def test_compute_full_statistic_yr_cube(self):
data = [self.cube1_yr, self.cube2_yr]
stats = multi_model_statistics(data, 'full', ['mean'])
expected_full_mean = np.ma.ones((4, 3, 2, 2))
expected_full_mean.mask = np.zeros((4, 3, 2, 2))
expected_full_mean.mask[2:4] = True
self.assert_array_equal(stats['mean'].data, expected_full_mean)
def test_edge_case_time_not_in_middle_of_months(span):
"""Test case when time coords are not on 15th for monthly data.
Expected behaviour: `multi_model_statistics` will set all dates to
the 15th.
"""
time_points = range(1, 4)
dates1 = [datetime(1850, i, 12, 0, 0, 0) for i in time_points]
dates2 = [datetime(1850, i, 25, 0, 0, 0) for i in time_points]
cubes = (
generate_cube_from_dates(dates1),
generate_cube_from_dates(dates2),
)
statistic = 'min'
statistics = (statistic, )
result = multi_model_statistics(cubes, span, statistics)
result_cube = result[statistic]
time_coord = result_cube.coord('time')
desired = np.array((14., 45., 73.))
np.testing.assert_array_equal(time_coord.points, desired)
# input cubes are updated in-place
for cube in cubes:
np.testing.assert_array_equal(cube.coord('time').points, desired)
def test_edge_case_different_time_offsets(span):
cubes = (
generate_cube_from_dates('monthly',
'360_day',
offset='days since 1888-01-01'),
generate_cube_from_dates('monthly',
'360_day',
offset='days since 1899-01-01'),
)
statistic = 'min'
statistics = (statistic, )
result = multi_model_statistics(cubes, span, statistics)
result_cube = result[statistic]
time_coord = result_cube.coord('time')
assert time_coord.units.calendar == 'gregorian'
assert time_coord.units.origin == 'days since 1850-01-01'
desired = np.array((14., 45., 73.))
np.testing.assert_array_equal(time_coord.points, desired)
# input cubes are updated in-place
for cube in cubes:
np.testing.assert_array_equal(cube.coord('time').points, desired)
def test_compute_overlap_statistic_yr_cube(self):
data = [self.cube4, self.cube4]
stats = multi_model_statistics(products=data,
statistics=['mean'],
span='overlap')
expected_ovlap_mean = np.ma.ones((2, 3, 2, 2))
self.assert_array_equal(stats['mean'].data, expected_ovlap_mean)
def multimodel_test(cubes, span, statistic):
"""Run multimodel test with some simple checks."""
statistics = [statistic]
result = multi_model_statistics(cubes, span=span, statistics=statistics)
assert isinstance(result, dict)
assert statistic in result
return result
def test_compute_full_statistic_mon_cube(self):
data = [self.cube1, self.cube2]
stats = multi_model_statistics(products=data,
statistics=['mean'],
span='full')
expected_full_mean = np.ma.ones((5, 3, 2, 2))
expected_full_mean.mask = np.ones((5, 3, 2, 2))
expected_full_mean.mask[1] = False
self.assert_array_equal(stats['mean'].data, expected_full_mean)
def test_compute_full_statistic_yr_cube(self):
data = [self.cube4, self.cube5]
stats = multi_model_statistics(products=data,
statistics=['mean'],
span='full')
expected_full_mean = np.ma.ones((4, 3, 2, 2))
expected_full_mean.mask = np.zeros((4, 3, 2, 2))
expected_full_mean.mask[2:4] = True
self.assert_array_equal(stats['mean'].data, expected_full_mean)
def test_edge_case_sub_daily_data_fail(span):
"""Test case when cubes with sub-daily time coords are passed."""
cube = generate_cube_from_dates('hourly')
cubes = (cube, cube)
statistic = 'min'
statistics = (statistic, )
with pytest.raises(ValueError):
_ = multi_model_statistics(cubes, span, statistics)
def test_edge_case_time_no_overlap_fail():
"""Test case when time coords do not overlap using span='overlap'.
Expected behaviour: `multi_model_statistics` should fail if time
points are not overlapping.
"""
cubes = generate_cubes_with_non_overlapping_timecoords()
statistic = 'min'
statistics = (statistic, )
with pytest.raises(ValueError):
_ = multi_model_statistics(cubes, 'overlap', statistics)
def test_edge_case_time_no_overlap_success():
"""Test case when time coords do not overlap using span='full'.
Expected behaviour: `multi_model_statistics` should use all
available time points.
"""
cubes = generate_cubes_with_non_overlapping_timecoords()
statistic = 'min'
statistics = (statistic, )
result = multi_model_statistics(cubes, 'full', statistics)
result_cube = result[statistic]
assert result_cube.coord('time').shape == (6, )
def test_multimodel_statistics(frequency, span, statistics, expected):
"""High level test for multicube statistics function.
- Should work for multiple data frequencies
- Should be able to deal with multiple statistics
- Should work for both span arguments
- Should deal correctly with different mask options
- Return type should be a dict with all requested statistics as keys
"""
cubes = get_cubes_for_validation_test(frequency)
if isinstance(statistics, str):
statistics = (statistics, )
expected = (expected, )
result = multi_model_statistics(cubes, span, statistics)
assert isinstance(result, dict)
assert set(result.keys()) == set(statistics)
for i, statistic in enumerate(statistics):
result_cube = result[statistic]
expected_data = np.ma.array(expected[i], mask=False)
assert_array_allclose(result_cube.data, expected_data)
def test_compute_overlap_statistic_yr_cube(self):
data = [self.cube1_yr, self.cube1_yr]
stats = multi_model_statistics(data, 'overlap', ['mean'])
expected_ovlap_mean = np.ma.ones((2, 3, 2, 2))
self.assert_array_equal(stats['mean'].data, expected_ovlap_mean)
