摘要： AbstractThe Imager of Sprites and Upper Atmospheric Lightning (ISUAL) on the FORMOSAT‐2 satellite, launched in 2004, records Transient Luminous Events (TLEs). ISUAL has an imager and a spectrophotometer that observe TLEs all over the globe. Among these phenomena, elves are particularly difficult to detect. ISUAL often records events that correspond to significant far ultraviolet (FUV) emissions in the spectrophotometer but have no discernible TLEs in the imager. These FUV events are called “dim” elves. Therefore, it is important to develop mathematical tools to analyze the data to obtain a better evaluation of the number of elves and their occurrence. Multivariate approaches are applied to characterize the unlabeled events. The first approach is the principal component analysis which distinguishes two different groups, one including elves and dim elves. The second approach is the PARallel FACtor analysis which provides a waveform model for each group. These methodologies confirm that FUV signal is the evidence of TLE presence. A crude classification method was then suggested taking into account these results. The proportion of elves, relatively to the considered ISUAL data set, is found to be about 40%. It is similar to previous results and confirms that relatively weak lightning peak current is sufficient to produce elves. This new strategy demonstrates the potential for discriminating between lightning and TLEs without prior knowledge within the selectivity of the FUV spectral band. 其他題名： J. Geophys. Res. Atmos 出版者： Washington: Blackwell Publishing Ltd 出版日期： 2015-08-16 出處： Journal of Geophysical Research: Atmospheres, 2015-08, Vol.120 (15), p.7454-7466 資源來源： Wiley Online Library - AutoHoldings Journals 版權： 2015. American Geophysical Union. All Rights Reserved. 版權： Copyright Blackwell Publishing Ltd. Aug 2015 識別號： ISSN: 2169-897X 識別號： ISSN: 2169-8996 識別號： EISSN: 2169-8996 識別號： DOI: 10.1002/2014JD023012
