Hi
The streamer tends to rotate when it reaches the lateral part of the sphere, so I believe that it has something to do with the rising hot current air from the spark.
http://www.youtube.com/watch?v=nj0ARi5JTgc
Higher break rates will definetively rise the spark channel temperature, so I suppose that the "tornado effect" can be done easier. One solution would be to rectify the NSTs and run it with a varible BPS rotary spark gap (I dont have one).
Another idea is to glue a little permanent magnet on the internal wall of the sphere, so the particles, no matter the moving direction, will have spiral movements. That will be an idea to experiment in the future..maibe..
Another idea is to have a sectionated sphere, with a laminar air flow through the thin section (on the upper 1/3 of the sphere) to create a lower pressure zone on the top of the sphere (through Coanda effect). I suppose that the streamer formation would be localised on this lower pressure zone (top of the sphere), so the single streamer can be emitted permanently on the same region of the sphere.