November 25, 2026
Europe/Bratislava timezone

KDE-Based Mode Estimation and Directional Analysis of Ambient-Noise HVSR

Not scheduled
1m
Vestibul pavilónu Fyziky

Vestibul pavilónu Fyziky

Zamestnanci Fyzika Poster session + coffee: works of research groups of the faculty Physics Poster session + káva: prezentácie vedeckých výsledkov FMFI UK - Fyzika

Speaker

Jozef Kristek (Comenius University Bratislava)

Description

Horizontal-to-vertical spectral ratios (HVSRs) from ambient noise are widely used to characterize near-surface structure and site effects. Most current workflows assume that the amplitudes obtained from individual time windows follow a lognormal distribution and then report a single “lognormal median” curve (geometric mean) with a symmetric ±1σ band. In many real data sets, however, the window-wise distributions of amplitudes and peak frequencies are skewed or show more than one preferred value. As a result, the lognormal summary may lie outside the region of highest probability, and its uncertainty band can be much wider or differently shaped than the data suggest.
We explore a different way to describe HVSR ensembles that does not rely on a specific parametric model. Instead of forcing a lognormal shape, we estimate the full distribution of amplitudes and directions and use its most probable value (the mode) and a range that contains most of the probability to describe the data. We apply this idea both to the usual, omnidirectional HVSR and to directional versions based on the main horizontal direction.
Using a near-lognormal microtremor example, we show that this mode-based representation is very close to the lognormal mode and exposes a modest upward bias in the lognormal median and its ±1σ band.
A second example relates to a complex mountainous site. When the amplitude and direction statistics are multimodal, the traditional lognormal summary falls between competing modes and overestimates uncertainty. Contrary, the mode-based curves follow the dominant branches, and the associated ranges span the full set of likely values.
Together, the results suggest that mode-based and directional summaries provide a more faithful and informative description of ambient-noise HVSR ensembles and offer simple diagnostics for weighting data and assessing the validity of one-dimensional site-response models.

Pracovisko fakulty (katedra)/ Department of Faculty KAFZM / DAPEM
Tlač postru/ Print poster Budem požadovať tlač /I hereby required to print the poster in faculty

Authors

Jozef Kristek (Comenius University Bratislava) Miriam Kristekova (Comenius University Bratislava) Peter Moczo (Comenius University Bratislava) Pierre-Yves Bard (Univ. Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, IRD, Univ. Gustave Eiffel, ISTerre, Grenoble, France)

Presentation materials

There are no materials yet.