SCEC Award Number 12124 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Stochastic Descriptions of Basin Velocity Structure from Analyses of Sonic Logs and the SCEC Community Velocity Model (CVM-H)
Investigator(s)
Name Organization
John H. Shaw Harvard University Thomas H. Jordan University of Southern California
Other Participants Andreas Plesch
SCEC Priorities 6c, 6a SCEC Groups GMP, Seismology, USR
Report Due Date 03/15/2013 Date Report Submitted N/A
Project Abstract
 Sustained increases in computational resources and the trend of numerical wave propagation studies to shorter periods create a demand for higher resolution velocity models. One way to parameterize such models is to characterize the local variability of the small length scale (<100m) structure with the goal of implementing a stochastic velocity overlay in regional velocity models. In this project, we analyzed sonic logs in the Los Angeles basin in conjunction with the velocity structure in the current CVMH (11.9) in order to develop a stochastic representation of variability to depths of 3 km.

Using the standard variation of a distribution as a measure of its variability, our analysis showed a standard variation of 20.8x10-6 s/m (around a mean of 1.25x10-6 s/m) for the delta between compressional wave slowness in well logs and the model. This variation characterizes aggregate variability at length scales between 3m (effective log resolution after despiking) and 100m (model resolution). A similar analyses using relative variations, that is the ratio of despiked log data and the model, results in +/- 6% variability in slowness at 1 sigma.

In order to define spatial length scales of variability, we attempted to measure correlation distances in vertical and horizontal directions through a variogram analyses. Our results show a maximum vertical correlation distance of about 80m at which variance levels reach about 430 µs*µs/m*m. The analyses suggest a possible background level of horizontal variance of about 750 µs*µs/m*m, yielding a maximum correlation distance of about 900m.
Intellectual Merit This project represents a new effort to develop stochastic representations of detailed velocity structure in the SCEC Community Velocity Models (CVMH), which seeks to enable ground motion simulations to higher frequencies.
Broader Impacts These enhanced velocity models will enable simulations at higher frequencies providing improved strong ground motion predictions, which are a central element of SCEC hazard assessment efforts.
Exemplary Figure Fig. 6: Histograms of logarithmic variability at various despiking length scales.