Please use this identifier to cite or link to this item:
Title: Source location and tracking capability of a small seismic array
Authors: Moran, Mark L.
Albert, Donald G.
Keywords: Beam forming
Seismic array processing
Capon method
Target location and tracking
Frequency-wavenumber estimation
Unattended ground sensors
Maximum likelihood method
Wavenumber bias effects
Seismic waves
Target acquistion
Publisher: Cold Regions Research and Engineering Laboratory (U.S.)
Engineer Research and Development Center (U.S.)
Series/Report no.: CRREL report ; 96-8.
Description: CRREL Report
Abstract: Recordings of seismic wavefields from various sources were obtained using a small array of vertical-component geophones under winter conditions at Grayling, Michigan. These data were processed using a frequency-wavenumber domain Capon minimum variance beamformer to estimate the bearing angle and propagation velocity of the waves emitted from the source. The cross power matrix was adaptively estimated using a tapered block-averaging procedure. The wave sources were sledgehammer blows on the ground surface, .45 caliber blank pistol shots, and an M60 tank moving at 4.5 m s^–1 along a road near the array. Reliable wavenumber spectra were obtained for all sources. Processing results for the hammer blows show that the dominant seismic arrival is a Rayleigh wave traveling at roughly 220 m s^–1. For the pistol shots, two arrivals corresponding to the airwave (338 m s^–1) and the air-coupled Rayleigh waves (220 m s^–1) were observed. The results for these sources were relatively insensitive to the processing parameters used. For the moving vehicle, the dominant signals observed were Rayleigh waves (220 m s^–1). Accurate locations were obtained for this moving source, although the processing parameters had to be carefully selected, and the choice of frequency parameters affected the accuracy of the wavenumber results. Maximizing the number of degrees of freedom and the coherence of the frequency estimates and minimizing the variation of the coherence across adjacent frequency bins provided the most consistently reliable strategy for obtaining accurate wavenumber estimates for the moving vehicle. The sensitivity of the wavenumber estimates to the frequency processing parameters seems to be related to the bias in the phase spectra of the signals and will potentially occur in any bearing estimation method that uses temporal frequncy phase spectra.
Rights: Approved for public release; distribution is unlimited.
Appears in Collections:CRREL Report

Files in This Item:
File Description SizeFormat 
CR-96-8.pdf1.2 MBAdobe PDFThumbnail