Wind Rose Charts in Python

A wind rose (also called a wind rose chart, wind rose plot, or wind rose diagram) summarizes how often observations arrive from each compass direction and how those observations are distributed across speed bands. XY bins the raw direction/speed pairs in Python and renders the result as stacked polar bars.

Use wind_rose() when you have one bearing and one magnitude per observation. If the directional counts are already aggregated, use a radial bar chart instead.

Jump to directional sectors, speed bands, or the input contract.

Directions are compass bearings in degrees: 0° is north and values increase clockwise. The helper applies that convention automatically:

Each stacked color band counts observations whose speed is above the previous edge and at or below the current edge. The legend labels show those inclusive upper edges.

sectors= controls the number of equally sized angular bins and must be at least 3. Each bin is centered on its compass bearing, so a value of exactly 0° belongs to the sector centered on north. Bearings outside 0..360 wrap around the circle.

More sectors reveal directional detail but need more observations to keep each bin stable. Common choices are 8, 12, 16, or 36 sectors, depending on sample size and the directional resolution of the source.

Pass increasing upper edges through speed_bins=:

The final edge should cover the fastest observation; values above it do not belong to a displayed band. XY removes duplicate edges and orders the remaining values. When speed_bins is omitted, it derives up to four readable bands from the speed quartiles and rounds the top edge upward so every finite observation is covered.

Lower the sector count when the sample is small or the source records only the eight principal bearings, and pass matching band edges so each petal stays thick enough to read:

With thousands of records a high sector count resolves the prevailing wind, and narrow speed bands separate calm air from gales — add xy.legend() to place the band labels where you want them:

• directions and speeds must have the same length.
• Each pair describes one observation.
• Non-finite pairs are dropped together.
• At least one finite pair must remain.
• sectors must be 3 or greater.
• speed_bins must contain at least one edge when supplied.

XY raises ValueError for mismatched arrays, an empty finite dataset, too few sectors, or an empty band definition.

The radial value is a count, not a speed. Each speed band becomes one stacked bar series and takes the next chart palette color. Pass chart keyword props such as title, width, height, padding, class_names, and styles through wind_rose().

The helper authors a degree-based theta axis with north at zero and clockwise rotation, plus an r axis labeled count. Build the equivalent sectors manually with polar_bar_chart() when you need custom pre-binning, non-count radial values, a different angular convention, or component children such as xy.theme(), xy.legend(), and xy.modebar().

Wind roses support hover, fixed-minimum radial zoom, reset, and browser/static export through the shared polar renderer. Theta rotation, box zoom, selection, brushing, and crosshairs are not available.

See the polar overview for the full interaction, renderer, annotation, and large-data boundary.

• Polar overview — numeric theta/r data and shared polar axes.
• Radial bar charts — pre-aggregated annular sectors.
• Pie and donut charts — share, progress-ring, revenue-mix, and gauge blocks.
• Radar charts — category profiles rather than directional frequencies.

wind_rose() always accepts compass bearings in degrees. It applies north-zero, clockwise theta settings automatically.

XY derives up to four bands from the finite speed quartiles and rounds the final edge upward to include the maximum observation.

Non-finite direction/speed pairs are dropped before the wind rose graph is binned. With authored speed_bins, make sure the final upper edge covers the fastest finite observation.