GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.2 GNGTS 2024 where H/V peak values are greater than 3, which should indicate a significant level of site amplificaCon. Peak values of this amount are found to be slightly more frequent for the sites on landslides (44%) than on stable areas of (38%). This can be expected because a pre-exisCng landslide creates velocity contrasts with the substratum. Fig. 2: DistribuCon of resonance frequency (a) and H/V peak amplitude (b) based on ambient noise in Daunia. However, resonance condiCons are not ubiquitous and 7.9% of landslide and 6.9% of stable ground areas showed no site effects. Furthermore, resonances appeared rather weak and this could be the effect of the “slow” flysch substratum, which causes weak velocity contrasts with the surficial material. Between the two most common flysch formaCons, FAE appear characterized by a slightly greater diversity of resonance frequencies with higher mean amplitudes, in comparison to FYR whose resonance frequencies are concentrated around lower values and amplitudes. The sites on landslides in FAE, however, show a larger recurrence of lower frequency and higher amplitude, likely as an effect of the reduced velocity of slope materials. The sites on FYR, besides an increase of mean amplitude, show a frequency re-distribuCon towards intermediate frequency from the lowest (< 2 Hz) and the highest (> 8 Hz) frequencies. Furthermore, resonance is absent in 9.4% of FAE and 22.5% of FYR sites. Given the differences in dynamic response, the esCmates of amplificaCon factors from the results of noise analysis may require the use of differenCated relaCons for the two units. However, addiConal efforts are needed considering that, with regard to H/V peak amplitude measurements, the HVNR technique suffers from a strong dependence on environmental condiCons. For this reason the recordings are being reanalysed with the HVIP technique (Del Gaudio, 2017), which extracts from the noise the Rayleigh waves and measures their ellipCcity, thus providing more stable H/V values beber correlated to the local amplificaCon factor. Fig. 3 shows a comparison between the results obtained by the two types of noise analyses.