[EMAIL PROTECTED] wrote: > What happens if there are 2+ `storms' (or storm centres or whatever you > want to call them). If they are nice and far apart, e.g. 5+ miles, you > can be confident that the readings won't get too badly munged as sound > starts to fade about that distance. However can anyone suggest a > general algorithm for identifying which readings are related? At the > minute I'm playing around with the idea of trying to get storm > hotspots that are not near to each other and then just attributing > readings to the ones that fit best.
Yes, this method of storm tracking gets discussed every time thers is one :) One possibility would be, instead of trying to computationally derive the location of each storm, to simply draw circles on the map centered at the position of each observer and of the range indicated by the flash/bang difference. The viewer could then use their on-board wetware to draw their own conclusions. This would also have the advantage of providing something to watch even if there's only a couple of observers (not enough to get a fix), and would be more useful than calculated positions which might be quite wrong (it would also give you an idea of the expected error). These circles could fade in colour over time to expire data and give people an idea of storm movements. aef