Spectral analysis of underwater explosions in the Dead Sea

The present study utilizes the Israel Seismic Network (ISN) as a spatially distributed multichannel system for the discrimination of low‐magnitude events ( M L  < 2.5), namely earthquakes and underwater explosions in the Dead Sea. In order to achieve this, we began with the application of conventional single‐station methods, such as spectral short‐period ratios. We then applied a newly developed, network‐oriented algorithm based on different spectral features of the seismic radiation from underwater explosions and earthquakes, i.e. spectral semblance statistics.  Twenty‐eight single‐shot underwater explosions (UWEs) and 16 earthquakes in the magnitude range M L  = 1.6–2.8, within distances of 10–150 km, recorded by the ISN, were selected for the analysis. The analysis is based on a smoothed (0.5 Hz window) Fourier spectrum of the whole signal (defined by the signal‐to‐noise criterion), without picking separate wave phases. It was found that the classical discriminant of the seismic energy ratio between the relatively low‐frequency (1–6 Hz) and high‐frequency (6–11 Hz) bands, averaged over an ISN subnetwork, showed an overlap between UWEs and earthquakes and cannot itself provide reliable identification.  We developed and tested a new multistation discriminant based on the low‐ frequency spectral modulation (LFSM) method. In our case the LFSM is associated with the bubbling effect in underwater explosions. The method demonstrates a distinct azimuth‐invariant coherency of spectral shapes in the low‐frequency range (1–12 Hz) of short‐period seismometer systems. The coherency of the modulated spectra for different ISN stations was measured by semblance statistics commonly used in seismic prospecting for phase correlation in the time domain. The modified statistics provided an almost complete separation between earthquakes and underwater explosions.