Suppression of Periodic Disturbances in Seismic Aftershock Recordings for Better Localisation of Underground Explosions

For precise localisation of a potential underground nuclear explosion, the Comprehensive Nuclear-Test-Ban Treaty Organization, during an on-site inspection, can set up seismic sensors to find the very small signals from aftershocks. These signals can be masked by periodic disturbances from, for example, helicopters. We present a new method to characterise every such disturbance by the amplitude, frequency and phase of the underlying sine in the time domain using a mathematical expression for its Hann-windowed discrete Fourier transform. The contributions of these sines are computed and subtracted from the complex spectrum sequentially. Two examples show the performance of the procedure: (1) synthetic sines superposed to a coal-mine induced event, orders of magnitude stronger than the latter, can be removed successfully, (2) removal of periodic content from the signals of a helicopter overflight reduces the amplitude by a factor 3.3 when the frequencies are approximately constant. The procedure cannot yet cope with peaks that change frequency too fast, for example by the Doppler effect when passing, and with peaks that lie too close to each other. Improvement to solve these problems seems possible.