Establishing region-specific N - Vs relationships through hierarchical Bayesian modeling

The blow count (N) measured from the standard penetration test is widely used to predict the shear wave velocity (V-s) of soils. As the N - V-s relationships vary from one region to another, the N - V-s relationship in a region is often determined based on the regional data. When there is no region-specific data or when the region-specific data is limited, such methods can hardly be applicable. In this study, a hierarchical Bayesian model based method is suggested to develop region-specific N - V-s relationships which is applicable to regions without region-specific data and regions with limited region-specific data. The validity of the suggested method is verified using new data that has not been used for model calibration. It is found that the suggested method can be used to predict V(s )for a region with no region-specific data, but the uncertainty associated with V-s could be large. On the other hand, a few region-specific data can significantly reduce the uncertainty associated with V-s. When the amount of region-specific data is small, the N - V-s relationship developed based on the method suggested in this paper is more reliable than the N - V-s relationships developed based on two alternative methods. When the amount of region-specific data is large, a comparison between the N - V-s relationship established by the suggested method and that developed based on regression analyses of the regional data may help identify the risk that the region-specific data may not be representative.

Automated summary

A hierarchical Bayesian model based method is suggested to develop region-specific N-Vs relationships which is applicable to regions without region-specific data and regions with limited region-specific data. The suggested method can be used to predict Vs for regions without region-specific data, but the uncertainty associated with Vs could be large. Additionally, a few region-specific data can significantly reduce the uncertainty associated with Vs.