Reduced Gaussian grid - plotting methods

The reduced Gaussian grid (reduced_gg) is the native model grid used by ECMWF’s IFS. Unlike a regular lat-lon grid, the number of points per latitude circle decreases towards the poles, keeping the physical spacing between grid points roughly constant across the globe. The octahedral variant (prefix O) is used by modern IFS versions.

This irregular point distribution means that naive plotting (e.g. reshaping to a 2-D array) does not work without interpolation. earthkit-plots handles this transparently for all methods:

Method	How it works	Best for
grid_cells	Draws each grid point as its Voronoi cell polygon	Seeing exact cell geometry
point_cloud	Plots a coloured marker at each grid point	Quick inspection
contourf	Interpolates to a regular grid then draws filled contours	Smooth maps

This notebook uses an O32 reduced Gaussian 2-metre temperature field (low resolution so the individual cells are clearly visible).

import earthkit.data as ekd

import earthkit.plots as ekp

data = ekd.from_source(
    "url",
    "https://get.ecmwf.int/repository/test-data/earthkit-regrid/test-data/global_0_360/O32.grib",
)

Shared style

A single Style object keeps the colour scale consistent across all panels.

style = ekp.styles.Style(levels=range(10, 31), colors="Spectral_r", units="celsius")

All methods side-by-side

We zoom to the Arctic where the reduced nature of the grid — fewer points near the pole — is most obvious.

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 reduced Gaussian grid data with various methods")

figure.legend(location="right")

figure.show()

What to notice

• ``grid_cells`` reveals the variable cell width at different latitudes — wider cells near the equator, narrower (and fewer) cells near the poles. This is the defining characteristic of the reduced Gaussian grid.

• ``point_cloud`` makes the irregular point spacing clear: rows near the pole have noticeably fewer dots than rows near the equator.

• ``contourf`` hides the underlying grid structure entirely, presenting a smooth interpolated field. This is often the right choice for final publication figures, but grid_cells is better for diagnosing data quality.