Great product! I wonder how they test this product...in the simulated lab environment? Or did they wait for the storms to perform field test?:-)
How accurately can it predict the location where the lightning is going to strike?
Sorry, a spark gap is very unreliable and short-range--it may tell you when lightning hits directly or very close to it, but it won't sense lightning at some distance. To do that, you need to sense the EM field of the lightning. By the time the spark gap flashes over, it will be too close to you and too late to do anything in anticipation.
Plus, it will depend on placement, whatever "antenna" you are using, and many other factors.
Keep in mind that a typical gap's sparkover threshold is about 5-10 kV/cm--so, how big is your gap, and how close/strong does the lightning have to be to jump that gap?
A spark gap makes a good lightning arrestor, but not a detector, sorry
[An aside: [snip] April 2002 QST "A Lightning Detector for the Shack"]
Once upon a time, I read a fascinating article in a old hardcover compendium of C.L. Strong's "Amateur Scientist" columns from /Scientific American/; the EM pulses were termed 'spherics (from "atmospheric"), and a vacuum-tube detector circuit was described.
I suppose that the AS3935 will also detect EMP from nuclear explosions -- God forbid...
Living on a hill in Salt Lake City, I used to use a vacuum tube am radio, tuned between stations, to monitor the August thunderstorms. Many broad-ranging storms produce a strike per second. The sensitivity was such that I could hear many strikes which did not accompany a visible flash, but in general the closer the arc the louder the sound pulse. In every case I saw a strike (most cloud-to-cloud with some I estimated to be 15 to 20 miles distant), I could instantly hear the rf noise.