cioos-siooc / cioos-siooc_data_transform

6 stars 5 forks source link

ODF Time Variable Timezone? #93

Open JessyBarrette opened 3 years ago

JessyBarrette commented 3 years ago

Issue

odf_parser for the moment does not handle any timezone information. So far, the only timezone information observed within the ODF files is sometimes available within the UNITS of the ODF SYTM variables.

All the test ODFs (~2500 files) also used the standard time format: dd-MMM-yyyy HH:mm:ss (ex: 31-DEC-2019 23:50:00.00) with no timezone information attached. This format is used within the data and the header.

The only timezone retrieved so far are:

As of now

We're not handling any timezone and just output a warning if the time variable has a unit different than the ones listed above. We could potentially replace the warning by an error to be more conservative.

Solution?

What should be done in the future? Is it an actual issue? What to assume:

  1. Time with no timezone information is by default in UTC. This follow CF convention, maybe that's what ODF is also.
  2. Handle no timezone data differently? Leave a default timezone within the config file
  3. Report the time zone just within the future ERDDAP dataset
guillotp commented 3 years ago

Assume it is UTC.

JessyBarrette commented 3 years ago

@guillotp always?

guillotp commented 3 years ago

For QO and Amundsen Science: yes. For MLI: I am waiting for a confirmation. For BIO: no idea.

guillotp commented 3 years ago

MLI is also always using UTC. But I guess QO and MLI must fix this lack a future ODF version.

JessyBarrette commented 3 years ago

@guillotp

But I guess QO and MLI must fix this lack a future ODF version.

You mean leaving UTC within the units. I think it's alright, it is still compliant with the CF standards where if no time zone is provided it defaults to UTC.

See note https://cfconventions.org/Data/cf-conventions/cf-conventions-1.8/cf-conventions.html#time-coordinate

guillotp commented 3 years ago

I did not know. Thank you.