On the correlation between earthquake coda horizontal-to-vertical spectral ratios and amplification functions at the KiK-net network

The Japanese KiK-net network comprises about 700 stations spread across the whole territory of Japan. For most stations, V P and S profiles were measured down to bottom borehole station. Using vast dataset earthquake recordings from 1997 2020 at a subset 428 seismic we compute horizontal-to-vertical spectral ratio coda, S-wave surface-to-borehole ratio, equivalent outcropping amplification function. de facto equivalence coda ambient vibration is assessed on homologous Swiss dataset. Based that, applied canonical correlation analysis between information all sites. aim test strategy predict local basing inference site condition indicators single-station recordings. Once frequency-dependent factors amplitudes ratios defined, each in selected with leave-one-out cross-validation approach. In particular, for site, three rounds predictions are performed, using as prediction target standard referred bedrock common reference rock profile. From our analysis, effective obtained when used, whereas strong mismatch reference. We ascribe this effect fact differently functions reference, fully site-dependent then their peak amplitude influenced by velocity contrast overlying sediments. Therefore, reduce discrepancy, add proxy inferred bedrock, which improves final prediction.