Update datetime handling to use dateTime with correct type and fill behavior and simplify environment setup

[amstokely]

This PR updates the goes_abi converter in the obs2ioda library to improve compliance with the IODA v3 format and address several compilation issues.

Changes

• Corrected dateTime representation:
  The dateTime variable is now written as an int64 array representing epoch seconds for each observation, instead of as a string. This aligns with the expected format for IODA v3 files.

• Added dateTime units attribute:
  The NetCDF attribute for dateTime units was added ("seconds since 1970-01-01T00:00:00Z"), as required by the IODA v3 specification.

• Compilation fixes:
  Resolved various minor compilation issues, including Fortran unit tests that previously failed to compile with the Intel Fortran compiler.

• Simplified output file naming:
  The output file name now includes only the satellite ID and the date/time with minute precision. This reduces filename verbosity, making the files more straightforward to work with in scripts and data workflows.

• Added a conda environment yaml for building the Python testing environment
  The conda environment file makes creating a conda environment for running the test suite seamlessly.

[jim-p-w]

It looks like this only addresses the goes_abi converter. What about the obs2ioda converter?
I see src/ncio_mod.f90 has this:
346: status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")

[amstokely]

It looks like this only addresses the goes_abi converter. What about the obs2ioda converter? I see src/ncio_mod.f90 has this: 346: status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")

While ncio_mod (and possibly gnssro_mod) define both datetime and dateTime variables, it's the dateTime variable that IODA recognizes and uses. IODA gives precedence to dateTime and ignores datetime if both are present. In a future PR, we should remove the unused datetime variable from ncio_mod, but for now it's harmless since it's ignored by IODA.

[jim-p-w]

It looks like this only addresses the goes_abi converter. What about the obs2ioda converter? I see src/ncio_mod.f90 has this: 346: status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")

While ncio_mod (and possibly gnssro_mod) define both datetime and dateTime variables, it's the dateTime variable that IODA recognizes and uses. IODA gives precedence to dateTime and ignores datetime if both are present. In a future PR, we should remove the unused datetime variable from ncio_mod, but for now it's harmless since it's ignored by IODA.

I don't see where dateTime is getting written in obs2ioda. Not being that familiar w/ the obs2ioda code,
I tried a case insensitive scan of the fortran code for 'cdfput.*datetime' and saw:

src/gnssro_mod.f90
460:      call check(netcdfPutAtt(ncid, 'min_datetime', gnssro_data%datetime(idx_min_time)))
461:      call check(netcdfPutAtt(ncid, 'max_datetime', gnssro_data%datetime(idx_max_time)))

src/ncio_mod.f90
150:      status = netcdfPutAtt(netcdfID, "min_datetime", xdata(ityp, itim)%min_datetime)
151:      status = netcdfPutAtt(netcdfID, "max_datetime", xdata(ityp, itim)%max_datetime)
346:               status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")

src/goes_abi_converter_mod.f90
147:        call check(netcdfPutVar(ncid, 'dateTime', datetime, 'MetaData'))

That isn't conclusive though. Can you point me to where obs2ioda writes dateTime?

[amstokely]

It looks like this only addresses the goes_abi converter. What about the obs2ioda converter? I see src/ncio_mod.f90 has this: 346: status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")

While ncio_mod (and possibly gnssro_mod) define both datetime and dateTime variables, it's the dateTime variable that IODA recognizes and uses. IODA gives precedence to dateTime and ignores datetime if both are present. In a future PR, we should remove the unused datetime variable from ncio_mod, but for now it's harmless since it's ignored by IODA.

I don't see where dateTime is getting written in obs2ioda. Not being that familiar w/ the obs2ioda code,

I tried a case insensitive scan of the fortran code for 'cdfput.*datetime' and saw:

src/gnssro_mod.f90

460:      call check(netcdfPutAtt(ncid, 'min_datetime', gnssro_data%datetime(idx_min_time)))

461:      call check(netcdfPutAtt(ncid, 'max_datetime', gnssro_data%datetime(idx_max_time)))



src/ncio_mod.f90

150:      status = netcdfPutAtt(netcdfID, "min_datetime", xdata(ityp, itim)%min_datetime)

151:      status = netcdfPutAtt(netcdfID, "max_datetime", xdata(ityp, itim)%max_datetime)

346:               status = netcdfPutVar(netcdfID, ncname, str_ndatetime, "MetaData")



src/goes_abi_converter_mod.f90

147:        call check(netcdfPutVar(ncid, 'dateTime', datetime, 'MetaData'))

That isn't conclusive though. Can you point me to where obs2ioda writes dateTime?

@jim-p-w dateTime is in the name_var_info array, which is defined in define_mod on line 127. ncio_mod uses this arrays to define variables. The validation test suite verifies that obs2ioda does write the dateTime variable for all file types in the MetaData group.

[ibanos90]

This branch successfully creates a GOES-ABI file that can be assimilated into our MPAS-JEDI applications. The output name also reflects better the extension and naming that follows closely other data types. Thanks for addressing all comments and fixing all issues, @amstokely!
For future reference, I tested compiling with GNU and INTEL and produces identical results with flags currently in use. Each file was tested within an MPAS-JEDI 3denvar run.

[liujake]

There are some conflicts need to be resolved before merge.

[amstokely]

There are some conflicts need to be resolved before merge.

@liujake I just rebased off main and all conflicts are resolved.

[liujake]

Looks like conflicts are resolved. Better merge this today with another approval.

[liujake]

Can someone having admin permission of this repo remove the restriction for two approvals before merging? I think it is unnecessary to have this restriction. One approval is Ok to me.

[ibanos90]

Can someone having admin permission of this repo remove the restriction for two approvals before merging? I think it is unnecessary to have this restriction. One approval is Ok to me.

Good point! I just changed it to 1