Detection And Source Location Of Unconventional Seismic Signals

In the records of seismometers,besides the distinct waveforms of earthquake events,unconventional seismic signals without clear phases or onsets exist during earthquake-quiet periods.Such unconventional seismic signals include ambient seismic wave fields generated by the atmosphere-ocean system interacting with the solid Earth,tectonic tremors(non-volcanic tremors)originated from slow fault slipping at plate interfaces,and volcanic tremors generally occurring when volcano activities are enhanced.Such signals provide important information about their sources and underground structures.This dissertation introduces our studies about detection and source location of unconventional seismic signals in the aspects of observations and method development.We observe and study the spatial and temporal distributions of two types of sources:microseismic sources excited by Hurricane Sandy and a persistent microseismic source in North Atlantic Ocean.Besides the observations,we also develop a new array processing method to locate unconventional seismic sources.We find that microseisms generated by Hurricane Sandy exhibit coherent energy within 1-hour time windows in the frequency band of 0.1-0.25 Hz,but with signals correlated among seismic stations aligned along close azimuths from the hurricane center.With the identification of this signal property,we show that travel-time difference can be measured between the correlated stations.These correlated seismic signals can be attributed to two types of seismic sources,with one group of the seismic signals from the hurricane center and the other from coastal region.The seismic sources in coastal region are diffusive and move northward along the coastline as Sandy moves northward.We further develop a hurricane seismic source model,to quantitatively describe the coupling among sea level pressure fluctuations,ocean waves and solid earth in the region of hurricane center,and determine the evolution of source's strength and pressure fluctuation in the region of hurricane center using seismic data.Strong seismic sources are also identified near the coastal region in New England after Sandy's dissipation,possibly related to subsequent storm surge in the area.The seismic method may be implemented as another practical means for hurricane monitoring,and seismological estimates of the hurricane seismic source model could be used as in-situ proxy measurements of pressure fluctuation in the region of hurricane center for hurricane physics studies.Using seismic data recorded by stations around North Atlantic Ocean,we carry out broadband(0.1-0.26 Hz)seismological observations of secondary microseismic sources in North Atlantic Ocean during whole 2012 year,with one-hour time resolution.The locations of equivalent point source of microseismic Rayleigh waves are determined by frequency-dependent polarization analysis method.We find that the equivalent point sources for different hours swarm within a region from Ireland,Iceland,to south end of Greenland.We estimate the values of product of group velocity and quality factor at different frequencies,and further determine the strengths of equivalent point source for each hour.We not only observe the seasonal variation pattern of source strengths,but also the variations pattern with several-day time scale.The single active period of this source is shorter than 5 days.There are also long quiet periods even during autumn and winter when the source is relatively strong.We propose a new array processing method:max-eigenvector analysis method,for detecting seismic signals without clear phases or onsets and locating simultaneous multiple sources using coherent signals.The array output function for coherent signal detection and source location is calculated from the eigenvector of the normalized-covariance matrix associated with the maximal eigenvalue.Synthetic tests indicate that our method can distinguish the main and side lobes which are hard to be distinguished by the beamforming method,and has fewer limitations than multiple signal classification(MUSIC)technique.Using this method,we observe the activities of long-period tremors(LPTs)with a period of 20 s at Aso volcano,Japan,during 1 hour in June 2009.The existences of 20s-period LPTs are detected with 2-minute time resolution,and the source positions are located at expected position of Aso volcano,with maximum location uncertainty about 15 km x 15 km.This new max-eigenvector analysis method may be implemented as a practical tool for detecting and locating weak unconventional seismic sources,such as sources of microseisms,tectonic tremors,volcanic tremors,anthropic seismic signals and so on.