| dc.description.abstract | A significant challenge in seismic site classification often arises from the conflicting shear wave velocity (VS) profiles often obtained using different survey methods, further complicated by uncertainties introduced when relying on correlations with SPT N-values. Despite the widespread use of these techniques, the inherent variability among test methods and its impact on site classification remain poorly understood.
This thesis details the complexities of VS profile estimation using conventional DH reduction methods, viz: Direct Method (DM) and Interval Method (IM), with model studies demonstrating that errors in VS estimation propagate below subsurface interfaces and increase with VS contrast. Accuracy of DH tests can be increased by adopting source offset of 1 to 2m and depth interval of 0.5 to 1m. To enhance DH interpretation, the Combined Direct Interval Method (CDIM) was developed, which eliminates the dependency on prior knowledge of overlying layer velocities. Both analytical model studies and field data validated that CDIM substantially reduces VS estimation errors, yielding VS profiles within a 5% error margin.
Further study using DH tests indicated that refraction correction resulted in a minimal change of less than 2% in average VS (VSz), suggesting a straight raypath is highly practical for estimating VSz and seismic site classification. The study also determined that errors in arrival time picking had only marginal effects on VSz, concluding that a single-point measurement suffices for estimating equivalent average VS, such as Vs30, irrespective of subsurface complexity.
Finally, the refined DH interpretation techniques were utilized to compare VS profiles derived from crosshole, downhole, MASW and SPT correlations. In residual and coastal alluvial region, crosshole and downhole tests gives similar VS values up to VS of 500 m/s. VS values from MASW were generally higher than other VS profiles due to the interface of shallower bedrock layer. This study found that if subsurface layers are spatially homogeneous and VS increases with depth then all methods lead to almost similar velocities. When subsurface layers are non-homogeneous and consist of alternate layers of varying stiffness, VS profiles are prone to deviations from the true profile with minimum deviation in CH followed by DH, SPT correlations and MASW.
Site classification largely remained unaffected by test selection at most locations, although minor scatter near site class boundaries could result in different classes. The conundrum of determining site class using correlations with SPT N-values and directly N-values up to 30m depth were also discussed. Ultimately, downhole test is found to be the most reliable for site classification, whereas crosshole tests is considered better for high resolution VS profiling. | en_US |
