Wind detection

The wind shown in SportsTrackLive is not taken from a weather model — it is derived from the flight itself. This page explains how it is computed and how it is presented as wind segments along the track.

How the wind is computed

When a glider, a bird, or a paraglider circles in a thermal, its GPS ground-speed vector (speed + heading) is constantly being pushed by the wind. Over a full circle, the tip of that velocity vector traces a circle whose centre is offset from zero by exactly the wind vector:

• the distance of the circle's centre from the origin is the wind speed;
• the direction of that centre (rotated 180°) is the wind bearing.

For every GPS fix, we convert the ground-speed vector into a point on this velocity plane and fits a circle through the accumulated points using circular regression (a Hyper-fit least-squares circle).

The fitted centre gives the instantaneous wind.

Fixes are filtered before they are used so that only clean circling data feeds the estimate — fixes with strong acceleration, a high turn rate, large altitude change (strong climb/sink) or with less than ~90° of arc completed are rejected or only used to "clean" the sample. The centre is then low-pass filtered from frame to frame for stability, and each estimate carries a quality score; only estimates above the quality threshold are kept.

The mean wind for the flight is the circular average of all the accepted per-fix bearings, together with the average speed — see Free-flight analytics.

Wind segments

A single flight rarely flies through one constant wind: the wind backs and veers and changes strength with altitude and terrain. Rather than collapse the whole flight to one arrow, SportsTrackLive splits it into wind segments — successive stretches of the track over which the wind was roughly steady.

• The smoothed per-fix wind is monitored continuously. A new segment starts whenever the mean bearing turns by more than a set threshold, or the mean speed changes by more than a set threshold.
• Adjacent segments that turn out to be similar (within half of those thresholds) are merged back together, so the result is a small number of meaningful segments rather than dozens of fragments.
• Each segment carries its own bearing, speed and quality, plus its start/end position and time, and is drawn along the track — letting you see, for example, a wind shift as the pilot climbed from valley level into the free-stream above the ridge.

In the player

The wind indicator and beacon stations on the 3D globe display the wind computed here at the current playback moment, and the wind segments are drawn along the track.