Evaluation of Existing Methods for Estimation of Shear Wave Velocity for Sites in Alabama

Abstract

Bridges are critical infrastructure systems and need to be resilient to even the rarest of hazards. Seismic damage to bridges and other infrastructure systems is determined by the seismic loading. Seismic loading is dependent on many factors, namely the magnitude and distance of the earthquake, site effects, and the structure itself. Site effects are of particular importance to this study as they influence the frequency and amplitude content of incoming ground motions. Site effects are included in seismic design with the consideration of shear wave velocity (VS). VS is used as a proxy for stiffness of the site as it is directly related to the shear modulus and density of the soil. VS profiles are included in design codes in the form of site classes based on a time averaged VS of the upper 30 meters of the subsurface, known as VS30. The most effective method for determining VS30 is by direct measurement. Direct measurement can be expensive and not economically feasible for every project. In more seismically active regions, many researchers have developed indirect methods for determining VS30 based on empirical relationships between VS and results of commonly performed in-situ tests, ambient noise recordings, and even with geospatial features of the site. These methods have proven to be effective in the regions for which they were developed but need validation before use in Alabama. To evaluate indirect methods for determining VS, direct measurements were taken at multiple sites across the state. Sites were selected to capture the behavior of soft alluvial sites and due to the availability of high quality boring logs. Direct measurements include multichannel analysis of surface waves (MASW) and seismic refraction. Indirect methods include SPT and RQD-based correlations, HVSR-based correlations, and geospatial correlations. Direct measurements were taken at 15 sites as a basis for comparison of indirect methods. Results of comparisons performed on the indirect methods indicate that geospatial correlations provide the best fit to the measured data from this study. Geospatial methods can be applied using GIS based maps and require the least time in application. Geospatial correlations are most effective within the coastal plains province in the state but will tend to be conservative for stiffer sites in other regions. Geospatial methods provide the most accurate estimates of site class with a bias towards conservatism. SPT based correlations are most accurate at shallower depths and tend to underestimate VS at greater depths. The correlations presented perform best on sands as reflected by calculated statistical measures. SPT based methods can provide reasonable estimates of site class but have a greater risk to provide unconservative estimates compared to geospatial methods. RQD based correlations are more limited in number than SPT based correlations. Results of analysis indicate the presented correlations all provide a poor fit to the database used in this study and are not recommended for use. Poor performance is not unexpected as geology in Alabama differs from the regions these correlations were intended to be used in. The sample size of HVSR curves with a clear peak in this study is limited so conclusions drawn may not be representative of overall trends in the state. High-dimensional models presented in Sharma Wagle et al. (2026) were able to be applied to all curves as they consider the entire HVSR curve as opposed to the peak frequency and its amplitude. The addition of more HVSR curves with clear peaks would allow for better evaluation of correlations that require a clear peak in application.