#886: Add cell measures to data based on CMIP7 style cell_measures file

mip_convert/mip_convert/new_variable.py

@@ -337,7 +337,7 @@ def ordered_coords(self):
 
         return self._ordered_coords
 
-    def process(self):
+    def process(self, saver=None, cell_measures_config=None):
         """Process the data.
 
         The units of the data of the |MIP requested variable| are the
@@ -367,6 +367,19 @@ def process(self):
             self._update_time_units()
         if hasattr(self.model_to_mip_mapping, 'valid_min'):
             self._apply_valid_min_correction()
+        if saver is not None and cell_measures_config is not None:
+            self.apply_cell_measures(saver, cell_measures_config)
+
+    def apply_cell_measures(self, saver, cell_measures_config):
+        """Apply cell measures to the CMOR variable if not already applied.        Parameters
+        ----------
+        saver:
+            The CMOR outputter for this variable.
+        cell_measures_config: tuple
+            Arguments forwarded to cmor_wrapper.apply_cell_measures,
+            in the form (mip_era, mip_table_dir, realm, variable, frequency, region).
+        """
+        saver.apply_cell_measures(self, *cell_measures_config)
 
     def _remove_alevhalf_bounds(self):
         """

mip_convert/mip_convert/requested_variables.py

@@ -128,8 +128,17 @@ def produce_mip_requested_variable(
     # Process the data by performing the appropriate 'model to MIP mapping', then save the 'MIP output variable'
     # to an 'output netCDF file'.
     period = user_config.slicing.get(stream_id, 'year')
+
+    cell_measures_config = (
+        user_config.mip_era,
+        user_config.inpath,
+        mip_table.id,
+        variable_name,
+        frequency,
+        user_config.global_attributes.get('region', ''))
+
     for time_slice in variable.slices_over(period):
-        time_slice.process()
+        time_slice.process(saver, cell_measures_config)
         logger.debug('MIP output variable contains: {}'.format(time_slice.info))
         save(time_slice, saver)
 

mip_convert/mip_convert/save/cmor/cmor_outputter.py

@@ -38,6 +38,7 @@
 from mip_convert import mip_parser
 import mip_convert.common
 from mip_convert.common import RelativePathChecker
+from mip_convert.save import create_cmor_variable
 from mip_convert.save.mip_config import MipTableFactory
 
 
@@ -301,6 +302,17 @@ def _optional_kwargs(self, variable):
             kwargs['comment'] = self._name_space.namespace_stamp(variable.comment)
         return kwargs
 
+    def apply_cell_measures(self, mip_output_variable, *cell_measures_config):
+        """Apply cell measures to this CMOR variable.
+
+        Ensures the CMOR variable ID exists (creating it if needed), then
+        sets the cell_measures attribute via the CMOR wrapper.
+        """
+        if self.varid is None:
+            cmor_variable = create_cmor_variable(mip_output_variable)
+            self._getVarId(cmor_variable)
+            self.cmor.apply_cell_measures(*cell_measures_config, self.varid)
+
     def _close_file(self):
         self.cmor.close(self.varid, preserve=True)
 

mip_convert/mip_convert/save/cmor/cmor_wrapper.py

@@ -1,6 +1,8 @@
 # (C) British Crown Copyright 2009-2025, Met Office.
 # Please see LICENSE.md for license details.
 from collections import OrderedDict
+import os
+import json
 
 import cmor
 
@@ -115,3 +117,38 @@ def zfactor(self, *args, **kwargs):
     def set_frequency(self, frequency, **kwargs):
         self._debug_on_args('frequency', [frequency], kwargs)
         cmor.cmor.set_cur_dataset_attribute('frequency', frequency)
+
+    def apply_cell_measures(self, mip_era, mip_table_dir, realm, variable, frequency, region, variable_id):
+        """
+        Set the cell_measures attribute on a CMOR variable using a lookup in
+        ``{mip_era}_cell_measures.json`` from the MIP tables directory. Returns
+        silently if the file does not exist.
+
+        The lookup key has the form ``{realm}.{root_label}.{branding}.{frequency}.{region}``
+        (variable name split on ``_``), so entries in the JSON must follow the
+        convention used for CMIP7.
+        """
+        self._debug_on_args('apply_cell_measures', [mip_era, mip_table_dir, realm, variable, frequency, region, variable_id], {})
+
+        cell_measures_file = os.path.join(mip_table_dir, f'{mip_era}_cell_measures.json')
+
+        if os.path.exists(cell_measures_file):
+            with open(cell_measures_file) as fh:
+                cell_measures = json.load(fh)
+            if 'cell_measures' not in cell_measures:
+                self.logger.debug(f'"cell_measures" key not found in {cell_measures_file}')
+                return
+            cell_measures = cell_measures['cell_measures']
+
+            root_label, branding = variable.split('_')
+            key = f'{realm}.{root_label}.{branding}.{frequency}.{region}'
+            if key in cell_measures:
+                retval = cmor.cmor.set_variable_attribute(
+                    variable_id,
+                    'cell_measures',
+                    'c',
+                    cell_measures[key])
+                if retval != 0:
+                    self.logger.debug('cell_measures assignment failed. Check cmor log file for details')
+        else:
+            self.logger.debug(f'Cell_measures file "{cell_measures_file}" not found. Continuing')