Predicting cloud-to-ground lightning with neural networks

Sie sind hier: Homepage > Suche

Predicting cloud-to-ground lightning with neural networks

Barnes, A.A. jun. / Frankel, D. / Draper, J.S.

Lightning is a hazard to ground operations, missile launch operations and recovery of the Space Shuttle at Cape Canaveral. The Air Force is responsible for providing the forecasts of lightning for these operations. In an effort to improve the forecasting of cloud-to-ground lightning, neural networks are being applied using the large data bases from the Cape Canaveral area which includes Cape Canaveral Air Force Station (CCAFS), Kennedy Space Center (KSC) and peripheral locations. The initial study (1-3) employed the wind data from a number of different levels on 32 towers to predict lightning strikes in 16 blocks over Cape Canaveral for four time periods; 0-15 min., 15-30 min., 30-60 min. and 1-2 hours. The network was trained by backpropagation using the data from one day, 24 July 1988, and was verified on independent data from 25 July 1988. Comparisons were made with the convergence method of Watson et al (4) and were found to give similar results. The neural network results should improve with larger training sets and with the addition of more of the readily available meteorological data. Results of further training and the addition of ground based field mill data are discussed.