Initializing WCSAxes with custom transforms¶
In Initializing axes with world coordinates, we saw how to make plots using
WCS
objects. However, the
WCSAxes
class can also be initialized
with more general transformations that don’t have to be represented by the
WCS
class. Instead, you can initialize
WCSAxes
using a Matplotlib
Transform
object and a dictionary
(coord_meta
) that provides metadata on how to interpret the transformation.
The Transform
should represent the conversion
from pixel to world coordinates, and should have input_dims=2
and
output_dims=2
. In addition, has_inverse
should be set to True
and
the inverted
method should be implemented.
The coord_meta
dictionary should include the following keys:
name
: an iterable of two strings giving the names for each dimensiontype
: an iterable of two strings that should be either'longitude'
,'latitude'
, or'scalar'
(for anything that isn’t a longitude or latitude).wrap
: an iterable of two values which indicate for longitudes at which angle (in degrees) to wrap the coordinates. This should beNone
unlesstype
is'longitude'
.unit
: an iterable of twoUnit
objects giving the units of the world coordinates returned by theTransform
.format_unit
: an iterable of twoUnit
objects giving the units to use for the formatting of the labels. These can be set toNone
to default to the units given inunit
, but can be set for example if theTransform
returns values in degrees and you want the labels to be formatted in hours.
The following example illustrates a custom projection using a transform and
coord_meta
:
from astropy import units as u
import matplotlib.pyplot as plt
from matplotlib.transforms import Affine2D
from astropy.visualization.wcsaxes import WCSAxes
# Set up an affine transformation
transform = Affine2D()
transform.scale(0.01)
transform.translate(40, -30)
transform.rotate(0.3) # radians
# Set up metadata dictionary
coord_meta = {}
coord_meta['name'] = 'lon', 'lat'
coord_meta['type'] = 'longitude', 'latitude'
coord_meta['wrap'] = 180, None
coord_meta['unit'] = u.deg, u.deg
coord_meta['format_unit'] = None, None
fig = plt.figure()
ax = WCSAxes(fig, [0.1, 0.1, 0.8, 0.8], aspect='equal',
transform=transform, coord_meta=coord_meta)
fig.add_axes(ax)
ax.set_xlim(-0.5, 499.5)
ax.set_ylim(-0.5, 399.5)
ax.grid()
ax.coords['lon'].set_axislabel('Longitude')
ax.coords['lat'].set_axislabel('Latitude')
()