They say lightning never strikes twice, but that is a spot of deceived wisdom. Indeed, many buildings, trees, and people have been struck multiple times. There are some 3 million lightning flashes around the world every single day. Predicting atmospheric electric activity is important for a number of reasons. Research published in the International Journal of Experimental Design and Process Optimisation looks at lightning prediction that could be vital for space launch operations at Kennedy Space Center and Cape Canaveral Space Force Station.
Jared Nystrom, Raymond R. Hill, Andrew Geyer, and Joseph J. Pignatiello Jr. of the Air Force Institute of Technology, Wright-Patterson Air Force Base, Ohio, USA, explain that near real-time prediction of lightning is critical to operations at these sites where the risk to life can be very high and delays and damage very costly. Current approaches to lightning prediction are inefficient, the team writes, as well as coming with large uncertainties in the predictions offered. A new model developed by the team uses the wavelet decomposition of chaotic weather sensor time series and semiparametric single-index models to mitigate the chaotic signal and any possible distributional misspecification.
The new approach benefits from the development of inexpensive sensors that can be deployed widely at a site and can be scalable. Such sensors produce noisy, high volume, and high-frequency time series, with which conventional processing often struggles. Specifically, assumptions made in the modeling process may not be valid and any smoothing of the data may overcompensate for the noise meaning the actual signal can be lost.
The team’s new approach offers a significant improvement over a persistence model, the team says, and can give positive identification of whether a given sight is likely to be struck by lightning within the next hour from the point at which the data is assimilated. Moreover, the system can predict the triggering events that might lead to an actual lightning strike. The team adds that the approach might also be adapted to other types of site with a lower density of sensors than a space centre and so allow remote lightning prediction.
Nystrom, J., Hill, R.R., Geyer, A., Pignatiello Jr., J.J. and Chicken, E. (2021) ‘Experimental design in complex model formulation for lightning prediction’, Int. J. Experimental Design and Process Optimisation, Vol. 6, No. 4, pp.304–332.