This project will evaluate an innovative neural-network, natural lightning predictor that uses new types of meteorological parameters (e.g. temperature, humidity) as inputs and training data histories from several epochs earlier than the 'current' epoch. The predictor will indicate where future lightning strikes will occur in time (T=0, 15 min., 30 min., 1 hr., 2 hrs.) over 16 different (5 x 5 nmi.) 'tiles' covering the Kennedy Space Center. The predictor's feasibility will be demonstrated with its increased probability of predicting a lightning strike above the 0.50 value (with a probability of false alarm < 0.001) over the current state-of-the-art system described by Frankel and Draper (1990). Based on Phase I results, will be made for a comprehensive (natural and initiated) lightning predictor will be recommended for the Phase II pre-commercialization prototype.
Potential Commercial Applications:A maker of tactical weather stations has indicated strong interest in the predictor's commercial possibilities, and the utility industry has also expressed interest. With the expertise gained in this project, a prediction capability for agricultural crops, rainfall, forest fires, and protection of commercial airports could be developed.