HR: 10:50h
AN: S22A-03    [Abstracts]
TI: Improving the quality of shallow velocities derived from VSP data by use of synthetic data
AU: Ge, J
EM: jge@memphis.edu
AF: Center for Earthquake Research and Information, The University of Memphis, Memphis, TN 38152 United States
AU: * Pujol, J
EM: jpujol@memphis.edu
AF: Dept. of Earth Sciences, The University of Memphis, Memphis, TN 38152 United States
AU: Pezeshk, S
EM: spezeshk@memphis.edu
AF: Dept. of Civil Engineering, The University of Memphis, Memphis, TN 38152 United States
AU: Stovall, S
EM: spstvll@memphis.edu
AF: Dept. of Civil Engineering, The University of Memphis, Memphis, TN 38152 United States
AB: We determined S wave interval velocities in shallow boreholes (depths up to about 60 m) drilled in the sediments of the Mississippi embayment using vertical seismic profiling (VSP) data. The source was an air-powered hammer that produced SH waves. In all cases the source offset was small (about 1.2 m). The depth interval was 1.5 m. The interval velocities were determined using first arrival times and two approaches, a zero-offset approximation and a layer-stripping method. The two sets of velocities are very close below about 5 m, and for some boreholes the velocity variations are quite large. Although in-situ velocities are affected by less error than those determined using surface data, the picking of arrival times is always affected by some uncertainty, which translates into errors in the computed velocities. In our case the most likely source of error was noise from human activities in the vicinity of the boreholes. To investigate the quality of the computed velocities we used synthetic VSP data. This was done using a forward approach by which the interval velocities computed using the layer-stripping method were modified until the synthetic data matched approximately the first cycle of the actual data. The synthetic data thus computed show large reflections that are not observed in the corresponding actual data, but their amplitudes become much smaller when attenuation is included in the computations. The quality factor (Q) used was determined using a spectral ratio method. As expected, the interval velocities determined using synthetic data show less variation than the original velocities. In some cases the differences between the two sets of velocities are significant. The synthetic data were also used to investigate the effect of realistic timing errors in the computation of interval velocities. The effect of using peak values instead of first arrivals was also considered.
DE: 7299 General or miscellaneous
SC: Seismology [S]
MN: Fall Meeting 2005