When are thermals not circular and do thermal helpers assume thatthey are?

On Tuesday, July 30, 2013 1:54:13 AM UTC+1, jfitch wrote:
Thermals may not be circular, but your glider flies in a pretty good approximation of a circle. So the cylindrical graph depicted in Winpilot is all the information you can really use. XCSoar's polar graph might fool you into thinking that is the shape of the lift - but it is a polar graph of strength around a fixed radius circle, not a map of the thermal. Oh yeah they are different!

'fixed radius circle' implies to me that XCSoar's assistant is trying to estimate the lift over a fixed area - eg a 250m wide circle. Just to clarify: XCSoar's thermal assistant is simply a 36-point plot of the lift recorded vs heading over the last turn. While your heading lies within a single 10 degree block, one of these points is continuously updated with your instantaneous vario value. When your heading crosses into the next 10 degree block, the next point is updated, and so on.

In practical terms this means that any sections where you hold a constant heading are ignored on the plot. As an extreme example, imagine you're flying a racetrack pattern between two cores. You've turned the first 180 degrees in the first core and have just completed the remaining 180 degrees in the second core. The top half of the thermal assistant would show the lift encountered during your 180-degree turn in the first core, and the bottom half would show the result of your 180-degree turn in the second core. Any lift or sink you encountered on the straight leg between the first and second cores would not be represented on the plot.

The line in the middle of the plot just points to the centre of the polygon..