How to read a wind rose (and use it properly in wind project assessment)

A wind rose is a polar chart that shows how often the wind blows from each direction at a site, and usually how strong it is when it does. If you assess wind projects, it's the first plot you look at and the one most people misread. This guide covers what the petals actually mean, how to build a rose from raw met-mast or reanalysis data, and three real wind roses from anonymised Vind projects showing how the shape of the rose changes everything downstream.

A wind rose shows the frequency of wind from each direction (petal length) and often the speed distribution within each direction (petal colour bands). Longer petal = wind comes from there more often, not that it's windier from there. Read it to find your prevailing direction, sanity-check sensor calibration, and inform turbine layout. The single most common mistake is confusing frequency with energy — and that mistake quietly costs yield.

‍

What a wind rose is — and what each part means

A wind rose is a polar chart that shows the frequency distribution of wind direction at a location, almost always with wind speed layered on top as colour. It answers one question fast: from which directions does the wind come, and how often? Mariners used the same idea on nautical charts centuries before it became a standard plot in wind resource assessment.

Three components carry the information, and reading the rose means reading all three at once:

• Petals (radial spokes) point in the direction the wind comes from. A petal pointing west means west winds. Most roses use 16 sectors of 22.5° each (N, NNE, NE, ENE…), though 8, 12, or 36 sectors are all used depending on data resolution.
• Petal length is the frequency — the percentage of time the wind blew from that sector. The concentric rings are the scale for that percentage.
• Colour bands within each petal split the speed distribution into bins (e.g. 0–3, 3–6, 6–9 m/s). A petal that's long but mostly pale tells a very different story from a petal that's short but mostly dark.

The key mental model: a wind rose is a 2D histogram wrapped into a circle. One axis is direction, the other is frequency, and colour adds the speed dimension. Once you see it that way, the common misreadings stop happening.

‍

How to read one without falling into the frequency trap

To read a wind rose, find the longest petals first — that's your prevailing wind direction, the directions worth most of your attention. Then read the colour inside those petals to understand whether the frequent wind is also strong wind. Those are two separate questions, and conflating them is the mistake that matters.

Here's the trap in one sentence: petal length is frequency, not strength. A site can have a long, pale north petal (wind comes from the north often, but gently) and a short, dark south-west petal (rare wind, but when it comes it's strong). For energy — which scales with the cube of wind speed — that short dark south-west petal may matter far more than the long pale northern one. A frequency rose alone won't tell you that; you need the speed bins, or better, an energy rose that weights each sector by power.

A practical reading sequence:

1. Orient yourself. Confirm the rose shows direction wind comes from (standard) and note the sector count.
2. Find the prevailing direction(s). The longest one or two petals.
3. Read the colour inside them. Frequent and dark = your bread-and-butter energy direction.
4. Scan for a secondary mode. Many sites have a bimodal rose (two opposing prevailing directions) — important for wake and layout.
5. Sanity-check the symmetry. A rose that's suspiciously offset from expected regional patterns can be a calibration flag (more on that below).

How to build a wind rose from raw data

Building a wind rose from raw data is a four-step pipeline: collect paired direction-and-speed records, bin them into directional sectors, count the frequency per sector, then plot. Each step is simple; the accuracy lives in the details.

Step 1 — Collect direction and speed data

Gather time-series records of wind direction and speed, each tagged with a timestamp. Sources are met-mast anemometers and wind vanes, LiDAR/SoDAR, or long-term reanalysis datasets (e.g. ERA5) when on-site data is short. Capture at least a full year to cover the seasonal cycle — anything shorter risks a rose that reflects one season, not the site.

Step 2 — Bin into directional sectors

Divide the 360° compass into discrete sectors — typically 16 sectors of 22.5° — and assign each record to the sector its direction falls in. Decide your bin edges deliberately: a common convention centres the first bin on 0° (north), so the N sector spans 348.75°–11.25°. Be consistent, because off-by-half-a-bin errors quietly distort the whole rose.

Step 3 — Calculate frequency per sector

For each sector, count the records that fall inside it and divide by the total number of valid records. That percentage is the petal length. If you're building a speed-resolved rose, do the same within each speed bin per sector, so each petal carries its internal colour breakdown. This is also where you'd fit a Weibull distribution per sector if you want to model the speed behaviour rather than just histogram it.

Step 4 — Plot the rose

Draw the polar chart: one petal per sector, length set by frequency, colour bands set by the per-sector speed distribution, concentric rings as the frequency scale. Tools range from Python (Matplotlib, the windrose library) and R (openair) to dedicated WRA platforms. The plotting is the easy part — by this stage the analytical decisions are already baked in.

Three roses: how site choice changes the shape

‍

Common pitfalls when interpreting a wind rose

Most wind-rose mistakes aren't about plotting — they're about interpretation. These four come up again and again in screening reviews.

• Reading frequency as strength. The longest petal is the most frequent direction, not the windiest or the highest-energy one. Always read the speed colour bands, and for layout decisions weight by energy (∝ speed³), not by frequency. This is the single most expensive misread.
• Too short a record. A rose built on three months captures one season's pattern and presents it as the site's character. Use at least a full year; for bankable assessment, correlate to a long-term reference. A short record produces a confident-looking rose that's quietly wrong.
• Uncorrected sensor offset. A vane misaligned by even 10–15° shifts every record into the wrong sector and rotates the whole rose. If your rose is oddly offset from the regional pattern, suspect calibration before you suspect the site. Cross-check against reanalysis or a second sensor.
• Ignoring data gaps and filtering. Sensor icing, downtime, or unfiltered tower-shadow sectors skew the frequencies. Document your data availability and flag any direction sectors affected by mast distortion before reading the rose as truth.

Summary

A wind rose tells you where your wind comes from and how often, and — read with its speed bands — whether the frequent directions are also the energetic ones. Read it in that order, build it on a full year of data, and treat an oddly offset rose as a calibration flag, not a site feature. The one habit that separates a good reading from a costly one: never confuse a long petal with a strong one.

If you'd rather not hand-build roses from raw time-series, Vind generates speed-resolved and energy wind roses in seconds from your site data as part of resource screening. See how it fits into a full assessment workflow →

‍