rouault via gdal-dev <gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>> wrote:
Hi,

The GDAL Color Interpretation enumeration is a good start, but is quite
limited regarding band spectral properties with just Red, Green, Bland,
and nothing for other band wavelengths, particularly for infra-red.

I was looking a bit at the classifications at
https://en.wikipedia.org/wiki/Infrared#Regions , and there are several
ones, like the "commonly used subdivision scheme" (NIR, SWIR, MWIR,
LWIR, FIR) or "CIE division scheme" (IR-A, IR-B, IR-C) or "ISO 20743
scheme" (NIR, MIR, FIR).

My inclination would rather to re-use the STAC EO classification at
https://github.com/stac-extensions/eo?tab=readme-ov-file#common-band-names,
which shows a lot of similarities with
https://github.com/awesome-spectral-indices/awesome-spectral-indices#expressions
, and has a nice mapping with a few popular instruments.

While we are it, should we standardize central wavelength and full width
half max (FWHM), has special properties of a bands rather than generic
text-based metadata, like:

double GDALRasterBand::GetCentralWaveLengthMicrometer() -> NaN if unknown
double GDALRasterBand::GetFWHMMicrometer() -> NaN if unknown
void GDALRasterBand::SetCentralWaveLengthMicrometer(double)
void GDALRasterBand::SetFWHMMicrometer(double)

Or maybe just standardize a "CENTRAL_WAVELENGTH" and "FWHM" metadata
items in the "IMAGERY" metadata domain:
https://gdal.org/en/latest/user/raster_data_model.html#imagery-domain-remote-sensing
?

Thoughts?

Even

--
http://www.spatialys.com<http://www.spatialys.com/>
My software is free, but my time generally not.

_______________________________________________
gdal-dev mailing list
gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>
https://lists.osgeo.org/mailman/listinfo/gdal-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20240831/5de06a00/attachment.htm>