astropy-xarray

Seamless interoperability between astropy and xarray.

Simple Usage

import astropy_xarray
import xarray as xr

ds = xr.Dataset({"a": ("x", [0, 1, 2]), "b": ("y", [-3, 5, 1], {"units": "m"})})
ds

output:

<xarray.Dataset> Size: 48B
Dimensions:  (x: 3, y: 3)
Dimensions without coordinates: x, y
Data variables:
    a        (x) int64 24B 0 1 2
    b        (y) int64 24B -3 5 1

To convert the variables of a Dataset to quantities, use accessor .astropy.quantify():

q = ds.astropy.quantify(a="s")
q

output:

<xarray.Dataset> Size: 48B
Dimensions:  (x: 3, y: 3)
Dimensions without coordinates: x, y
Data variables:
    a        (x) float64 24B [s] 0.0 1.0 2.0
    b        (y) float64 24B [m] -3.0 5.0 1.0

to convert to different units, use accessor .astropy.to():

c = q.astropy.to({"a": "ms", "b": "km"})
c

output:

<xarray.Dataset> Size: 48B
Dimensions:  (x: 3, y: 3)
Dimensions without coordinates: x, y
Data variables:
    a        (x) float64 24B [ms] 0.0 1e+03 2e+03
    b        (y) float64 24B [km] -0.003 0.005 0.001

to convert back to non-quantities for portability, use accessor .astropy.dequantify():

d = c.astropy.dequantify()
d

output:

<xarray.Dataset> Size: 48B
Dimensions:  (x: 3, y: 3)
Dimensions without coordinates: x, y
Data variables:
    a        (x) float64 24B 0.0 1e+03 2e+03
    b        (y) float64 24B -0.003 0.005 0.001

SkyCoord Usage

To convert a astropy.skyCoord to a Dataset, use skycoord_to_dataset().

import astropy.units as u
from astropy.coordinates import ICRS, SkyCoord
from astropy.time import Time

from astropy_xarray.coordinates.sky_coord import (
    skycoord_to_dataset,
)

sc = skycoord_to_dataset(
    SkyCoord(
        ra=[[2, 6, 7, 4]] * u.deg,
        dec=[[4, 7, 4, 3]] * u.deg,
        pm_ra_cosdec=[[1, 1, 1, 1]] * u.mas / u.yr,
        pm_dec=[[1, 1, 1, 1]] * u.mas / u.yr,
        frame="icrs",
    ),
    coords={
        "timestamp": ("time", Time([1.7e9], format="unix")),
        "field_label": ("field", ["a", "b", "c", "d"]),
    },
)
sc

output:

<xarray.Dataset> Size: 152B
Dimensions:       (time: 1, field: 4)
Coordinates:
    timestamp     (time) float64 8B [utc unix] 2023-11-14T22:13:20.000000000
    field_label   (field) <U1 16B 'a' 'b' 'c' 'd'
Dimensions without coordinates: time, field
Data variables:
    ra            (time, field) float64 32B [°] 2.0 6.0 7.0 4.0
    dec           (time, field) float64 32B [°] 4.0 7.0 4.0 3.0
    pm_ra_cosdec  (time, field) float64 32B [mas yr⁻¹] 1.0 1.0 1.0 1.0
    pm_dec        (time, field) float64 32B [mas yr⁻¹] 1.0 1.0 1.0 1.0
Attributes:
    frame:    {'name': 'icrs', 'representation_type': 'spherical', 'different...

To convert SkyCoord-like datasets back to astropy types, use .astropy.to_skycoord():

sc.astropy.to_skycoord()

output:

<SkyCoord (ICRS): (ra, dec) in deg
    [[(2., 4.), (6., 7.), (7., 4.), (4., 3.)]]
 (pm_ra_cosdec, pm_dec) in mas / yr
    [[(1., 1.), (1., 1.), (1., 1.), (1., 1.)]]>

For more, see the documentation