Local polynomial approximation and intersection of confidence intervals for removing noise of lightning electric field measurements

Abstract

Lightning electric field (LEF) measurements are aperiodic signals characterized by inherent noise of different sources, i.e., it is not possible to register a noise-free signal. In the last decade, the denoising of LEF measurements has been achieved using some time-frequency representations such as short-time Fourier transform (STFT), wavelet transform (WT) and fractional Fourier transform (FRFT) without definitive results. In this paper, a denoising process applied on LEF measurements using the Local Polynomial Approximation (LPA) is proposed. The window size selection is made by combining the LPA with the intersection of confidence intervals (ICI) algorithm. Furthermore, a cross-validation criterion is used to select the optimal value of the threshold parameter in the LPA-ICI denoising method. It is shown that for different signal-to-noise ratio (SNR) values, the proposed method significantly reduces the noise present in the recorded signals. Finally, a discussion about the processed signatures in terms of some lightning electric field temporal features is performed.

Keywords

• electric field
• intersection of confidence intervals
• lightning flash
• local polynomial approximation
• noise reduction