Influence of subsurface refraction and wave travel path on the estimation of average <i>V_S</i> profiles from the seismic downhole test

Abstract The Downhole (DH) seismic test is an important geophysical exploration method and has become an essential tool for determining subsurface dynamic properties and seismic site class. To analyse the acquired data, an approximation of a straight raypath is often assumed between the source and receiver, which neglects refractions along the travel path due to changes in subsurface properties. Further investigation is needed to fully understand the effects of subsurface refraction on the determination of total travel length and average S‐wave velocity ( V S ). Hence, DH tests at several locations up to a depth of over 30 m were conducted in this study. Average V S profiles were determined considering a straight travel path and refraction occurring along the travel path. It was observed that the average V S values remain fairly independent of the inversion method used. The total refracted raypath length up to 30 m depth also shows no significant difference compared to a straight raypath. Similar observations were noted when the average V S and raypath length from surface up to sediment depth and bedrock depth was calculated. Further, the ambiguity associated with the selection of a model for DH data reduction was studied, and minimal change was observed in average V S with a change in depth of different subsurface interfaces. This observation concludes that the travel time from the source up to the 30 m depth and bedrock depth are sufficient for site classification if a detailed profile is not needed immediately. Further, the average V S values obtained from DH tests were compared with commonly used MASW tests. The difference in average V S up to bedrock and sediment depth from MASW and from DH test, considering both refracted raypath and straight raypath, was observed to be high for shallow bedrock sites. To study the effect of error in arrival time picking on average V S profiles, random error within ±0.50 ms was introduced in arrival time data and V S profiles were estimated.