Updates following review comments

Author: gavinevans

improver/calibration/ensemble_calibration.py

@@ -862,14 +862,15 @@ def _add_predictor_coords(
 
         Returns:
             A cube with the predictor_index and predictor_name
-            coordinates added.
+            coordinates added. Single value dimension coordinates
+            are converted to non-dimension coordinates.
         """
         template_cubes = iris.cube.CubeList()
         fp_names = []
         for index, fp in enumerate(forecast_predictors):
             template_cube_copy = template_cube.copy()
             predictor_index = iris.coords.DimCoord(
-                np.array(index, dtype=np.int32), long_name="predictor_index", units="1",
+                np.array(index, dtype=np.int8), long_name="predictor_index", units="1",
             )
             template_cube_copy.add_aux_coord(predictor_index)
             template_cube_copy = iris.util.new_axis(template_cube_copy, predictor_index)
@@ -882,7 +883,7 @@ def _add_predictor_coords(
         template_cube.add_aux_coord(
             predictor_name, data_dims=template_cube.coord_dims("predictor_index"),
         )
-        return template_cube
+        return iris.util.squeeze(template_cube)
 
     def _create_cubelist(
         self,
@@ -930,7 +931,10 @@ def _create_cubelist(
                 template_cube = self._add_predictor_coords(
                     template_cube, forecast_predictors
                 )
+                optimised_coeff = np.reshape(optimised_coeff, template_cube.shape)
                 replacements += ["predictor_index", "predictor_name"]
+            else:
+                optimised_coeff = np.array(optimised_coeff)
 
             for coord in coords_to_replace:
                 template_cube.replace_coord(coord)
@@ -952,9 +956,7 @@ def _create_cubelist(
                 template_cube,
                 generate_mandatory_attributes([historic_forecasts]),
                 optional_attributes=self._set_attributes(historic_forecasts),
-                data=np.reshape(optimised_coeff, template_cube.shape)
-                if "beta" == coeff_name
-                else np.array(optimised_coeff),
+                data=optimised_coeff,
             )
             cubelist.append(cube)
         return cubelist
@@ -1021,7 +1023,6 @@ def compute_initial_guess(
         truths: ndarray,
         forecast_predictor: ndarray,
         predictor: str,
-        number_of_forecast_predictors: int,
         number_of_realizations: Optional[int],
     ) -> List[float]:
         """
@@ -1065,9 +1066,6 @@ def compute_initial_guess(
                 the location parameter when estimating the EMOS coefficients.
                 Currently the ensemble mean ("mean") and the ensemble
                 realizations ("realizations") are supported as the predictors.
-            number_of_forecast_predictors:
-                Number of forecast predictors. This includes all additional
-                fields to include in the calibration.
             number_of_realizations:
                 Number of realizations within the forecast predictor. If no
                 realizations are present, this option is None.
@@ -1089,7 +1087,7 @@ def compute_initial_guess(
 
         if predictor == "mean" and default_initial_guess:
             initial_beta = np.repeat(
-                1.0 / number_of_forecast_predictors, number_of_forecast_predictors
+                1.0 / forecast_predictor.shape[0], forecast_predictor.shape[0]
             ).tolist()
             initial_guess = [0] + initial_beta + [0, 1]
         elif predictor == "realizations" and default_initial_guess:
@@ -1175,22 +1173,18 @@ def guess_and_minimise(
 
         """
         if self.point_by_point and not self.use_default_initial_guess:
-            index = [
-                truths.coord(axis="y"),
-                truths.coord(axis="x"),
-            ]
             y_name = truths.coord(axis="y").name()
             x_name = truths.coord(axis="x").name()
 
             initial_guess = []
-            for truth_slice in truths.slices_over(index):
+            for truth_slice in truths.slices_over([y_name, x_name]):
                 constr = iris.Constraint(
                     coord_values={
                         y_name: lambda cell: any(
-                            np.isclose(cell.point, truth_slice.coord(axis="y").points)
+                            np.isclose(cell.point, truth_slice.coord(y_name).points)
                         ),
                         x_name: lambda cell: any(
-                            np.isclose(cell.point, truth_slice.coord(axis="x").points)
+                            np.isclose(cell.point, truth_slice.coord(x_name).points)
                         ),
                     }
                 )
@@ -1209,7 +1203,6 @@ def guess_and_minimise(
                         truth_slice.data,
                         forecast_predictors_data,
                         self.predictor,
-                        len(forecast_predictors),
                         number_of_realizations,
                     )
                 )
@@ -1227,7 +1220,6 @@ def guess_and_minimise(
                 truths.data,
                 forecast_predictor_data,
                 self.predictor,
-                len(forecast_predictors),
                 number_of_realizations,
             )
             if self.point_by_point:
@@ -1334,14 +1326,14 @@ def process(
                 "forecast_reference_time",
                 "realization",
             ]
-            for af_cube in additional_fields:
-                if any([af_cube.coords(c) for c in disallowed_coords]):
+            for cube in additional_fields:
+                if any([cube.coords(c) for c in disallowed_coords]):
                     coords = [
-                        af_cube.coord(c) for c in disallowed_coords if af_cube.coords(c)
+                        cube.coord(c) for c in disallowed_coords if cube.coords(c)
                     ]
                     msg = (
                         "Only static additional predictors are supported. "
-                        f"The {af_cube.name()} cube provided contains {coords}."
+                        f"The {cube.name()} cube provided contains {coords}."
                     )
                     raise ValueError(msg)
 
@@ -1360,9 +1352,8 @@ def process(
 
         number_of_realizations = None
         if self.predictor == "mean":
-            forecast_predictors = iris.cube.CubeList()
-            forecast_predictors.append(
-                collapsed(historic_forecasts, "realization", iris.analysis.MEAN)
+            forecast_predictors = iris.cube.CubeList(
+                [collapsed(historic_forecasts, "realization", iris.analysis.MEAN)]
             )
         elif self.predictor == "realizations":
             number_of_realizations = len(historic_forecasts.coord("realization").points)
@@ -1417,6 +1408,7 @@ def __init__(self, predictor: str = "mean") -> None:
 
         self.coefficients_cubelist = None
         self.current_forecast = None
+        self.additional_fields = None
 
     def __repr__(self) -> str:
         """Represent the configured plugin instance as a string."""
@@ -1491,29 +1483,18 @@ def _calculate_location_parameter_from_mean(self) -> ndarray:
             Location parameter calculated using the ensemble mean as the
             predictor.
         """
-        forecast_predictors = iris.cube.CubeList()
-        forecast_predictors.append(
-            collapsed(self.current_forecast, "realization", iris.analysis.MEAN)
+        forecast_predictors = iris.cube.CubeList(
+            [collapsed(self.current_forecast, "realization", iris.analysis.MEAN)]
         )
         if self.additional_fields:
             forecast_predictors.extend(self.additional_fields)
 
+        beta_cube = self.coefficients_cubelist.extract_cube("emos_coefficient_beta")
+
         fp_names = [fp.name() for fp in forecast_predictors]
-        if len(forecast_predictors) != len(
-            self.coefficients_cubelist.extract_cube("emos_coefficient_beta")
-            .coord("predictor_index")
-            .points
-        ):
-            n_coord_points = len(
-                self.coefficients_cubelist.extract_cube("emos_coefficient_beta")
-                .coord("predictor_index")
-                .points
-            )
-            coord_names = (
-                self.coefficients_cubelist.extract_cube("emos_coefficient_beta")
-                .coord("predictor_name")
-                .points
-            )
+        if len(forecast_predictors) != len(beta_cube.coord("predictor_index").points):
+            n_coord_points = len(beta_cube.coord("predictor_index").points)
+            coord_names = beta_cube.coord("predictor_name").points
             msg = (
                 "The number of forecast predictors must equal the number of "
                 "beta coefficients in order to create a calibrated forecast. "
@@ -1527,12 +1508,7 @@ def _calculate_location_parameter_from_mean(self) -> ndarray:
         location_parameter = np.zeros(forecast_predictors[0].shape)
         for fp in forecast_predictors:
             constr = iris.Constraint(predictor_name=fp.name())
-            location_parameter += (
-                self.coefficients_cubelist.extract_cube("emos_coefficient_beta")
-                .extract(constr)
-                .data
-                * fp.data
-            )
+            location_parameter += beta_cube.extract(constr).data * fp.data
         location_parameter += self.coefficients_cubelist.extract_cube(
             "emos_coefficient_alpha"
         ).data

improver_tests/acceptance/SHA256SUMS

@@ -8,23 +8,23 @@ bee125371a9db4c80fb7f8d1b0644a2710903aad27023b5939331f7ec5f3b617  ./apply-emos-c
 fa8efb4b54430cf6f5e4782b9ec1a0c49dd4102d228dae04bd250ec7b0b9b481  ./apply-emos-coefficients/land_sea/realizations_kgo.nc
 4eaedd69b7d0bbc4b99fa598ffffdb39454962d8e9c33c3d75f58d1de9f25313  ./apply-emos-coefficients/normal/input.nc
 0c5839e3f3a9d557a24757c292099af859ff7ce27c75de7ba8b98a46828b3912  ./apply-emos-coefficients/normal/kgo.nc
-f2af75abf277e1c270bc4921b092d72c1d3423ac850110e372cfd745852e46ef  ./apply-emos-coefficients/normal/normal_coefficients.nc
+7c4033340c89788ba3e55f560fbe2f47ed8bac14c7b47cdf71312e115caf7cfc  ./apply-emos-coefficients/normal/normal_coefficients.nc
 d51fde4566a4d9695a3ba26622ef8505a879131e10f438ab19d4594a57997ad7  ./apply-emos-coefficients/percentiles/input.nc
 7e673a16dd80edb42f890771184a4ad8592a53bd0d0d18d7cd316c489c9b4d7d  ./apply-emos-coefficients/percentiles/kgo.nc
 bd56f09033f7a37323e04be7b509e8e363eb2474c04fb8aa4fb5e3769acb81f7  ./apply-emos-coefficients/probabilities/input.nc
 8eedf1e3cbee70e33fc281f2da06f89895e68c0d8a6ad3099b6c9522b6c4ee45  ./apply-emos-coefficients/probabilities/kgo.nc
-6a4af7032dee64f5b9ea0e3790788814cca11d83a1a0227af08b33233c04609d  ./apply-emos-coefficients/realizations/realizations_coefficients.nc
+12110777cf3db3c8d76daa06f78760f8e314038262f359e659d16fa5d49b98bf  ./apply-emos-coefficients/realizations/realizations_coefficients.nc
 7c5cd2382cad998e22ef0d4210bf65d131587514de2b5cf4ad1fe3c30fede188  ./apply-emos-coefficients/realizations/realizations_kgo.nc
 615ddb59a838024bb055b9cbbb44b7c18ce28d5734952777774980760e0f36fc  ./apply-emos-coefficients/rebadged_percentiles/input.nc
 0bc91af1e7003d696aa440a07d79e653bca566733cdf9bf525ca92cff821548f  ./apply-emos-coefficients/sites/input.nc
-f3f2b6ccafab05468751e7efa9ce42c56a27b4543419bb14acb205381ca2ed1f  ./apply-emos-coefficients/sites/offset/coefficients.nc
+a0112b5a48ba99a1a4a345d43b0c453caaf25181504fb9a13786b5722f84cc10  ./apply-emos-coefficients/sites/offset/coefficients.nc
 22d950494dc9d76d642b0ecc75414e95bfb78fa693e5701fe7cd2fc462a0c6a7  ./apply-emos-coefficients/sites/offset/kgo.nc
 2722b1d08a87cebf1f36ac914d5badb9af38859b2150da83311dc202ce9be4dd  ./apply-emos-coefficients/sites/offset/offset_input.nc
-b587f6f057d979eebef1c2351e86ba67b99caaa2b80d2792a73b6d8691dd8eeb  ./apply-emos-coefficients/sites/point_by_point/coefficients.nc
+63291b3435066f0b0fc56bc78b9774e0e7457d5a6b20086a67ecc6eb826d618d  ./apply-emos-coefficients/sites/point_by_point/coefficients.nc
 833a703e14a02999a7d70e9a37ce199ab36a09b13b96f87295a0fb91739658ce  ./apply-emos-coefficients/sites/point_by_point/kgo.nc
 965a1f0565f9192d95eb01d0a61dc7bede6902f5e1a0481d92611e677841a139  ./apply-emos-coefficients/truncated_normal/input.nc
 feb2d93cf96b05889571bcb348dfbb4d60cfc1f3d9b7343dcf7d85cf34339746  ./apply-emos-coefficients/truncated_normal/kgo.nc
-32f112547dae560e966413b1e326564adac86fda349f1b4256ac1c5910b2d90a  ./apply-emos-coefficients/truncated_normal/truncated_normal_coefficients.nc
+d2cab1d3d8aa588be08a3d7d65e95e859fed37daa767e8d4a2bdaae25702b9a8  ./apply-emos-coefficients/truncated_normal/truncated_normal_coefficients.nc
 182891343e44dbccb637cc668946fad2902d25bc401bf3e539c9d8d68fe4ef39  ./apply-lapse-rate/basic/kgo.nc
 d72e62922cd2d97b7ad82f6b58706b9bfebf1baa0a3118a3cd26d54ca846c3ad  ./apply-lapse-rate/basic/ukvx_lapse_rate.nc
 d81b1559d07cc9ff9086fd8fe3d72b0228701eae706be755307d18c7bfcb7df2  ./apply-lapse-rate/basic/ukvx_orography.nc
@@ -100,19 +100,19 @@ bf7e42be7897606682c3ecdaeb27bf3d3b6ab13a9a88b46c88ae6e92801c6245  ./create-grid-
 d84aaed836c80f55e11207c8618d556b8aaca0c9dfd4320a3ea88387de71670c  ./estimate-emos-coefficients/combined_input/historic_forecast.json
 eeda3b2a092f3bc78bf7dc8cc22efd5e885f94b8b8e04ccc4740978dfc79e04d  ./estimate-emos-coefficients/combined_input/truth.json
 ff3a00a16fa94697d6e529a6f98384e4602f70a7d6ce26669c3947c8ac2e6f7e  ./estimate-emos-coefficients/landmask.nc
-cb0e56a8f4e1830f9462255bbb6e6ab5e1dd647942ec62677fa509ffad528f6d  ./estimate-emos-coefficients/normal/default_initial_guess_kgo.nc
+4907ffd47370afdcc7aaf001e633240ec89e092463ad7ae25e910af10d6c28e4  ./estimate-emos-coefficients/normal/default_initial_guess_kgo.nc
 f878fca6cb6b8af8967d1bcdae28d9933b27cfd25e8e9bcf93498cae3e2163df  ./estimate-emos-coefficients/normal/history/20170602T0300Z-PT0012H-temperature_at_screen_level.nc
 c72e48ae88fb3a9c909323376c20d9bcaecec5e1b55a42accdab7201ce8d7cc8  ./estimate-emos-coefficients/normal/history/20170603T0300Z-PT0012H-temperature_at_screen_level.nc
 a7bd5d01ed80297fb5b28be9a308fd8de20c162a87220be5e59c8ac3ffa800f0  ./estimate-emos-coefficients/normal/history/20170604T0300Z-PT0012H-temperature_at_screen_level.nc
-f2af75abf277e1c270bc4921b092d72c1d3423ac850110e372cfd745852e46ef  ./estimate-emos-coefficients/normal/kgo.nc
-4964511af78cd064396e897b2efee748c30bdcaa925882f9d31006f23f4a14ef  ./estimate-emos-coefficients/normal/land_only_kgo.nc
-6a4af7032dee64f5b9ea0e3790788814cca11d83a1a0227af08b33233c04609d  ./estimate-emos-coefficients/normal/realizations/kgo.nc
+7c4033340c89788ba3e55f560fbe2f47ed8bac14c7b47cdf71312e115caf7cfc  ./estimate-emos-coefficients/normal/kgo.nc
+67d50642e9a49a0c9115c786cc74f62fc27e10aaaff7ba7e2329d2580c63aefd  ./estimate-emos-coefficients/normal/land_only_kgo.nc
+12110777cf3db3c8d76daa06f78760f8e314038262f359e659d16fa5d49b98bf  ./estimate-emos-coefficients/normal/realizations/kgo.nc
 42fe8daffb997848086cb08b3d39dd366b561bb1c24fc7ed3e2e54737d772b9a  ./estimate-emos-coefficients/normal/sites/history/20201209T1500Z-PT0012H00M-temperature_at_screen_level.nc
 3cda6a3244db09530585454d323ec13bde6b4cfae24940a8b069469f47aec60c  ./estimate-emos-coefficients/normal/sites/history/20201210T1500Z-PT0012H00M-temperature_at_screen_level.nc
 e5dab98673996a3663093a0353b4efac7910b8176921bd0fbf10aa0f53ae5800  ./estimate-emos-coefficients/normal/sites/history/20201211T1500Z-PT0012H00M-temperature_at_screen_level.nc
-b587f6f057d979eebef1c2351e86ba67b99caaa2b80d2792a73b6d8691dd8eeb  ./estimate-emos-coefficients/normal/sites/point_by_point/kgo.nc
-980b19f4e5dcc557e3e499ee884ade51e6a6052fca2d7d2d95259ea442ec99b5  ./estimate-emos-coefficients/normal/sites/point_by_point/realizations_kgo.nc
-cb2452279a335a09edc0fc9aced714dddd8cc1472344a67f0487e7e9b55cd32d  ./estimate-emos-coefficients/normal/sites/point_by_point_default_initial_guess/kgo.nc
+63291b3435066f0b0fc56bc78b9774e0e7457d5a6b20086a67ecc6eb826d618d  ./estimate-emos-coefficients/normal/sites/point_by_point/kgo.nc
+20ba2f8b50ccafb6e10af1db9ecef72f2d529afd131a652bf2cfa70c64d5f95b  ./estimate-emos-coefficients/normal/sites/point_by_point/realizations_kgo.nc
+36412f957fb44761e389432c401d49a443d50add44fc1b40ce47456962173fba  ./estimate-emos-coefficients/normal/sites/point_by_point_default_initial_guess/kgo.nc
 c93cb58b443c90fec46135715c322d2a4a214868f13bcb073871690c0a59aeaa  ./estimate-emos-coefficients/normal/sites/truth/20201209T1500Z-PT0000H00M-temperature_at_screen_level.nc
 d128c7ae795bbc5c7281d4140f1d7826e82e82160ec6320317f72bccd08bcb83  ./estimate-emos-coefficients/normal/sites/truth/20201210T1500Z-PT0000H00M-temperature_at_screen_level.nc
 bca3f986233d44f4c008f1113d40e808f730b80428a55fe4eb4c6f9ee8241e16  ./estimate-emos-coefficients/normal/sites/truth/20201211T1500Z-PT0000H00M-temperature_at_screen_level.nc
@@ -122,7 +122,7 @@ ee8638a56d5c8866106ed0e0e9578f391300d9271b78b49e70baeae78ebb9e2d  ./estimate-emo
 4d6f2a70e895017464765f7c9ec1df67e4fc33992783241971a4d630177b6e08  ./estimate-emos-coefficients/truncated_normal/history/20170602T0300Z-PT0012H-horizontal_wind_speed_at_10m.nc
 11e0ff3506e0456abd4b169846cf1f5bcc9596fcb50eb520ba9f68c988fa786d  ./estimate-emos-coefficients/truncated_normal/history/20170603T0300Z-PT0012H-horizontal_wind_speed_at_10m.nc
 db4aa83b225fd02aa6d15b5ea1500b960d2da46378290ebba9d45b8a29c189f2  ./estimate-emos-coefficients/truncated_normal/history/20170604T0300Z-PT0012H-horizontal_wind_speed_at_10m.nc
-32f112547dae560e966413b1e326564adac86fda349f1b4256ac1c5910b2d90a  ./estimate-emos-coefficients/truncated_normal/kgo.nc
+d2cab1d3d8aa588be08a3d7d65e95e859fed37daa767e8d4a2bdaae25702b9a8  ./estimate-emos-coefficients/truncated_normal/kgo.nc
 397b0e3712041863f9d8ac45480b67162f6e740cf9b7d2e1521a6844074146e8  ./estimate-emos-coefficients/truncated_normal/truth/20170602T1500Z-PT0000H-horizontal_wind_speed_at_10m.nc
 1032b6e41987c08ad4caace25e182a02701177408461c7b573f3e8b304b31c16  ./estimate-emos-coefficients/truncated_normal/truth/20170603T1500Z-PT0000H-horizontal_wind_speed_at_10m.nc
 fe84db49d69ea62e2e82312c7b3d860c7329c627d79d76c2543d365f69d2e0d8  ./estimate-emos-coefficients/truncated_normal/truth/20170604T1500Z-PT0000H-horizontal_wind_speed_at_10m.nc