Unit test data for um2nc

[truth-quark]

The um2nc code requires some small datasets for implementing unit testing.

Ideally, test input files can be processed in a few seconds, with the goal of keeping the full test suite run time down.

Questions:

• Do any real datasets exist for use?
• Will synthetically created datasets be required?
• How large are the input files?

[atteggiani]

The The files are not too large

• Do any real datasets exist for use?
• Will synthetically created datasets be required?

Real datasets exist, but depending on what you want to test, it might get difficult to have exactly the file that you want. Therefore, I think preparing datasets "synthetically" is suggested (and I am sure we can re-use some datasets for multiple tests).

• How large are the input files?

In general, it depends on the data shape (lat, lon) and number of fields. UM files can be relatively small (if we make them small), say around 50-150 KB. But I think for some specific tests we might require more fields, which means the files can get up to a few MB.