Allowed usage of multi_model_statistics on single cubes/products

esmvalcore/preprocessor/_multimodel.py

@@ -19,7 +19,7 @@
 import numpy as np
 from iris.cube import Cube, CubeList
 from iris.exceptions import MergeError
-from iris.util import equalise_attributes
+from iris.util import equalise_attributes, new_axis
 
 from esmvalcore.iris_helpers import date2num
 from esmvalcore.preprocessor import remove_fx_variables
@@ -302,6 +302,12 @@ def _combine(cubes):
 
     cubes = CubeList(cubes)
 
+    # For a single cube, merging returns a scalar CONCAT_DIM, which leads to a
+    # "Cannot collapse a dimension which does not describe any data" error when
+    # collapsing. Thus, treat single cubes differently here.
+    if len(cubes) == 1:
+        return new_axis(cubes[0], scalar_coord=CONCAT_DIM)
+
     try:
         merged_cube = cubes.merge_cube()
     except MergeError as exc:
@@ -411,9 +417,10 @@ def _multicube_statistics(cubes, statistics, span):
     Cubes are merged and subsequently collapsed along a new auxiliary
     coordinate. Inconsistent attributes will be removed.
     """
-    if len(cubes) == 1:
-        raise ValueError('Cannot perform multicube statistics '
-                         'for a single cube.')
+    if not cubes:
+        raise ValueError(
+            "Cannot perform multicube statistics for an empty list of cubes"
+        )
 
     # Avoid modifying inputs
     copied_cubes = [cube.copy() for cube in cubes]

tests/unit/preprocessor/_multimodel/test_multimodel.py

@@ -669,19 +669,6 @@ def test_edge_case_sub_daily_data_fail(span):
         _ = multi_model_statistics(cubes, span, statistics)
 
 
-@pytest.mark.parametrize('span', SPAN_OPTIONS)
-def test_edge_case_single_cube_fail(span):
-    """Test that an error is raised when a single cube is passed."""
-    cube = generate_cube_from_dates('monthly')
-    cubes = (cube, )
-
-    statistic = 'min'
-    statistics = (statistic, )
-
-    with pytest.raises(ValueError):
-        _ = multi_model_statistics(cubes, span, statistics)
-
-
 def test_unify_time_coordinates():
     """Test set common calendar."""
     cube1 = generate_cube_from_dates('monthly',
@@ -1045,3 +1032,100 @@ def test_arbitrary_dims_0d(cubes_with_arbitrary_dimensions):
     stat_cube = stat_cubes['sum']
     assert stat_cube.shape == ()
     assert_array_allclose(stat_cube.data, np.ma.array(0.0))
+
+
+def test_empty_input_multi_model_statistics():
+    """Check that ``multi_model_statistics`` fails with empty input."""
+    msg = "Cannot perform multicube statistics for an empty list of cubes"
+    with pytest.raises(ValueError, match=msg):
+        mm.multi_model_statistics([], span='full', statistics=['mean'])
+
+
+def test_empty_input_ensemble_statistics():
+    """Check that ``ensemble_statistics`` fails with empty input."""
+    msg = "Cannot perform multicube statistics for an empty list of cubes"
+    with pytest.raises(ValueError, match=msg):
+        mm.ensemble_statistics(
+            [], span='full', statistics=['mean'], output_products=[]
+        )
+
+
+STATS = ['mean', 'median', 'min', 'max', 'p42.314', 'std_dev']
+
+
+@pytest.mark.parametrize('stat', STATS)
+@pytest.mark.parametrize(
+    'products',
+    [
+        CubeList([generate_cube_from_dates('monthly')]),
+        set([PreprocessorFile(generate_cube_from_dates('monthly'))]),
+    ],
+)
+def test_single_input_multi_model_statistics(products, stat):
+    """Check that ``multi_model_statistics`` works with a single cube."""
+    output = PreprocessorFile()
+    output_products = {'': {stat: output}}
+    kwargs = {
+        'statistics': [stat],
+        'span': 'full',
+        'output_products': output_products,
+        'keep_input_datasets': False,
+    }
+
+    results = mm.multi_model_statistics(products, **kwargs)
+
+    assert len(results) == 1
+
+    if isinstance(results, dict):  # for cube as input
+        cube = results[stat]
+    else:  # for PreprocessorFile as input
+        result = next(iter(results))
+        assert len(result.cubes) == 1
+        cube = result.cubes[0]
+
+    if stat == 'std_dev':
+        assert_array_allclose(
+            cube.data, np.ma.masked_invalid([np.nan, np.nan, np.nan])
+        )
+    else:
+        assert_array_allclose(cube.data, np.ma.array([1.0, 1.0, 1.0]))
+
+
+@pytest.mark.parametrize('stat', STATS)
+@pytest.mark.parametrize(
+    'products',
+    [
+        CubeList([generate_cube_from_dates('monthly')]),
+        {PreprocessorFile(generate_cube_from_dates('monthly'))},
+    ],
+)
+def test_single_input_ensemble_statistics(products, stat):
+    """Check that ``ensemble_statistics`` works with a single cube."""
+    cube = generate_cube_from_dates('monthly')
+    attributes = {
+        'project': 'project',
+        'dataset': 'dataset',
+        'exp': 'exp',
+        'ensemble': '1',
+    }
+    products = {PreprocessorFile(cube, attributes=attributes)}
+    output = PreprocessorFile()
+    output_products = {'project_dataset_exp': {stat: output}}
+    kwargs = {
+        'statistics': [stat],
+        'output_products': output_products,
+    }
+
+    results = mm.ensemble_statistics(products, **kwargs)
+
+    assert len(results) == 1
+    result = next(iter(results))
+    assert len(result.cubes) == 1
+    cube = result.cubes[0]
+
+    if stat == 'std_dev':
+        assert_array_allclose(
+            cube.data, np.ma.masked_invalid([np.nan, np.nan, np.nan])
+        )
+    else:
+        assert_array_allclose(cube.data, np.ma.array([1.0, 1.0, 1.0]))