Application of Kohonen self-organizing map for clustering negative cloud-to-ground lightning electric field waveforms

Lightning electric field (E-field) waveforms are widely used in understanding the physical processes that take place during different lightning events and for improving engineering return-stroke models that helps the design of better lightning protection systems. The Lightning Detection and Waveform Storage System (LDWSS) is a low cost system that record relatively wideband E-fields of lightning. In this work we use the Kohonen self-organizing map for clustering electric field waveforms of negative cloud-to-ground lightning (-CGs) recorded by the LDWSS. This approach allowed us to determine a standard/typical E-field waveform for distinct range of distance. The data was splitted into two subsets, one for events recorded under daytime conditions and other for events recorded at nighttime.The results showed that it is possible to identify twenty-five standard/typical E-field waveforms of -CGs, being 15 groups of waveforms that occurred during daytime and 10 groups that occurred at nighttime conditions.