LTDP SAFE

GMV S.A.U.

The metadata currently used by SAFE is no not standard and would be better exploitable in any interoperability framework if using standardisation.

The HMA FO (i.e. Earth Observation profile of Observation and Measurement) and ISO element Lineage defined in ISO 19115-2 shall be considered for SAFE metadata alignment.

HMA FO is an OGC standard adopted in the HMA framework and ISO element Lineage defined in ISO 19115-2 is an internationally recognised standard for representation of product history and could be used to cover the relevant part of SAFE metadata.

In addition, the following fields shall be analysed to be added/extended and documented as standard metadata in SAFE products:
- Satellite Phase
- WRS (World Reference System)
- Repeat cycle
- Last update of the metadata
- Organisation responsible for product management

During the review of the HMA work some criticism was raised on the approach initially taken for the product metadata. To make a long story short, we (ESA and partners) find very difficult to extend the ISO 19115 approach to product metadata due to the our requirement to define mission specific attributes. So generally ISO 19115 is considered only suitable to the "collection" metadata and not for the "EO product" metadata.

If 19115 is not considered sufficient then a so-called "model driven approach" shall be used to define the metadata and extend it. This means defining a conceptual model - ideally using UML - and then possibly deriving the XML encoding from it. Two communities we consider as reference: INSPIRE - Guidelines for the use of Observations & Measurements and Sensor Web Enablement-related standards in INSPIRE Annex II and III data specification development (D2.9_v1.0) and the Open Geospatial Consortium. They are active in the geospatial domain. They have identified in the Observation and Measurements the model to be used when (e.g.) when provenance and quality information should be provided with the data.
The Observation and Measurement is an ISO and OGC standard (ISO 19156, OGC 10-004r3).
The extension for the Earth Observation product metadata is being published. A preview is in the attached file.

In attachment you can find some documents providing metadata encoding examples produced in the GSCDA context (GSC_DAR_ICD, GSC_DAR_OPT, GSC_DAR_ATM). They are based on the old version of the HMA metadata but the changes from the old to the new version are not too broad so this examples can help you to better understand the metadata model proposed.
In addition you can find an example of encoding of the HMA FO metadata standard (EO O&M Meris example).

You can find the preview of the Earth Observation Metadata Profile of Observation and Measurement in the attached file.

HMA metadata model has been developed for products and not auxiliary files. It should be verified if the HMA model can be tailored for auxiliary files or a specific model should be introduced for them (possibly based on Observation and Measurement as the HMA model). Anyway the difference between products and auxiliary files should be taken into account in SAFE data model.

Mai-Eli Johansen

The KSAT HMA experience is restricted to the HMA-EO standard. We have implemented parts of the ordering standard in our on-line ordering system. We are also producing metadata for some satellites that it compliant with the HMA-EO standard. So far KSAT is content with the HMA parts that we have used.

GMV S.A.U.

The following technical note has been developed by GMV including a metadata alignment between OGC/INSPIRE/ISO/WCS. This information will conform the final metadata model to update the SAFE Core Specification document.

We think that this could be of intereset for the community. All your comments are appreciated.

Adrian Sanz (GMV)
LTDP SAFE Project Manager

Stephan Zinke

Comments from Stephan Zinke (EUMETSAT) attached.
Regards
--
Stephan

GMV S.A.U.

Hi Stephan,

I answer to your points below.

> Section 4: Operational vs. LDTP conformance class...

I share your view. There is no harm in adding those metadata as optional for the Operational class. We can promote them as optional to the SAFE Abstract Data class, and just restrict them to mandatory for the SAFE LTDP class.
Concerning the representation information, in principle we just say that it is not required in the operational class, but we don’t forbid its use. However, I think that the figures of the information models should be updated in order to reflect this properly, because at the moment the representation information doesn’t even appear as optional for the operational classes.

> General: (entered as ESA-PANEL-118)...

The approach of ISO/INSPIRE extensions to the OGC EOP was chosen at the beginning taking into account that the only base standard required was the OGC EOP, hence only focusing on EO Products. Anyway, the issue of collections is important, so I think that it should be discussed with the rest to see their opinion. On our side we are open to apply the needed changes if it is agreed.

I understand the point about the separation between EO Product metadata and Collection metadata, but there is something that is not clear for me. The O&M model seems to cover already the possibility to reference/include ISO 19139 metadata through the OM_CommonProperties/metadata/gmd:MD_Metadata. Then the OGC EOP model allows this as well, since it just extends the O&M model. Therefore, the EO product metadata and collection metadata don't seem to be so clearly separated, unless this mechanism only allows to make reference to external gmd metadata. Is this the case? I am just trying to understand why breaking the aforementioned separation would be a problem when it seems that the OGC EOP model already breaks it, but I am sure that I am missing or misunderstanding something here.

I understand the following of the approach proposed (please confirm if it is right):
1. We should take the ISO and INSPIRE metadata of the extensions out to an independent schema (not an extension) and change them to follow ISO 19139 (or ISO19115-2 and ISO19139-2)
2. We should remain with the main XML metadata file compliant to the OGC EO Profile, and define an extension for the rest of metadata (particular SAFE own metadata, QA4EO metadata)
3. We should define a reference in the EO Product package to the EO Collection Metadata file, choosing one of the options proposed (I think that both options would fit well in the XFDU structure, personally I prefer the first one, because it keeps each file at its place and it reduces redundancy, but this would have to be agreed).

Regards,

Héctor.

Stephan Zinke

Dear Hector,
> General: (entered as ESA-PANEL-118)...
The approach you described is what we would propose as this follows exactly what was done in HMA: separation of EO product metadata (defined in EO O&M) and EO collection metadata (defined in ISO 1939).
We checked again the HMA specs. They define the linkage via the EO O&M parentIdentifier attribute!
Here the details:
The EO O&M eop:EarthObservationMetadata type includes an element "parentIdentifier" and this one links to the identifier of the collection metadata (the ISO19139 document). The XPath to this identifier in the ISO19139 document is as follows: /gmd:MD_Metadata/gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:identifier

So it would be best to have an EO O&M XML file which is linked by parentIdentifier to the ISO19139 describing the collectionMetadata.

Regards
--
Stephan

GMV S.A.U.

Dear all,

Some questions concerning collections:

For a set of EO products which belong to the same EO collection, can it happen that the collection metadata is updated at any time?
If yes, should all the EO products of the collection be aware of those changes?

I presume that the "yes-yes" answers have perfect sense, but I would like to have your confirmation, since it is an important issue for the design of the collections in SAFE 2.0.

Thanks.
Best regards,

Héctor.

> Dear Hector,
> > General: (entered as ESA-PANEL-118)...
> The approach you described is what we would propose as this follows exactly what was done in HMA: separation of EO product metadata (defined in EO O&M) and EO collection metadata (defined in ISO 1939).
> We checked again the HMA specs. They define the linkage via the EO O&M parentIdentifier attribute!
> Here the details:
> The EO O&M eop:EarthObservationMetadata type includes an element "parentIdentifier" and this one links to the identifier of the collection metadata (the ISO19139 document). The XPath to this identifier in the ISO19139 document is as follows: /gmd:MD_Metadata/gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:identifier
>
> So it would be best to have an EO O&M XML file which is linked by parentIdentifier to the ISO19139 describing the collectionMetadata.
>
> Regards
> --
> Stephan

GMV S.A.U.

The following comments have been done offline by Diego Lozano (Deimos) after the PDR-C collocation meeting:

In the SAFE volume 2 (PGSI-GSEG-EOPG-FS-05-0002), 8.4. SAFE QA4EO Extension Types, it has already defined some safe metadata for Quality. There are already other projects defining quality metadata for GML EO profile, for instance ngEO (metadataUpdate reports), GMES QII reports, … It may be worth to have a look to them and reused some of the metadata if applicable.