Fix handling of the time coordinate in outputs of recipe_ref_fire.yml (#4538)

Co-authored-by: Manuel Schlund <32543114+schlunma@users.noreply.github.com>

Author: jlenh

doc/sphinx/source/recipes/recipe_ref_fire.rst

@@ -63,22 +63,26 @@ User settings in recipe
 
    *Required settings for script*
 
-   * var_order: list of climate drivers in the order corresponding to the one
+   * `var_order`: list of climate drivers in the order corresponding to the one
      specified in the corresponding file from the confire_param directory.
 
    *Optional settings for script*
 
-   * confire_param: path to the directory containing the required files to run
+   * `confire_param`: path to the directory containing the required files to run
      the ConFire model or Zenodo URL to retrieve files from a Zenodo archive.
      If custom files are used, the corresponding directory needs to be present
      inside the auxiliary data directory defined inside the user configuration.
      This defaults to the original Zenodo archive otherwise.
-   * remove_vpd_files: Removing or not the computed vapor pressure deficit files.
+   * `remove_vpd_files`: Removing or not the computed vapor pressure deficit files.
      It will only apply if the vapor pressure deficit is part of var_order.
-     This defaults to False.
-   * remove_confire_files: Removing or not the files produced during the ConFire
+     This defaults to `False`.
+   * `remove_confire_files`: Removing or not the files produced during the ConFire
      model evaluation.
-     This defaults to False.
+     This defaults to `False`.
+   * `keep_original_time`: Keeping or not the time coordinate (origin, calendar)
+     from the input. If not, then it defaults to: unit
+     `"days since 1850-01-01 00:00:00"` and calendar `"proleptic_gregorian"`.
+     This defaults to `False`.
 
    *Required settings for variables*
 

esmvaltool/diag_scripts/fire/diagnostic_run_confire.py

@@ -151,47 +151,57 @@ def _select_key_or_default(
 
 
 # /libs/iris_plus.py
-def _sort_time(
-    cube: iris.cube.Cube,
-    field: str,
-    filename: str,
-) -> iris.cube.Cube:
-    """Sort time dimension in the iris cube.
+def make_sort_time_callback(*, keep_original_time: bool):
+    """Create a _sort_time callback with configurable time coordinate."""
 
-    Parameters
-    ----------
-    cube: iris cube
-        Input cube.
-    field: str
-        Variable name in cube.
-    filename: str
-        Filename of cube.
+    def _sort_time(
+        cube: iris.cube.Cube,
+        field: str,
+        filename: str,
+    ) -> iris.cube.Cube:
+        """Sort time dimension in the iris cube.
 
-    Returns
-    -------
-    cube: iris cube
-        Cube with sorted and added time dimensions.
-    """
-    logger.debug("Sorting time for variable %s in cube %s", field, filename)
-
-    cube.coord("time").bounds = None
-    tcoord = cube.coord("time")
-    tcoord.units = cf_units.Unit(tcoord.units.origin, calendar="gregorian")
-    tcoord.convert_units("days since 1661-01-01 00:00:00")
-    tcoord.units = cf_units.Unit(
-        tcoord.units.origin,
-        calendar="proleptic_gregorian",
-    )
-    cube.remove_coord("time")
-    cube.add_dim_coord(tcoord, 0)
+        Parameters
+        ----------
+        cube: iris cube
+            Input cube.
+        field: str
+            Variable name in cube.
+        filename: str
+            Filename of cube.
 
-    if not cube.coords("year"):
-        iris.coord_categorisation.add_year(cube, "time")
+        Returns
+        -------
+        cube: iris cube
+            Cube with sorted and added time dimensions.
+        """
+        logger.debug(
+            "Sorting time for variable %s in cube %s",
+            field,
+            filename,
+        )
+
+        cube.coord("time").bounds = None
+        tcoord = cube.coord("time")
+        if not keep_original_time:
+            tcoord.units = cf_units.Unit(
+                "days since 1850-01-01 00:00:00",
+                calendar="proleptic_gregorian",
+            )
+        cube.remove_coord("time")
+        cube.add_dim_coord(tcoord, 0)
+
+        if not cube.coords("month"):
+            iris.coord_categorisation.add_month_number(
+                cube,
+                "time",
+                name="month",
+            )
 
-    if not cube.coords("month"):
-        iris.coord_categorisation.add_month_number(cube, "time", name="month")
+        if not cube.coords("year"):
+            iris.coord_categorisation.add_year(cube, "time")
 
-    return cube
+    return _sort_time
 
 
 def _insert_data_into_cube(
@@ -412,6 +422,7 @@ def _read_variable_from_netcdf(
     make_flat: bool = False,
     return_time_points: bool = False,
     return_extent: bool = False,
+    keep_original_time: bool,
     verbose: bool = True,
 ) -> iris.cube.Cube:
     """Read data from a netCDF file.
@@ -440,6 +451,9 @@ def _read_variable_from_netcdf(
     time_series: list
         List comtaining range of years. If making flat and
         returned a time series, checks if that time series contains year.
+    keep_original_time: bool
+        If True, keep time coordinate from inputs, otherwise replace by
+        Unit("days since 1850-01-01 00:00:00", calendar="proleptic_gregorian")
 
     Returns
     -------
@@ -457,15 +471,19 @@ def _read_variable_from_netcdf(
     if isinstance(filename, str):
         dataset = iris.load_raw(
             Path(directory) / filename,
-            callback=_sort_time,
+            callback=make_sort_time_callback(
+                keep_original_time=keep_original_time,
+            ),
         )
     else:
         # Fallback for CMIP7 data for tasmax
         var = "tas" if "tasmax/tas_tmaxavg" in filename[0] else filename[1]
         dataset = iris.load_raw(
             Path(directory) / filename[0],
             var,
-            callback=_sort_time,
+            callback=make_sort_time_callback(
+                keep_original_time=keep_original_time,
+            ),
         )
     dataset = dataset[0]
 
@@ -565,6 +583,7 @@ def _read_all_data_from_netcdf(
     add_1s_columne: bool = False,
     x_normalise01: bool = False,
     check_mask: bool = True,
+    keep_original_time: bool,
     **kw: dict,
 ) -> tuple[np.array]:
     """Read data from netCDF files.
@@ -596,6 +615,9 @@ def _read_all_data_from_netcdf(
         you dont want. This could be different in some circumstances.
     frac_random_sample: int
         fraction of data to be returned
+    keep_original_time: bool
+        If True, keep time coordinate from inputs, otherwise replace by
+        Unit("days since 1850-01-01 00:00:00", calendar="proleptic_gregorian")
     args: tuple
         See _read_variable_from_netcdf comments.
     kw: dict
@@ -615,6 +637,7 @@ def _read_all_data_from_netcdf(
         return_time_points=True,
         return_extent=True,
         verbose=False,
+        keep_original_time=keep_original_time,
         **kw,
     )
 
@@ -626,6 +649,7 @@ def _read_all_data_from_netcdf(
             time_points=time_points,
             extent=extent,
             verbose=False,
+            keep_original_time=keep_original_time,
             **kw,
         )
 
@@ -642,6 +666,7 @@ def _read_all_data_from_netcdf(
             make_flat=True,
             time_points=time_points,
             extent=extent,
+            keep_original_time=keep_original_time,
             **kw,
         )
 
@@ -880,6 +905,7 @@ def _get_parameters(config: dict) -> tuple:
     """
     work_dir = config["work_dir"]
     confire_param = config["confire_param_dir"]
+    keep_original_time = config["keep_original_time"]
     # **Define Paths for Parameter Files  and for outputs**
     output_dir = work_dir + "/ConFire_outputs/"
     # Parameter files (traces, scalers, and other model parameters)
@@ -901,12 +927,14 @@ def _get_parameters(config: dict) -> tuple:
         x_filename_list=nc_files,
         scalers=scalers,
         directory=nc_dir,
+        keep_original_time=keep_original_time,
     )
     # Load a sample cube (used for inserting data)
     eg_cube = _read_variable_from_netcdf(
         nc_files[0],
         directory=nc_dir,
         verbose=False,
+        keep_original_time=keep_original_time,
     )
     # **Extract Model Parameters**
     logger.info("Loading ConFire model parameters...")

esmvaltool/diag_scripts/fire/fire_diagnostic.py

@@ -346,6 +346,9 @@ def main(cfg: dict) -> None:
     # default is False
     if "remove_confire_files" not in cfg:
         cfg["remove_confire_files"] = False
+    # Keeping or not the original time coordinates and calendar form the inputs
+    # Default is False
+    cfg.setdefault("keep_original_time", False)
 
     for model_dataset, group in datasets.items():
         # 'model_dataset' is the name of the model dataset.