Failure of applying masks to OBS datasets

[bascrezee]

When trying to apply mask_landsea to an Observational dataset, it fails.

datasets:
  - {dataset: CDS-SATELLITE-SOIL-MOISTURE, type: sat, project: OBS, mip: Lmon, version: COMBINED-TCDR-v201812.0.0, tier: 3, start_year: 2017,  end_year: 2018}

preprocessors:
  pp_test_mask:
    mask_landsea:
      mask_out: sea 
      always_use_ne_mask: True

Why is _get_correct_fx_file called for OBS datasets? This shouldn't happen, right?

  File "/ESMValCore/esmvalcore/_recipe.py", line 873, in __init__
    self.tasks = self.initialize_tasks() if initialize_tasks else None
  File "/ESMValCore/esmvalcore/_recipe.py", line 1220, in initialize_tasks
    task_name=task_name,
  File "/ESMValCore/esmvalcore/_recipe.py", line 844, in _get_preprocessor_task
    name=task_name,
  File "/ESMValCore/esmvalcore/_recipe.py", line 712, in _get_single_preprocessor_task
    config_user=config_user,
  File "/ESMValCore/esmvalcore/_recipe.py", line 662, in _get_preprocessor_products
    config_user=config_user)
  File "/ESMValCore/esmvalcore/_recipe.py", line 429, in _update_fx_settings
    'sftlf': _get_correct_fx_file(variable, 'sftlf', config_user),
  File "/ESMValCore/esmvalcore/_recipe.py", line 402, in _get_correct_fx_file
    fx_files = get_input_filelist(variable=fx_var,
UnboundLocalError: local variable 'fx_var' referenced before assignment
2019-10-07 11:55:20,619 UTC [21865] INFO    If you suspect this is a bug or need help, please open an issue on https://github.com/ESMValGroup/ESMValTool/issues and attach the run/recipe_*.yml and run/main_log_debug.txt files from the output directory.

[valeriupredoi]

a bug! But not really an accidental one since some OBS's actually have fx files associated (@mattiarighi pointed me to ERA-Interim I believe) - this is something that we've not actually agreed on - ie what to do when fx data files are not found. @bouweandela had the very Dutch opinion to stop the code and tell the user there are none 😁 My take is to let it go in certain cases (like this one here)

[bascrezee]

a bug!

Yes, I am especially surprised that it occurs when I explicitly tell to use the ne_mask (i.e. the shapefiles).

[valeriupredoi]

pls try out #295 - it's a bit of a hack but nothing better before we reach an agreement how to treat the OBS-fx relation 😁

[valeriupredoi]

a bug!

Yes, I am especially surprised that it occurs when I explicitly tell to use the ne_mask (i.e. the shapefiles).

that request is a lower level one that gets superseded by the calls in _recipe.py that invariably fail since there is no if project == OBS clause. Bleh 😁

[bascrezee]

pls try out #295 - it's a bit of a hack but nothing better before we reach an agreement how to treat the OBS-fx relation

This fixes the problem indeed for my case. Thanks !

[valeriupredoi]

#296 supersedes this so can we close pls? 🍺