Add SAR swath information

[emmanuelmathot]

This PR adds the field sar:swath composed of beam objects to describe precisely subswaths.

It fixes #17

[m-mohr]

@emmanuelmathot Is any of this something users would search on?

[fabricebrito]

@m-mohr yes, user searchs for S1 Stripmap with beam S6 to create an interferometric pair

[fabricebrito]

@emmanuelmathot I suggest adding an example of a Stripmap product in the README and stripmap-item.json. It eases the adoption

[m-mohr]

@m-mohr yes, user searchs for S1 Stripmap with beam S6 to create an interferometric pair

Okay. For search the current structure of an array of objects is not ideal. If we want to enable easy searching we should probably make the structure simpler or so.

[fabricebrito]

Yes, I agree with going for a simpler structure and keep the focus on discovery rather than extensive metadata in the STAC Item.
Additional information can be found in the metadata.xml (or equivalent) of the SAR product.

[m-mohr]

Maybe just sar:beam_ids: ["IW1", "IW2"] (or sar:swatch_ids)? (I assume beams and swaths is the same, but I'm probably wrong?)
The view:* fields can be provided in Item properties.
How useful is it to know the relative burst and polarization per beam?

[davidraleigh]

Is this useful for satellites other than sentinel-1?

[davidraleigh]

seems like this is maybe more appropriate for the sentinel-1 extension. unless there are other data providers where this is useful?

[emmanuelmathot]

The idea is to have a scalable description of the swath with an object describing the beams. Each beam is a specific steering of the instrument and thus has it's own parameters (i.e. view).
Concerning the search, I do not see any limitation or complexity for filtering

A_CONTAINS(sar:swath.id,["S1","S2"])

Shall work like

A_CONTAINS(sar:subswath_ids,["S1","S2"])

no?

Is this useful for satellites other than sentinel-1?

Most SAR mission uses different instrument modes with swath composed of many beams. For instance, ScanSAR technique is a quite used mode in many missions.

[fabricebrito]

@davidraleigh It's relevant for all SAR missions.

See pages 14-15 of https://www.asi.it/wp-content/uploads/2019/08/COSMO-SkyMed-Mission-and-Products-Description_rev3-2.pdf for CSK and this is just an example.

[fabricebrito]

Maybe just sar:beam_ids: ["IW1", "IW2"] (or sar:swatch_ids)? (I assume beams and swaths is the same, but I'm probably wrong?) The view:* fields can be provided in Item properties. How useful is it to know the relative burst and polarization per beam?

beams and swath ids cannot be seen as synonyms

[m-mohr]

Concerning the search, I do not see any limitation or complexity for filtering

@emmanuelmathot I'm not so sure. In theory yes, but in practice I still see implementations struggle to search bands or assets, so I don't think in practice you can currently easily search for the swath IDs in STAC API Filter implementations. Do you have different information? :-)

beams and swath ids cannot be seen as synonyms

@fabricebrito Okay, I tried to understand this through some googling but can't find a good resource that explains the differences. I
I'm asking because I'm looking into CEOS ARD and they ask me to provide beam IDs, but I'm not sure how to provide that for S1 for example, I thought it's swath IDs in the end. SAR is always so confusing ;-)

The other question is: If this is rather specific to one/some(?) SAR modes, should this go into a separate extension (e.g. ScanSAR)?

[fabricebrito]

@fabricebrito Okay, I tried to understand this through some googling but can't find a good resource that explains the differences. I I'm asking because I'm looking into CEOS ARD and they ask me to provide beam IDs, but I'm not sure how to provide that for S1 for example, I thought it's swath IDs in the end. SAR is always so confusing ;-)

The other question is: If this is rather specific to one/some(?) SAR modes, should this go into a separate extension (e.g. ScanSAR)?

beam ids are relevant for StripMap acquisitions so it comes down to how the satellite is programmed to acquire data over the areas. Sentinel-1 Stripmap for instance is typically (not exclusively) used over small islands.
My view is that at the sar extension level we need this information but this doesn't hamper having more advanced extensions for certain modes

[fabricebrito]

We need a Stripmap example as this is the original request

[m-mohr]

Please remove

[fabricebrito]

@emmanuelmathot I hadn't seen the S1-S4-GRD example, it contains the wrong beam id (s6 instead of s4), so just a minor fix and we're good to go

[m-mohr]

I want to check whether this could be re-used for the CEOS ARD work, but this could take a bit longer... just fyi

[fabricebrito]

I want to check whether this could be re-used for the CEOS ARD work, but this could take a bit longer... just fyi

will this create a delay in the approval?

[fabricebrito]

All good! Thanks @emmanuelmathot

[fabricebrito]

@m-mohr any updates on your check with CEOS ARD work?