phase_method as part of ROProf necessary?

[marq]

Hi,

Armin discovered that we have a 'phase_method' in the documentation of the ROprof structure, but none in the actual implementation; we do have pod_method, bangle_method and refrac_method, though. In both the ASCII and the netCDF versions, there isn't any phase_method data or attribute either. I wonder if having one would make sense; one could specify things like zero / single / double differencing. Adding it wouldn't be difficult for the structure itself or for the netCDF; but it would break the current ASCII format as we would have to add a new line or something.

From that point of view, it might be simpler to simply drop the phase method and ask people to put the information into the bangle_method.

Please let me know what you prefer,

Chris.

[Dave Offiler]

I noticed this anomoly when checking over Torsten's CHAMP file in ROPP v1 text files. It would be desirable to include this if only for completemness. Including it in the ROPP-type should be simple enough, and presumably not much on an issue to include the info in the netCDF files. The text file is the main issue, and while I would in one sense like to freeze the format where it is (as GFZ are using it for real), it's not formally released, so in principle subject to change anyway. My feeling is that all the *_method meta-info should be treated consistently throughout. [The text-format spec is not yet described in the ROPP User Guide; the seperate ROPP-format document is anyway out of date with respect to the code...]

[marq]

Ok,

as discussed offline with Dave, I have implemented the change in the following way (in [794]):

• The ROprof structure got a new element (phase_method).
• netCDF-based data files carry the iformation in a global attribute, also named phase_method.
• Text format files are not affected; their format doesn't change to maintain compatibility with files already delivered by GFZ.

The change in the set of global netCDF attributes means that all netCDF data files used for any kind of testing in the software distribution itself or in the test suite need to be modified to carry this attribute, or tests will fail. I'll do that for the data files in the source code; someone else will have to address the test suite. I'll discuss this offline with Dave and Axel to find out what's the easiest way to achieve this before closing this ticket.

[Dave Offiler]

Since 'phase_method' has been implemented and (presumably) all files used for testing have this attribute, it should be safe to close this Ticket (re-open it if any issues remain that haven't been noted).

[(none)]

Milestone 0.9 deleted