Seismic cone penetrometer (SCPT) technique for hazard studies

We have found that high‐quality vertical seismic profile (VSP) data can be collected for near‐surface applications using the seismic cone penetrometer. Cone‐mounted accelerometers are used as the VSP receivers, and a sledgehammer against the cone truck baseplate is used as a source. This technique eliminates the need to drill a borehole, thereby reducing the cost of the survey, and results in a less invasive means of obtaining VSP data. Two SH-wave VSP surveys were acquired over a deltaic sand/silt sequence and compared to an SH-wave common‐depth‐point (CDP) reflection profile. The VSP data were processed using a combination of singular‐value‐decomposition filtering, deconvolution, and f-k filtering to produce the final VSP extracted traces. The VSP traces correlate well with cone geotechnical logs and the CDP surface‐seismic data. The first breaks from the VSP can be used to generate shear‐wave velocity profiles that are important for time‐to‐depth conversion and the velocity correction of the CDP surface data.

Download Full-text

A Bayesian approach for determining velocity and uncertainty estimates from seismic cone penetrometer testing or vertical seismic profiling data

Canadian Geotechnical Journal ◽

10.1139/t11-012 ◽

2011 ◽

Vol 48 (7) ◽

pp. 1061-1069

Author(s):

Adam Pidlisecky ◽

Seth S. Haines

Keyword(s):

Travel Time ◽

Model Uncertainty ◽

Velocity Model ◽

Cone Penetrometer ◽

Uncertainty Estimates ◽

Data Error ◽

Cone Penetrometer Testing ◽

Seismic Cone Penetrometer ◽

Measurement Location ◽

Ray Paths

Conventional processing methods for seismic cone penetrometer data present several shortcomings, most notably the absence of a robust velocity model uncertainty estimate. We propose a new seismic cone penetrometer testing (SCPT) data-processing approach that employs Bayesian methods to map measured data errors into quantitative estimates of model uncertainty. We first calculate travel-time differences for all permutations of seismic trace pairs. That is, we cross-correlate each trace at each measurement location with every trace at every other measurement location to determine travel-time differences that are not biased by the choice of any particular reference trace and to thoroughly characterize data error. We calculate a forward operator that accounts for the different ray paths for each measurement location, including refraction at layer boundaries. We then use a Bayesian inversion scheme to obtain the most likely slowness (the reciprocal of velocity) and a distribution of probable slowness values for each model layer. The result is a velocity model that is based on correct ray paths, with uncertainty bounds that are based on the data error.

Download Full-text