One‐dimensional ground response analyses (GRAs) can introduce model error to site response estimates when wave propagation is not dominated by vertically propagating shear waves. We investigate whether the variability of spatially distributed microtremor‐based horizontal‐to‐vertical spectral ratios (mHVSR) is related to GRA effectiveness using data from 19 vertical array sites across California. For each site, multiple mHVSR curves were measured at variable distances from the vertical array (average of 20/site). The variability of spatially distributed mHVSR curves is evaluated for various frequency intervals centered on peaks as a function of separation distance using various correlation metrics. We select the longest common subsequence metric for its ability to align with independent analysts’ visual assessments and improve upon previously proposed metrics such as Pearson's r , mean absolute error, and variability of fundamental site frequency derived from HVSR. GRA effectiveness is evaluated using results from prior work in which predicted and observed surface/downhole transfer functions were derived across multiple events for each vertical array, with a wide range of outcomes spanning from excellent alignment to strong misfits. We observe that vertical array sites exhibiting spatial similarity in the mHVSR curves, particularly within a low frequency interval corresponding to the fundamental peak of the observed transfer function, tend to show alignment between the predicted and observed transfer functions. Moreover, sites with pronounced low‐frequency mHVSR variability were less likely to be adequately modeled by GRA. These results show that spatially distributed mHVSR measurements can provide useful insights regarding site complexity with implications for the reliability of site response modeling.
Ornelas et al. (Sun,) studied this question.