Discriminating cloud to ground lightning flashes based on wavelet analysis of electric field signals

Abstract Lightning discharges produce electromagnetic radiation in a wide frequency range, but its propagation in a certain frequency range are usually used by lightning detection networks. Investigation of lightning activities in time-frequency domain can be obtained by using the wavelet transform. This study proposes a new approach using the discrete wavelet transform (DWT) algorithm to classify the detected lightning strikes. The measuring station would capture lightning electric field in 500 ms time scale and then utilizes a wavelet based recognizer algorithm to duly differentiate the cloud to ground flash from other cloud activities. Wavelet transform allows the expansion of transient events into a small number of coefficients. A total of 200 lightning flashes were randomly selected among the captured lightning discharges in South of Malaysia in one year. Initially, the cloud to ground and other cloud activities were manually analysed and discriminated. Then, these lightning flashes were analysed using different mother wavelets such as Haar, symmlet, Coiflet, and Daubechies by means of MATLAB program. Haar mother wavelet gives the best result for CG decomposition analysis. A total of 24 decomposition layers were chosen and the energy level of each layer was calculated to obtain the correlation between energy fluctuation and type of signal. The investigations reveal that the CG discharges have higher energy in 17th to 20th layers compared to the rest. However, the opposite results were obtained for the case of other cloud activities. To increase the accuracy of the wavelet transform approach algorithm, another filter was added to the algorithm flowchart. The proposed CG discrimination algorithm successfully classified 92% of the randomly selected flashes.