HEALPix — plotting methods

HEALPix (Hierarchical Equal Area isoLatitude Pixelisation) divides the sphere into equal-area pixels arranged in rings of constant latitude. It is used as the native output grid of the ECMWF AIFS machine-learning forecast model.

Because HEALPix data is stored as a 1-D array of unordered pixel values, different plotting methods make different trade-offs between accuracy and speed:

Method	How it works	Best for
grid_cells	Draws each pixel as its true diamond-shaped polygon	Seeing exact cell geometry
point_cloud	Plots a coloured marker at each pixel centre	Quick inspection, sparse data
contourf	Interpolates to a regular grid then draws filled contours	Smooth, publication-quality maps

This notebook loads a HEALPix H128 2-metre temperature field and shows all three methods side-by-side.

import earthkit.data as ekd

import earthkit.plots as ekp

data = ekd.from_source("sample", "healpix-h128-nested-2t.grib")

Shared style

We define a single Style object and reuse it across all subplots so the colour scale is identical in every panel.

style = ekp.styles.Style(levels=range(0, 21), colors="Spectral_r", units="celsius")

All methods side-by-side

Each subplot uses the same domain (France and Spain) so the different cell geometries and interpolation effects are easy to compare.

figure = ekp.Figure(rows=2, columns=2, domain=["France", "Spain"])

for method in ["point_cloud", "grid_cells", "contourf", "grid_points"]:
    subplot = figure.add_map()
    getattr(subplot, method)(data, style=style)
    subplot.title(method)

figure.coastlines()
figure.gridlines()

figure.title("Plotting HEALPix grid data with various methods")

figure.legend(location="right")

figure.show()

What to notice

• ``grid_cells`` shows the true HEALPix pixel boundaries — the distinctive diamond shapes that give HEALPix its equal-area property.

• ``point_cloud`` places one dot per pixel centre. At this resolution (H128 ≈ 0.46°) the dots are close enough to fill in, but zooming out would reveal gaps.

• ``contourf`` interpolates to a regular grid before contouring, producing a smooth result at the cost of some spatial accuracy near sharp gradients.