Time domain polarization filters

As well known, each type of seismic waves has a specific parade motion. This property of seismic waves could be used to identify the wave phases in three component seismo grams, and. to consturct the basic construction the approximation for enhance of the signal- noise ratio. For instance, noises generally have elliptic polarization while body waves show linear polarization. Noises some time may have linear polarization, but their direction is random. Polarization filters have been developed using these properties of signal and noise. Information of the seismic wave polarization can be obtain both in time and in fre quency domain. In this study, both to analyse of the seismic wave polarization and to design the polarization filters have been used the time domain approximation. This technique bases on to determine an ellipsoid that best fits the data points over the time window in least square sense. For this purpose, covariance matrix is obtained for the specified time window. The attributes such as rectilinearity, planarity, azimuth and incidence angles of waves are calculated from eigenvalues and eigenvectors of this matrix. After that, these calculations are repeated for windows that are moved for chosen the moving interval, until end of seismograms. The polarization attributes that are obtained as above could be used to determine the phases that are including seismograms and their arrival times. The rectilinearity and directivity functions calculate from polarization attributes of seismic waves to design the time domain polarization filters. The filtered traces obtain with multiplication of these functions and seismograms. Since rectilinearity and directivity func tions obtain for moving time windows, these functions act as a point by point gain control. For this reason, the polarization filtering is a non-linear filtering. Since polarization attributes are calculated for moving time window, sensitivity and stability of results are depend on length and moving interval of window. For this reason, in this study, firstly it has been tried to find a general criterion to choose these parameters. For this purpose, the technique firstly applied synthetic data. Similarly, the time domain polari zation filter applied synthetic data to determine the optimum filter parameters. Finally, both of these techniques applied local and far field earthquake data and discussed their results. Key Words: Particle motion, polarization analysis, polarization filters.