def test_non_integer_width(self):
"""Test when the width of the triangle does not fall on a grid point.
This only affects the slope of the triangle slightly."""
width = 3.5
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.0,
0.0,
0.04,
0.12,
0.2,
0.28,
0.2,
0.12,
0.04,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_unevenly_spaced_coord(self):
"""Test the case where the input coordinate is not equally spaced.
This represents the case where the data changes to 3 hourly. In this
case the weights are assigned according to the value in the
coordinate."""
width = 5
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(10)
coord_vals = np.append(coord_vals, [12, 15, 18, 21, 24])
midpoint = 8
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.0,
0.0,
0.0,
0.0,
0.05,
0.1,
0.15,
0.2,
0.25,
0.2,
0.05,
0.0,
0.0,
0.0,
0.0,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_large_width(self):
"""Test the case where the width of the triangle is larger than the
coordinate input.
In this case all the weights are non-zero but still form the
shape of a triangle."""
width = 10
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.055556,
0.066667,
0.077778,
0.088889,
0.1,
0.111111,
0.1,
0.088889,
0.077778,
0.066667,
0.055556,
0.044444,
0.033333,
0.022222,
0.011111,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_non_integer_midpoint(self):
"""Test the case where the midpoint of the triangle is not a point in
the input coordinate.
In this case we do not sample the peak of the triangle."""
width = 2
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 3.5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.0,
0.0,
0.125,
0.375,
0.375,
0.125,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_midpoint_at_edge(self):
"""Test that the correct triangular weights are returned for a case
where the midpoint is close to the end of the input coordinate.
In this case the triangle is cut off at the end of the coordinate"""
width = 3
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 1
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.25,
0.375,
0.25,
0.125,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def __init__(self, coord, central_point, parameter_units, width):
"""Set up for a Weighted Blending plugin
Args:
coord (str):
The name of a coordinate dimension in the cube that we
will blend over.
central_point (float or int):
Central point at which the output from the triangular weighted
blending will be calculated.
parameter_units (str):
The units of the width of the triangular weighting function
and the units of the central_point.
This does not need to be the same as the units of the
coordinate we are blending over, but it should be possible to
convert between them.
width (float):
The width of the triangular weighting function we will use
to blend.
"""
self.coord = coord
self.central_point = central_point
self.parameter_units = parameter_units
self.width = width
# Set up a plugin to calculate the triangular weights.
self.WeightsPlugin = ChooseDefaultWeightsTriangular(
width, units=parameter_units)
# Set up the blending function, based on whether weighted blending or
# maximum probabilities are needed.
self.BlendingPlugin = (WeightedBlendAcrossWholeDimension(
coord, timeblending=True))
def test_basic_weights(self):
"""Test that the function returns the correct triangular weights in a
simple case"""
width = 3
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
expected_weights = np.array([
0.0,
0.0,
0.0,
0.11111111,
0.22222222,
0.33333333,
0.22222222,
0.11111111,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_same_units(self):
"""Test plugin produces the correct weights when the parameters for
the triangle are in the same units as the input cube's coordinate"""
width = 2
WeightsClass = ChooseDefaultWeightsTriangular(width, units=self.units)
midpoint = 1
weights = WeightsClass.process(self.cube, self.coord_name, midpoint)
expected_weights = np.array([0.33333333, 0.66666667])
self.assertArrayAlmostEqual(weights, expected_weights)
def test_unconvertable_units(self):
""""Test plugin produces the correct weights when the parameters for
the triangle cannot be converted to the same units as the
coordinate"""
width = 7200
weights_instance = ChooseDefaultWeightsTriangular(width, units="m")
midpoint = 3600
message = r"Unable to convert from 'Unit\('m'\)' to 'Unit\('hours'\)'"
with self.assertRaisesRegex(ValueError, message):
weights_instance.process(self.cube, self.coord_name, midpoint)
def test_different_units(self):
""""Test plugin produces the correct weights when the parameters for
the triangle (width and midpoint are in different units to the
input cube's coordinate"""
width = 2
weights_instance = ChooseDefaultWeightsTriangular(width, units="hours")
midpoint = 1
weights = weights_instance.process(self.cube, self.coord_name,
midpoint)
expected_weights = np.array([0.33333333, 0.66666667])
self.assertArrayAlmostEqual(weights.data, expected_weights)
def test_basic(self):
"""Test that the function returns a numpy array.
Also check that the length of the weights is correct and they add
up to 1.0"""
width = 3
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
self.assertIsInstance(weights, np.ndarray)
self.assertEqual(len(weights), len(coord_vals))
self.assertEqual(weights.sum(), 1.0)
def test_basic(self):
"""Test that the function returns a numpy array.
Also check that the length of the weights is correct and they add
up to 1.0"""
width = 3
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
coord_vals = np.arange(15)
midpoint = 5
weights = triangular_weights_instance.triangular_weights(
coord_vals, midpoint, width)
self.assertIsInstance(weights, np.ndarray)
self.assertEqual(len(weights), len(coord_vals))
# Using float32s for weights can give 1.0000001 here - good enough.
self.assertAlmostEqual(weights.sum(), 1.0, places=6)
def __init__(self, coord, central_point, parameter_units, width,
weighting_mode):
"""Set up for a Weighted Blending plugin
Args:
coord (string):
The name of a coordinate dimension in the cube that we
will blend over.
central_point (float or int):
Central point at which the output from the triangular weighted
blending will be calculated.
parameter_units (string):
The units of the width of the triangular weighting function
and the units of the central_point.
This does not need to be the same as the units of the
coordinate we are blending over, but it should be possible to
convert between them.
width (float):
The width of the triangular weighting function we will use
to blend.
weighting_mode (string):
The mode of blending, either weighted_mean or
weighted_maximum. Weighted average finds the weighted mean
across the dimension of interest. Maximum probability
multiplies the values across the dimension of interest by the
given weights and returns the maximum value.
Raises:
ValueError : If an invalid weighting_mode is given
"""
self.coord = coord
self.central_point = central_point
self.parameter_units = parameter_units
self.width = width
if weighting_mode not in ['weighted_maximum', 'weighted_mean']:
msg = ("weighting_mode: {} is not recognised, must be either "
"weighted_maximum or weighted_mean").format(weighting_mode)
raise ValueError(msg)
self.mode = weighting_mode
# Set up a plugin to calculate the triangular weights.
self.WeightsPlugin = ChooseDefaultWeightsTriangular(
width, units=parameter_units)
# Set up the blending function, based on whether weighted blending or
# maximum probabilities are needed.
self.BlendingPlugin = (WeightedBlendAcrossWholeDimension(
coord, weighting_mode, timeblending=True))
def process(self, cube):
"""
Apply the weighted blend for each point in the given coordinate.
Args:
cube : iris.cube.Cube
Cube to blend.
Returns:
cube: iris.cube.Cube
The processed cube, with the same coordinates as the input
cube. The points in one coordinate will be blended with the
adjacent points based on a triangular weighting function of the
specified width.
"""
# We need to correct all the coordinates associated with the dimension
# we are collapsing over, so find the relevant coordinates now.
dimension_to_collapse = cube.coord_dims(self.coord)
coords_to_correct = cube.coords(dimensions=dimension_to_collapse)
coords_to_correct = [coord.name() for coord in coords_to_correct]
# We will also need to correct the bounds on these coordinates,
# as bounds will be added when the blending happens, so add bounds if
# it doesn't have some already.
for coord in coords_to_correct:
if not cube.coord(coord).has_bounds():
cube.coord(coord).guess_bounds()
# Set up a plugin to calculate the triangular weights.
WeightsPlugin = ChooseDefaultWeightsTriangular(
self.width, units=self.parameter_units)
# Set up the blending function, based on whether weighted blending or
# maximum probabilities are needed.
BlendingPlugin = WeightedBlendAcrossWholeDimension(self.coord,
self.mode)
result = iris.cube.CubeList([])
# Loop over each point in the coordinate we are blending over, and
# calculate a new weighted average for it.
for cube_slice in cube.slices_over(self.coord):
point = cube_slice.coord(self.coord).points[0]
weights = WeightsPlugin.process(cube, self.coord, point)
blended_cube = BlendingPlugin.process(cube, weights)
self.correct_collapsed_coordinates(cube_slice, blended_cube,
coords_to_correct)
result.append(blended_cube)
result = concatenate_cubes(result)
return result
def test_basic_no_units(self):
"""Test the repr function formats the arguments correctly"""
width = 3.0
triangular_weights_instance = ChooseDefaultWeightsTriangular(width)
result = str(triangular_weights_instance)
expected = ("<ChooseDefaultTriangularWeights "
"width=3.0, parameters_units=no_unit>")
self.assertEqual(result, expected)
def test_cf_unit_input(self):
"""Test the case where an instance of cf_units.Unit is passed in"""
units = Unit("hour")
width = 5
weights_instance = ChooseDefaultWeightsTriangular(width, units=units)
expected_width = 5
expected_unit = units
self.assertEqual(weights_instance.width, expected_width)
self.assertEqual(weights_instance.parameters_units, expected_unit)
def test_basic(self):
"""Test the repr function formats the arguments correctly"""
width = 3
TriangularWeightsClass = ChooseDefaultWeightsTriangular(width,
units='hours')
result = str(TriangularWeightsClass)
expected = ("<ChooseDefaultTriangularWeights width= 3.0,"
" parameters_units=hours>")
self.assertEqual(result, expected)
def test_string_input(self):
"""Test the case where a string is passed and gets converted to a
Unit instance"""
units = "hour"
width = 5
weights_instance = ChooseDefaultWeightsTriangular(width, units=units)
expected_width = 5
expected_unit = Unit("hour")
self.assertEqual(weights_instance.width, expected_width)
self.assertEqual(weights_instance.parameters_units, expected_unit)
def __init__(
self,
coord: str,
central_point: Union[int, float],
parameter_units: str,
width: float,
) -> None:
"""Set up for a Weighted Blending plugin
Args:
coord:
The name of a coordinate dimension in the cube to be blended
over.
central_point:
Central point at which the output from the triangular weighted
blending will be calculated. This should be in the units of the
units argument that is passed in. This value should be a point
on the coordinate for blending over.
parameter_units:
The units of the width of the triangular weighting function
and the units of the central_point.
This does not need to be the same as the units of the
coordinate being blending over, but it should be possible to
convert between them.
width:
The width from the triangle’s centre point, in units of the units
argument, which will determine the triangular weighting function
used to blend that specified point with its adjacent points. Beyond
this width the weighting drops to zero.
"""
self.coord = coord
self.central_point = central_point
self.parameter_units = parameter_units
self.width = width
# Set up a plugin to calculate the triangular weights.
self.WeightsPlugin = ChooseDefaultWeightsTriangular(
width, units=parameter_units
)
# Set up the blending function, based on whether weighted blending or
# maximum probabilities are needed.
self.BlendingPlugin = WeightedBlendAcrossWholeDimension(
coord, timeblending=True
)
class TriangularWeightedBlendAcrossAdjacentPoints(object):
"""
Apply a Weighted blend to a coordinate, using triangular weights at each
point in the coordinate. Returns a cube with the same coordinates as the
input cube, with each point in the coordinate of interest having been
blended with the adjacent points according to a triangular weighting
function of a specified width.
There are two modes of blending:
1. Weighted mean across the dimension of interest.
2. Weighted maximum across the dimension of interest, where
probabilities are multiplied by the weights and the maximum is
taken.
"""
def __init__(self, coord, central_point, parameter_units, width,
weighting_mode):
"""Set up for a Weighted Blending plugin
Args:
coord (string):
The name of a coordinate dimension in the cube that we
will blend over.
central_point (float or int):
Central point at which the output from the triangular weighted
blending will be calculated.
parameter_units (string):
The units of the width of the triangular weighting function
and the units of the central_point.
This does not need to be the same as the units of the
coordinate we are blending over, but it should be possible to
convert between them.
width (float):
The width of the triangular weighting function we will use
to blend.
weighting_mode (string):
The mode of blending, either weighted_mean or
weighted_maximum. Weighted average finds the weighted mean
across the dimension of interest. Maximum probability
multiplies the values across the dimension of interest by the
given weights and returns the maximum value.
Raises:
ValueError : If an invalid weighting_mode is given
"""
self.coord = coord
self.central_point = central_point
self.parameter_units = parameter_units
self.width = width
if weighting_mode not in ['weighted_maximum', 'weighted_mean']:
msg = ("weighting_mode: {} is not recognised, must be either "
"weighted_maximum or weighted_mean").format(weighting_mode)
raise ValueError(msg)
self.mode = weighting_mode
# Set up a plugin to calculate the triangular weights.
self.WeightsPlugin = ChooseDefaultWeightsTriangular(
width, units=parameter_units)
# Set up the blending function, based on whether weighted blending or
# maximum probabilities are needed.
self.BlendingPlugin = (WeightedBlendAcrossWholeDimension(
coord, weighting_mode, timeblending=True))
def __repr__(self):
"""Represent the configured plugin instance as a string."""
msg = ('<TriangularWeightedBlendAcrossAdjacentPoints:'
' coord = {0:s}, central_point = {1:.2f}, '
'parameter_units = {2:s}, width = {3:.2f}, mode = {4:s}>')
return msg.format(self.coord, self.central_point, self.parameter_units,
self.width, self.mode)
def _find_central_point(self, cube):
"""
Find the cube that contains the central point, otherwise, raise
an exception.
Args:
cube (iris.cube.Cube):
Cube containing input for blending.
Returns:
central_point_cube (iris.cube.Cube):
Cube containing central point.
Raises:
ValueError: Central point is not available within the input cube.
"""
# Convert central point into the units of the cube, so that a
# central point can be extracted.
central_point = (Unit(self.parameter_units).convert(
self.central_point,
cube.coord(self.coord).units))
constr = iris.Constraint(coord_values={self.coord: central_point})
central_point_cube = cube.extract(constr)
if central_point_cube is None:
msg = ("The central point {} in units of {} not available "
"within input cube coordinate points: {}.".format(
self.central_point, self.parameter_units,
cube.coord(self.coord).points))
raise ValueError(msg)
return central_point_cube
def process(self, cube):
"""
Apply the weighted blend for each point in the given coordinate.
Args:
cube (iris.cube.Cube):
Cube containing input for blending.
Returns:
blended_cube (iris.cube.Cube):
The processed cube, with the same coordinates as the input
central_cube. The points in one coordinate will be blended
with the adjacent points based on a triangular weighting
function of the specified width.
"""
# Extract the central point from the input cube.
central_point_cube = self._find_central_point(cube)
# Calculate weights and produce blended output.
weights = self.WeightsPlugin.process(cube, self.coord,
self.central_point)
blended_cube = self.BlendingPlugin.process(cube, weights)
# With one threshold dimension (such as for low cloud), the threshold
# axis is demoted to a scalar co-ordinate by BlendingPlugin. This line
# promotes threshold to match the dimensions of central_point_cube.
blended_cube = check_cube_coordinates(central_point_cube, blended_cube)
blended_cube = central_point_cube.copy(blended_cube.data)
return blended_cube