Here's a brief explanation of the physical causes behind lightning in a supercell.
1.) Strong updrafts are necessary to lift precipitation particles into the "mixed phase" region (typically -10 to -20 C) where ice crystals, graupel (soft hail particles) and supercooled water droplets (water that exists in temperatures below 0 C) can co-exist.
2.) The collision of riming graupel (graupel with a thin layer of water) and ice crystals exchanges charge -- this is known as the Non-inductive charging mechanism and is now widely accepted.
3.) Wind shear and the particles' differential sizes re-distribute these ice crystals and graupel to different parts of the storm building up a charge potential.
4.) Once the electrical breakdown threshold (~100,000 V/m) is reached, lightning occurs to "re-balance" the charge structure and environment of the supercell.
Climatology shows that ~ 70% of lightning are in-cloud flashes. In-cloud lightning almost always proceeds cloud-to-ground lightning. The initial charge separation usually occurs in this mixed phase region, atleast 4-6 km above ground level -- this promotes in-cloud flashes. Later in a supercell's lifecycle the updrafts weaken and the charge carrying precipitation particles are allowed to descend where they interact with charge potential differences at the surface -- leading to cloud-to-ground strikes.
The typical schematic of a supercell will see: Peak in in-cloud flashes ---> 5 to 15 min. later Peak in cloud-to-ground flashes ---> 5 to 15 min. later Peak in rainfall rate. Since a supercell is a self-sustaining entity it will go thru cycles of lightning activity and rainfall rates.
New sensing technology known as "Lightning Mapping Arrays" use a network of 7 to 12 sensors usually to passively receive the VHF (similar to television frequencies: 50-100 MHz) sources that the electrical breakdown associated with lightning emit. These can be mapped in 3D using time-of-arrival techniques or interferometry to pin-point the locations and paths of these in-cloud flashes, adding large amounts of data to the NLDN cloud-to-ground strike data which is extremely sparse in comparison.
Here is my paper I'm presenting at the AMS Radar Conf. later this month that shows ways in which "total lightning" data can be used in a complementary fashion with radar for use in forecasting and identification of storm features.
http://ams.confex.com/ams/32Rad11Meso/tech...paper_97291.htm
