hear wave velocity (VS) is one of the essential parameters for site characterization studies. This study aims to generate reliable equations for estimating VS of top 30 m of soil profile (VS30) in absence of site-specific measurements and investigates the relation between VS and some of the geotechnical properties (e.g. the number of SPT blow counts (N), depth (Z) and VS in the upper layer of soil (VSu) in Mashhad. Group method of data handling (GMDH) type neural network optimized using the genetic algorithm (GA) was used to model these relationships. A database containing 1657 data points compiled from 206 boreholes was used for training and testing of the models. The performance of the proposed correlations compared with the previously published correlations for VS showed aconsiderable improvement in the prediction of VS. Sensitivity analysis of the obtained models was performed to study the influence of input parameters on model output. Results show that VSu is the most important parameter in predicting VS and SPT is a more effective parameter than depth in predicting VS in coarse- grained soils.

Keywords