xrframes

Similar to (and inspired by) jbusecke/xmovie, make animations from xarray objects by applying a matplotlib-based plotting function along a dimension.

Install

pip install xrframes

Install and use Dask¹ or joblib² to parallelize the frame creation step.

.to_gif() uses ImageMagick (magick or convert required on PATH); .to_mp4() uses FFmpeg (ffmpeg required on PATH). Both are available via conda-forge³, as well as other package managers.

Example

Basic example, using just the xarray object's plot method:

import xarray as xr
from xrframes import Frames

# Note: `pooch` required
ta = xr.tutorial.open_dataset("air_temperature").air.isel(time=slice(0, 10))

frames = Frames(ta, lambda da: da.plot(size=2.5, aspect=1.7), dim="time")
frames.write(dpi=120)  # serial
frames.to_gif("./ta_basic.gif", fps=5)

A bit fancier example with the same data, using Cartopy:

import cartopy.crs as ccrs
import cartopy.feature as cfeature
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import xarray as xr
from xrframes import Frames

# Note: `pooch` required
ta = xr.tutorial.open_dataset("air_temperature").air.isel(time=slice(0, 10))

proj = ccrs.Mercator()
tran = ccrs.PlateCarree()

def plot(da, **kwargs):
    fig, ax = plt.subplots(figsize=(6, 3), layout="constrained", subplot_kw=dict(projection=proj))

    ax.coastlines()
    ax.add_feature(cfeature.BORDERS)
    ax.add_feature(cfeature.STATES)
    ax.gridlines(draw_labels=True)

    da.plot.contourf(**kwargs, levels=np.arange(230, 305, 5), extend="both", ax=ax, transform=tran)

    ax.set_title("")
    ax.set_title(pd.Timestamp(da.time.item()).strftime(r"%Y-%m-%d %H:%M"), loc="left", size=10)

    return fig

frames = Frames(ta, plot, dim="time")
frames.write()  # serial
frames.to_mp4("./ta_cartopy.mp4", fps=5)