Query SOAR metadata

[ebuchlin]

Describe the feature

My understanding is that sunpy-soar currently only supports queries by instrument / time / level / product, as this is basically what is available in the SOAR web query form and in the v_sc_data_item and v_ll_data_item tables. However, the user should also be able to do queries with different metadata (other Fido attributes).

Proposed solution

The list of all tables and their columns is available from SOAR with TAP. I attach a human-readable version (tree by schema / table / column), generated by XSLT with this XSL stylesheet.

This shows that more complete metadata are available in SOAR, in instrument-specific tables, e.g. v_spi_sc_fits. Example query: http://soar.esac.esa.int/soar-sl-tap/tap//sync?REQUEST=doQuery&LANG=ADQL&FORMAT=json&QUERY=SELECT+TOP+10+%2A+FROM+v_spi_sc_fits

Fido attributes should be linked to columns in these different instrument-specific tables. For a query with multiple instruments, multiple tables should be queried... (or should this not be supported?). LL files metadata can also be queried, from still different tables.

[ebuchlin]

There is a draft documentation for the tables, views, and columns of the SOAR TAP interface: https://www.cosmos.esa.int/web/soar/tables-views-and-columns

[ebuchlin]

There are now new columns soop_name and soop_type available in the SOAR TAP interface.

[wtbarnes]

There are now new columns soop_name and soop_type available in the SOAR TAP interface.

Is this covered by #84?

[ebuchlin]

There are now new columns soop_name and soop_type available in the SOAR TAP interface.
Is this covered by #84?

For queries by SOOP name it seems that #84 covers it, yes.

[wtbarnes]

@ebuchlin sorry for the lack of traffic on this issue. I definitely agree we should be supporting more complex queries against the SOAR through Fido, but I'm a bit confused as to the scope. Looking at the docs you linked above, it is not quite clear to me what attributes should be supported through the attrs interface. Could you provide an example of what a Fido query would look like with these additional metadata?

The example from @hayesla in #66 makes it a bit more clear to me, but again the issue is what subset of that metadata we should support. I don't think it is practical to try and translate each bit of SOAR metadata to a Fido attribute. However, maybe there could be some sort of interface to specifying these filters as strings, similar to what we allow with JSOC keywords in the sunpy.net.jsoc.attrs.

[ebuchlin]

This is a generic issue meant to tell that there were more possibilities with the SOAR TAP interface than the ones initially used by sunpy-soar (the details of the TAP interface were undocumented at that time). Now that we have some documentation and that queries by SOOP name have been implemented, we can be more specific about the potentially other useful attributes, starting from existing sunpy.net ones:

• Detector: from the v_<instrument>_<ll/sc>_fits tables, column detector. Partially overlaps the use cases for a.soar.Product.
• Wavelength: from the v_<instrument>_<ll/sc>_fits tables, column wavelength
  • SPICE windows wavelengths (one range per window) are not all accessible through SOAR, I think that only the first window is, unless the full list is in the undocumented v_<instrument>_<sc/ll>_extension_fits tables for <instrument> = SPICE (just a guess).
  • STIX rather has energy bands
• Resolution: for AIA and HMI, this is a factor from the highest resolution. There is some information in the cdelt[n], total_binning_factor and binning_factor columns (in one row per dimension), but not sure how to combine this into something meaningful, and consistent with the existing meaning for AIA and HMI. Also, should this be limited to spatial resolution?
• Phyobs: not in SOAR; could be deduced from a.soar.Product?
• Extent (as in sunpy.net.vso.attrs) could be useful, but the meaning should be clarified (some overlap with FOV?) and it might be difficult to implement.

An issue is that v_<instrument>_<ll/sc>_fits is actually multiple tables, one per instrument and per data type (low-latency or science), and that there must then be join operations with the v_<ll/sc>_data_item tables.

In case one would like to have access to previous versions of the files (instead of only the latest version), the v_<ll/sc>_repository_file tables would also have to be considered.

For a start, we can of course ignore previous versions of files, ignore low-latency observations, and prioritize attributes in the above list. The efforts should also be balanced with those put on access to Solar Orbiter data through VSO as data provider.

[ebuchlin]

For complex SOAR TAP queries, here is a tutorial on TAP queries that we did at IAS; it could provide ideas for how to do some of the queries we would like to be doable using Fido.

[nabobalis]

@ebuchlin we are going to add this as a GSoC project and I have a really rough draft here: https://github.com/OpenAstronomy/openastronomy.github.io/pull/350/files#diff-03a99800468bb348b3741103deee0d442348ced2997c4a20c1aa6479cd7729e9

If you had time could you review it and would you be willing to help with the project in an advisory capacity?

[ebuchlin]

I have added a small comment to the GSoC project, and yes I am willing to help.

[nabobalis]

Thanks!

[MetaphorC]

Hey! This issue is part of the GSoC projects. I would like to work on it, and with the organisation in general. I am new to working with open-source projects, but I will try my best to help. Where should I start?