Bilinear interpolation

Bilinear interpolation resamples data onto a regular pixel grid by computing a weighted average of the four surrounding source values for each output pixel. The result is a smooth, continuous-looking field with no visible cell boundaries — well suited to filled contours and shaded plots where a visually clean output is desired.

Use Bilinear when:

• your source data is on a regular lat/lon (or other rectilinear) grid, and

• you want smooth, interpolated rendering rather than sharp grid-cell edges.

For unstructured or native HEALPix/reduced-Gaussian grids, combine Bilinear with Regrid via a Chain (covered in the Regrid notebook).

Example: 2 m temperature over Europe

We will use a sample GRIB file containing 2 m temperature on a regular lat/lon grid.

import earthkit.data as ekd

import earthkit.plots as ekp

data = ekd.from_source("sample", "test.grib").to_fieldlist()
data.ls()

	parameter.variable	time.valid_datetime	time.base_datetime	time.step	vertical.level	vertical.level_type	ensemble.member	geography.grid_type
0	2t	2020-05-13 12:00:00	2020-05-13 12:00:00	0 days	0	surface	0	regular_ll
1	msl	2020-05-13 12:00:00	2020-05-13 12:00:00	0 days	0	surface	0	regular_ll

Default pixel count

Calling Bilinear() with no arguments uses 1000 × 1000 pixels — a good default for most maps. The pixel grid is always aligned with the map’s coordinate reference system (CRS), not the source grid.

from earthkit.plots.resample import Bilinear

chart = ekp.Map(domain="Europe")

chart.contourf(data, resample=Bilinear())

chart.coastlines()
chart.gridlines()
chart.legend()

chart.show()

Controlling the pixel count

Pass a single integer to set both nx and ny to the same value, or use nx/ny keyword arguments to specify them independently. Lowering the count reveals the individual pixels; raising it gives a finer result.

figure = ekp.Figure(rows=1, columns=2, domain="Europe")

# Coarse — individual pixels are visible
ax = figure.add_map()
ax.contourf(data, resample=Bilinear(50))
ax.title("Bilinear(50)")

# Fine — very smooth appearance
ax = figure.add_map()
ax.contourf(data, resample=Bilinear(500))
ax.title("Bilinear(500)")

figure.coastlines()
figure.legend(location="right")

figure.show()

Resolution-based specification

Instead of a fixed pixel count, you can specify the pixel spacing in degrees using Bilinear.at_resolution(dx). The pixel count then adapts automatically to the map extent — useful when you want consistent detail at different zoom levels.

chart = ekp.Map(domain="Europe")

# One output pixel per 0.5 degrees
chart.contourf(data, resample=Bilinear.at_resolution(0.5))

chart.coastlines()
chart.gridlines()
chart.legend()

chart.show()

NOTE: Unlike Subsample, Bilinear can upsample — you can request more pixels than source grid points. The interpolated values are estimated from the surrounding source data.