NDCube#
- class ndcube.NDCube(data, wcs=None, uncertainty=None, mask=None, meta=None, unit=None, copy=False, **kwargs)[source]#
Bases:
NDCubeBase
Class representing N-D data described by a single array and set of WCS transformations.
- Parameters:
data (
numpy.ndarray
) – The array holding the actual data in this object.wcs (
astropy.wcs.wcsapi.BaseLowLevelWCS
,astropy.wcs.wcsapi.BaseHighLevelWCS
, optional) – The WCS object containing the axes’ information, optional only ifdata
is anastropy.nddata.NDData
object.uncertainty (Any, optional) – Uncertainty in the dataset. Should have an attribute uncertainty_type that defines what kind of uncertainty is stored, for example “std” for standard deviation or “var” for variance. A metaclass defining such an interface is NDUncertainty - but isn’t mandatory. If the uncertainty has no such attribute the uncertainty is stored as UnknownUncertainty. Defaults to None.
mask (Any, optional) – Mask for the dataset. Masks should follow the numpy convention that valid data points are marked by False and invalid ones with True. Defaults to None.
meta (dict-like object, optional) – Additional meta information about the dataset. If no meta is provided an empty collections.OrderedDict is created. Default is None.
unit (Unit-like or str, optional) – Unit for the dataset. Strings that can be converted to a Unit are allowed. Default is None.
extra_coords (iterable of
tuple
, each with three entries) – (str
,int
,astropy.units.quantity
or array-like) Gives the name, axis of data, and values of coordinates of a data axis not included in the WCS object.copy (bool, optional) – Indicates whether to save the arguments as copy. True copies every attribute before saving it while False tries to save every parameter as reference. Note however that it is not always possible to save the input as reference. Default is False.
Attributes Summary
Returns the WCS physical types that vary along each array axis.
The WCS transform for the NDCube, including the coordinates specified in
.extra_coords
.The stored dataset.
Coordinates not described by
NDCubeABC.wcs
which vary along one or more axes.Coordinate metadata which applies to the whole cube.
Mask for the dataset, if any.
A
MatplotlibPlotter
instance providing visualization methods.Image representation of the PSF for the dataset.
Uncertainty in the dataset, if any.
Unit for the dataset, if any.
A world coordinate system (WCS) for the dataset, if any.
Methods Summary
axis_world_coords
(*axes[, pixel_corners, wcs])Returns objects representing the world coordinates of pixel centers for a desired axes.
axis_world_coords_values
(*axes[, ...])Returns the world coordinate values of all pixels for desired axes.
crop
(*points[, wcs])Crop using real world coordinates.
crop_by_values
(*points[, units, wcs])Crop using real world coordinates.
explode_along_axis
(axis)Separates slices of NDCubes along a given axis into an NDCubeSequence of (N-1)DCubes.
plot
([axes, plot_axes, axes_coordinates, ...])Visualize the
NDCube
.rebin
(bin_shape[, operation, ...])Downsample array by combining contiguous pixels into bins.
reproject_to
(target_wcs[, algorithm, ...])Reprojects the instance to the coordinates described by another WCS object.
to
(new_unit, **kwargs)Convert instance to another unit.
Attributes Documentation
- array_axis_physical_types#
- combined_wcs#
- dimensions#
- extra_coords#
- global_coords#
- mask#
Mask for the dataset, if any.
Masks should follow the
numpy
convention that valid data points are marked byFalse
and invalid ones withTrue
.- Type:
any type
- meta = None#
- plotter = None#
A
MatplotlibPlotter
instance providing visualization methods.The type of this attribute can be changed to provide custom visualization functionality.
- psf#
- uncertainty#
Uncertainty in the dataset, if any.
Should have an attribute
uncertainty_type
that defines what kind of uncertainty is stored, such as'std'
for standard deviation or'var'
for variance. A metaclass defining such an interface isNDUncertainty
but isn’t mandatory.- Type:
any type
- wcs#
A world coordinate system (WCS) for the dataset, if any.
- Type:
any type
Methods Documentation
- axis_world_coords(*axes, pixel_corners=False, wcs=None)#
Returns objects representing the world coordinates of pixel centers for a desired axes.
- Parameters:
axes (
int
orstr
, or multipleint
orstr
, optional) – Axis number in numpy ordering or unique substring ofndcube.NDCube.wcs.world_axis_physical_types
of axes for which real world coordinates are desired. Not specifying axes inputs causes results for all axes to be returned.pixel_corners (
bool
, optional) – IfTrue
then instead of returning the coordinates at the centers of the pixels, the coordinates at the pixel corners will be returned. This increases the size of the output by 1 in all dimensions as all corners are returned.wcs (
astropy.wcs.wcsapi.BaseHighLevelWCS
, optional) – The WCS object to used to calculate the world coordinates. Although technically this can be any valid WCS, it will typically beself.wcs
,self.extra_coords
, orself.combined_wcs
combining both the WCS and extra coords. Default=self.wcs
- Returns:
axes_coords (iterable) – An iterable of “high level” objects giving the real world coords for the axes requested by user. For example, a tuple of
SkyCoord
objects. The types returned are determined by the WCS object. The dimensionality of these objects should match that of their corresponding array dimensions, unlesspixel_corners=True
in which case the length along each axis will be 1 greater than the number of pixels.
Examples
>>> NDCube.axis_world_coords('lat', 'lon') >>> NDCube.axis_world_coords(2)
- axis_world_coords_values(*axes, pixel_corners=False, wcs=None)#
Returns the world coordinate values of all pixels for desired axes. In contrast to
ndcube.NDCube.axis_world_coords()
, this method returnsQuantity
objects. Which only provide units rather than full coordinate metadata provided by high-level coordinate objects.- Parameters:
axes (
int
orstr
, or multipleint
orstr
, optional) – Axis number in numpy ordering or unique substring ofndcube.NDCube.wcs.world_axis_physical_types
of axes for which real world coordinates are desired. axes=None implies all axes will be returned.pixel_corners (
bool
, optional) – IfTrue
then coordinates at pixel corners will be returned rather than at pixel centers. This increases the size of the output along each dimension by 1 as all corners are returned.wcs (
BaseHighLevelWCS
orExtraCoordsABC
, optional) – The WCS object to be used to calculate the world coordinates. Although technically this can be any valid WCS, it will typically beself.wcs
,self.extra_coords
, orself.combined_wcs
, combing both the WCS and extra coords. Defaults to the.wcs
property.
- Returns:
axes_coords (
tuple
ofQuantity
) – An iterable of raw coordinate values for all pixels for the requested axes. The returned units are determined by the WCS object. The dimensionality of these objects should match that of their corresponding array dimensions, unlesspixel_corners=True
in which case the length along each axis will be 1 greater than the number of pixels.
Examples
>>> NDCube.axis_world_coords_values('lat', 'lon') >>> NDCube.axis_world_coords_values(2)
- crop(*points, wcs=None)#
Crop using real world coordinates. This method crops the NDCube to the smallest bounding box in pixel space that contains all the provided world coordinate points.
This function takes the points defined as high-level astropy coordinate objects such as
SkyCoord
,SpectralCoord
, etc.- Parameters:
points (iterable of iterables) – Tuples of high level coordinate objects e.g.
SkyCoord
. Each iterable of coordinate objects represents a single location in the data array in real world coordinates.The coordinates of the points as they are passed to
world_to_array_index
. Therefore their number and order must be compatible with the API of that method, i.e. they must be passed in world order.wcs (
BaseHighLevelWCS
orExtraCoordsABC
) – The WCS to use to calculate the pixel coordinates based on the input. Will default to the.wcs
property if not given. While any valid WCS could be used it is expected that either the.wcs
or.extra_coords
properties will be used.
- Returns:
Examples
An example of cropping a region of interest on the Sun from a 3-D image-time cube: >>> point1 = [SkyCoord(-50*u.deg, -40*u.deg, frame=frames.HeliographicStonyhurst), None] # doctest: +SKIP >>> point2 = [SkyCoord(0*u.deg, -6*u.deg, frame=frames.HeliographicStonyhurst), None] # doctest: +SKIP >>> NDCube.crop(point1, point2) # doctest: +SKIP
- crop_by_values(*points, units=None, wcs=None)#
Crop using real world coordinates. This method crops the NDCube to the smallest bounding box in pixel space that contains all the provided world coordinate points.
This function takes points as iterables of low-level coordinate objects, i.e.
Quantity
objects. This differs fromcrop()
which takes high-level coordinate objects requiring all the relevant coordinate information such as coordinate frame etc. Hence this method’s API is more basic but less explicit.- Parameters:
points (iterable) – Tuples of coordinate values, the length of the tuples must be equal to the number of world dimensions. These points are passed to
wcs.world_to_array_index_values
so their units and order must be compatible with that method.units (
str
orUnit
) – If the inputs are set without units, the user must set the units inside this argument asstr
orUnit
objects. The length of the iterable must equal the number of world dimensions and must have the same order as the coordinate points.wcs (
BaseHighLevelWCS
orExtraCoordsABC
) – The WCS to use to calculate the pixel coordinates based on the input. Will default to the.wcs
property if not given. While any valid WCS could be used it is expected that either the.wcs
or.extra_coords
properties will be used.
- Returns:
Examples
An example of cropping a region of interest on the Sun from a 3-D image-time cube: >>> NDCube.crop_by_values((-600, -600, 0), (0, 0, 0), units=(u.arcsec, u.arcsec, u.s)) # doctest: +SKIP
- explode_along_axis(axis)#
Separates slices of NDCubes along a given axis into an NDCubeSequence of (N-1)DCubes.
- Parameters:
axis (
int
) – The array axis along which the data is to be changed.- Returns:
result (
ndcube.NDCubeSequence
)
- plot(axes=None, plot_axes=None, axes_coordinates=None, axes_units=None, data_unit=None, wcs=None, **kwargs)#
Visualize the
NDCube
.- Parameters:
axes (
WCSAxes
or None:, optional) – The axes to plot onto. If None the current axes will be used.plot_axes (
list
, optional) – A list of length equal to the number of pixel dimensions in array axis order. This list selects which cube axes are displayed on which plot axes. For an image plot this list should contain'x'
and'y'
for the plot axes andNone
for all the other elements. For a line plot it should only contain'x'
andNone
for all the other elements.axes_unit (
list
, optional) – A list of length equal to the number of world dimensions specifying the units of each axis, orNone
to use the default unit for that axis.axes_coordinates (
list
, optional) – A list of length equal to the number of pixel dimensions. For each axis the value of the list should either be a string giving the world axis type orNone
to use the default axis from the WCS.data_unit (
astropy.units.Unit
) – The data is changed to the unit given or theNDCube.unit
if not given.wcs (
astropy.wcs.wcsapi.BaseHighLevelWCS
) – The WCS object to define the coordinates of the plot axes.kwargs – Additional keyword arguments are given to the underlying plotting infrastructure which depends on the dimensionality of the data and whether 1 or 2 plot_axes are defined: - Animations:
mpl_animators.ArrayAnimatorWCS
- Static 2-D images:matplotlib.pyplot.imshow
- Static 1-D line plots:matplotlib.pyplot.plot
- rebin(bin_shape, operation=<function mean>, operation_ignores_mask=False, handle_mask=<function all>, propagate_uncertainties=False, new_unit=None, **kwargs)[source]#
Downsample array by combining contiguous pixels into bins.
Values in bins are determined by applying a function to the pixel values within it. The number of pixels in each bin in each dimension is given by the bin_shape input. This must be an integer fraction of the cube’s array size in each dimension. If the NDCube instance has uncertainties attached, they are propagated depending on binning method chosen.
- Parameters:
bin_shape (array-like) – The number of pixels in a bin in each dimension. Must be the same length as number of dimensions in data. Each element must be in int. If they are not they will be rounded to the nearest int.
operation (function) – Function applied to the data to derive values of the bins. Default is
numpy.mean
operation_ignores_mask (
bool
) – Determines how masked values are handled. If False (default), masked values are excluded when calculating rebinned value. If True, masked values are used in calculating rebinned value.handle_mask (
None
or function) – Function to apply to each bin in the mask to calculate the new mask values. IfNone
resultant mask isNone
. Default isnumpy.all
propagate_uncertainties (
bool
or function.) – If False, uncertainties are dropped. If True, default algorithm is used (propagate_rebin_uncertainties
) Can also be set to a function which performs custom uncertainty propagation. Additional kwargs provided to this method are passed onto this function. See Notes section on how to write a custompropagate_uncertainties
function.new_unit (
astropy.units.Unit
, optional) – If the rebinning operation alters the data unit, the new unit can be provided here.kwargs – All kwargs are passed to the error propagation function.
- Returns:
new_cube (
NDCube
) – The resolution-degraded cube.
Notes
Rebining Algorithm Rebinning is achieved by reshaping the N-D array to a 2N-D array and applying the function over the odd-numbered axes. To demonstrate, consider the following example. Let’s say you have an array:
x = np.array([[0, 0, 0, 1, 1, 1], [0, 0, 1, 1, 0, 0], [1, 1, 0, 0, 1, 1], [0, 0, 0, 0, 1, 1], [1, 0, 1, 0, 1, 1], [0, 0, 1, 0, 0, 0]])
and you want to sum over 2x2 non-overlapping sub-arrays. This summing can be done by reshaping the array:
y = x.reshape(3,2,3,2)
and then summing over the 1st and third directions:
y2 = y.sum(axis=3).sum(axis=1)
which gives the expected array:
array([[0, 3, 2], [2, 0, 4], [1, 2, 2]])
Defining Custom Error Propagation To perform custom uncertainty propagation, a function must be provided via the propgate_uncertainty kwarg. This function must accept, although doesn’t have to use, the following args:
- uncertainty:
astropy.nddata.NDUncertainty
but notastropy.nddata.UnknownUncertainty
The uncertainties associated with the data.
- data: array-like
The data associated with the above uncertainties. Must have same shape as uncertainty.
- mask: array-like of
bool
orNone
Indicates whether any uncertainty elements should be ignored in propagation. True elements cause corresponding uncertainty elements to be ignored. False elements cause corresponding uncertainty elements to be propagated. Must have same shape as above. If None, no uncertainties are ignored.
All kwarg inputs to the rebin method are also passed on transparently to the propagation function. Hence additional inputs to the propagation function can be included as kwargs to
ndcube.NDCube.rebin()
.The shape of the uncertainty, data and mask inputs are such that the first dimension represents the pixels in a given bin whose data and uncertainties are aggregated by the rebin process. The shape of the remaining dimensions must be the same as the final rebinned data. A silly but informative example of a custom propagation function might be:
def my_propagate(uncertainty, data, mask, **kwargs): # As a silly example, propagate uncertainties by summing those in same bin. # Note not all args are used, but function must accept them. n_pixels_per_bin = data.shape[0] # 1st dimension of inputs gives pixels in bin. final_shape = data.shape[1:] # Trailing dims give shape of put rebinned data. # Propagate uncerts by adding them. new_uncert = numpy.zeros(final_shape) for i in range(n_pixels_per_bin): new_uncert += uncertainty.array[i] # Alternatively: new_uncerts = uncertainty.array.sum(axis=0) return type(uncertainty)(new_uncert) # Convert to original uncert type and return.
- reproject_to(target_wcs, algorithm='interpolation', shape_out=None, return_footprint=False, **reproject_args)#
Reprojects the instance to the coordinates described by another WCS object.
- Parameters:
target_wcs (
astropy.wcs.wcsapi.BaseHighLevelWCS
,astropy.wcs.wcsapi.BaseLowLevelWCS
,) – orastropy.io.fits.Header
The WCS object to which thendcube.NDCube
is to be reprojected.algorithm ({‘interpolation’ | ‘adaptive’ | ‘exact’}) – The algorithm to use for reprojecting. When set to “interpolation”
reproject_interp
is used, when set to “adaptive”reproject_adaptive
is used and when set to “exact”reproject_exact
is used.shape_out (
tuple
, optional) – The shape of the output data array. The ordering of the dimensions must follow NumPy ordering and not the WCS pixel shape. If not specified,array_shape
attribute (if available) from the low level API of thetarget_wcs
is used.return_footprint (
bool
) – IfTrue
the footprint is returned in addition to the newNDCube
. Defaults toFalse
.**reproject_args – All other arguments are passed through to the reproject function being called. The function being called depends on the
algorithm=
keyword argument, see that for more details.
- Returns:
reprojected_cube (
ndcube.NDCube
) – A new resultant NDCube object, the suppliedtarget_wcs
will be the.wcs
attribute of the outputNDCube
.footprint (
numpy.ndarray
) – Footprint of the input array in the output array. Values of 0 indicate no coverage or valid values in the input image, while values of 1 indicate valid values.
Notes
This method doesn’t support handling of the
mask
,extra_coords
, anduncertainty
attributes yet. However,meta
andglobal_coords
are copied to the outputndcube.NDCube
.
- to(new_unit, **kwargs)[source]#
Convert instance to another unit.
Converts the data, uncertainty and unit and returns a new instance with other attributes unchanged.
- Parameters:
new_unit (
astropy.units.Unit
) – The unit to convert to.kwargs – Passed to the unit conversion method, self.unit.to.
- Returns:
NDCube
– A new instance with the new unit and data and uncertainties scales accordingly.