PRINCIPAL COMPONENT ANALYSIS OF SEISMIC DATA AND DIRECTION OF THE PRINCIPAL COMPONENT FOR SEISMIC DATA
Seismic data laboratory rept.
TELEDYNE INDUSTRIES INC ALEXANDRIA VA EARTH SCIENCES DIV
Pagination or Media Count:
The first part of this report develops the theory of principal components as applied to noise backgrounds measured at seismic arrays. It is noted that the main use of principal components is the reduction of random variables. A summary of statistical interpretation of the principal component is given. An example of principal component analysis is given for LASA noise data. Results are given concerning the proportion of the total variance power from 10 seismometers as explained by the first four principal components. At 0.2 cps approximately 80 of the total variance is accounted for by first principal component. In the summary it is concluded that it seems quite worthwhile to investigate the applications of the principal component to seismic noise study. In the second part the seismic signal data used are from LONGSHOT and a geographically nearby earthquake recorded at a LASA subarray. The phase shifts of the first principal component at 0.2 cps corrected for LASA instrument response prove to be interesting. From these phase shifts its appears that the general direction of the main noise source can be estimated. Computed examples for LONGSHOT and the nearby earthquake are given to verify this claim.
- Statistics and Probability