Addition of new diagnostics to lfric_atm metadata

[SamanthaPullen]

I have a couple of questions about procedure for adding to the lfric_atm metadata: https://code.metoffice.gov.uk/trac/lfric_apps/browser/main/trunk/applications/lfric_atm/metadata

For the new land surface data assimilation system we are developing (JEDI-based JOPA-land and JADA-land) there are some additional land surface diagnostic variables that we need to output from the model and will need defining in the lfric_atm metadata.

The new variables are:
soil__soil_moist_wilt
soil__soil_moist_crit
soil__soil_moist_sat
soil__soil_moisture_top_layer
surface__urban_tile
surface__land_ice_tile

I plan to add field definitions for these variables in https://code.metoffice.gov.uk/trac/lfric_apps/browser/main/trunk/applications/lfric_atm/metadata/field_def_diags.xml
The bottom 3 also need a zoom axis defining in https://code.metoffice.gov.uk/trac/lfric_apps/browser/main/trunk/applications/lfric_atm/metadata/axis_def_main.xml

Questions:

1. Should I also be adding the new variables to lfric_dictionary.xml in main/trunk/applications/lfric_atm/metadata – LFRic Apps?
2. Is there any particular order to be followed for adding field definitions to field_def_diags.xml or can I just add my new variables onto the bottom of the soil__ and surface__ sections?

[stevemullerworth]

1. No. Fields should be defined only once, and the field_def_diags file is the right place for defining diagnostics. The lfric_dictionary.xml is used for holding "prognostic" fields: fields that the model may need to initialise at the start of a run and retain in memory throughout. Such "prognostic" fields are those that are typically initialised in the create_physics_prognostics routine.
2. I don't believe any order is mandated, so insert them where best, I would say. I guess one might group them with similarly-named existing diagnostics, but I don't believe anyone has defined any rules.

[SamanthaPullen]

Great, thanks for the answers @stevemullerworth