Sick shots dude...probably the best lightning shots I've ever seen.
I'm going to speculate based on what I know of lightning polarity, that these are almost certainly all positive strikes, which possess roughly 10x the electrical current that negative strikes do, and therefore they would probably also make a hell of a noise. These type of strikes also only account for 10% of all CG strikes, and usually occur in severe (supercell) storms, in the trailing regions of QLCSs, near mountain tops or in winter (when the + region is closer to the earth). Note to self - when watching equatorial thunderstorms don't be outside within 10 miles of the storm!
In the usual thunderstorm "dipole", the top of the storm is positively charged (where the cloud consists mainly of ice crystals), while the main bulk of the storm below is negatively charged. An induced positive charge at the surface also follows along under the cloud like a shadow, which is where the majority of negative CGs occur. This is also why positive strikes don't tend to strike directly beneath a thunderstorm - instead they are very tall (explaining the strength of the current), originating from high in the cloud and then outward some distance from the cloud, where the ground has its usual negative charge again after the storm passes.
I am willing to bet this is the reason we see the "dog leg" shape, due to the current of the stroke flowing around the positively charged ground/electric field below the storm. Sure, the majority of discharges probably occur as IC strokes between the + and - region, but in a super strong updraft tower when charge separation is occurring at its most intense (as you mentioned when they tend to happen), a few zots shoot out and away from the storm. It probably has to do with updraft strength, or even sheer depth of convection (which in the tropics can be an excess of 60Kft) in tropical thunderstorms. Again, just long-winded speculation!
