An analysis of the effects of noise on three-component VSP /
| Main Author: | |
|---|---|
| Other Authors: | , , , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1990.
|
| Subjects: | |
| Online Access: | ProQuest, Abstract Link to OAKTrust copy |
| Abstract: | A gradient technique is presented for finding the orientation of particle motion on a three-component vertical seismic profile (VSP). A least-squares residual is calculated which indicates the rectilinearity; the lower the residual, the more rectilinear is the particle motion. The procedure was applied to synthetic data and to a field VSP recorded using three offset sources to determine how the uncertainty in polarization measurement varies with signal-to-noise (S/N) ratio, noise bandwidth, and time window length. The downhole orientation of the sonde was found by measuring the azimuth of the direct P arrival. The most accurate azimuth measurements on the field VSP were generally found not at the peak of maximum amplitude, but just after the first break, where the residual was at its lowest value. The uncertainty in azimuth increased, and the S/N ratio decreased, with depth. The uncertainty in azimuth measurement was 1.2° at a S/N ratio of 9.3, and 6.9° at a S/N ratio of 3.7. By comparing the measured azimuths with the known azimuth angles between the sources, errors in one source azimuth measurement at a given depth were identified and removed. The uncertainty in azimuth was thereby reduced to less than 0.4 times the original value. The uncertainties on the field VSP agreed with those found by modeling the noise as band-limited noise having a Gaussian amplitude distribution. The polar angle of particle polarization was measured on the field VSP, and the uncertainty in measurement was estimated. Changes in apparent polarization of a P event due to P-SV conversion at an impedance contrast were measured on a VSP model. Velocity filtering was used to try to reduce the polarization error on the field VSP and the model. The error due to P-SV conversion appeared to be reduced, but a new error was introduced by Rieber mixing. The probability of the residual for band-limited Gaussian noise being less than a given value was found for varying noise bandwidth and time window length. The probability was constant for a constant ratio of time window length to minimum noise period. The duration of minima on the residual time record was investigated as a criterion for separating signal from noise on synthetic signals contaminated with band-limited Gaussian noise. The technique was successful when the duration of the rectilinear signals was longer than the mean noise period. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. "Major subject: Geophysics." |
| Physical Description: | xiv, 190 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references. |