Pre-estimation of Distance-Based Lightning Using Effective Meteorological Parameters

Abstract

Lightning can be described as a high voltage electrical discharge event that occurs between the cloud and the ground. The development of a warning system that enables us to predict lightning before it occurs and to take safety precautions is essential to minimize undesired lightning-related events. This study aims to predict lightning events 1 h in advance by using meteorological data recorded by a single meteorological station. The data used in this study consist of a total of ten atmospheric features, and lightning strikes were examined in three groups by considering distances from the station as 0–2 km (data group-1, DG-1), 2–4 km (DG-2), and 4–6 km (DG-3). The sequential forward selection (SFS) method was used to determine the most effective among the ten features. The probability of lightning for each data group was then estimated with the artificial neural networks algorithm. As a result, we found that the best lightning estimation rate among the three groups was obtained at 0–2 km distance, with 90% accuracy with only four effective features determined by the SFS method. As far as the authors are aware, a detailed analysis of the data and a distance-based lightning prediction system in the literature does not exist, and therefore, the originality and success of this study are expected to contribute to the literature for future studies.