H/V Variability of Ambient Seismic Noise : Sources, Structure and Polarization

Background
The H/V method, based on the spectral ratio between the horizontal and vertical components of ambient seismic noise, is widely used to characterize site effects and identify frequencies that may be amplified during an earthquake. However, the interpretation of H/V amplitude and its use to characterize subsurface structure and quantify strong ground-motion amplification remain active research topics.
In a recent study of the Mygdonian Basin in northern Greece, we showed from 3D numerical simulations that H/V measurements can depend on both the local geological structure and the spatial distribution of ambient noise sources. In particular, the presence of a blind fault and strong lateral contrasts can modify H/V amplitude and polarization, while non-isotropic noise sources can produce temporal and spatial variations in the measurements and, in some cases, mask the signature of the structure.
To better understand these effects, a two-month measurement campaign was carried out at the Volvi site in the Mygdonian Basin using an array of 30 seismic sensors. This dataset provides a unique opportunity to simultaneously investigate the temporal and spatial variability of H/V around the blind fault, as well as to identify and locate the noise sources active during the campaign.

Internship objectives
The main objective of the internship will be to better understand the origin of the variability observed in H/V measurements at the Volvi site, distinguishing, as far as possible, between effects related to subsurface structure and those related to the distribution of noise sources.
The work will focus in particular on :
• analysing the seismic recordings acquired during the two-month campaign ;
• computing and analysing H/V curves as a function of time and sensor location ;
• investigating H/V polarization and its robustness under different recording conditions ;
• analysing ambient seismic noise using the array in order to identify and locate active noise sources ;
• studying the relationships between source distribution, H/V variability, and station location relative to the blind fault ;
• comparing the observations with the results of 3D numerical simulations performed using a simplified representation of the basin ;
• identifying effects related to non-isotropic sources and exploring methods to distinguish, and eventually account for or correct, these effects in order to better isolate the structural signature.
Particular attention will be paid to H/V polarization, which remains a relatively underused observable. The simulations performed in our study show that both its direction and amplitude can be jointly controlled by laterally heterogeneous structure and source distribution. The dense measurements acquired at Volvi will provide an opportunity to directly test these results against real data.

Approach and Methods
The internship will combine real-data processing, array-based ambient-noise analysis, and the use of numerical simulation results.
The first stage will consist of characterizing the temporal variability of the seismic signals and H/V curves across the entire array. The measurements will be analysed at different time scales in order to identify variations related to changes in ambient-noise conditions.
The second stage will focus on the spatial analysis. The density of the array will make it possible to investigate H/V variations around the blind fault and to search for signatures consistent with the lateral structural variations predicted by the simulations.
Array-based analysis will then be used to characterize the spatial distribution of noise sources and to track their evolution over time. This approach will make it possible in particular to relate the observed H/V variations to changes in the composition of the ambient seismic wavefield.
Finally, the results will be compared with the available numerical simulations in order to test the identified mechanisms and assess the extent to which H/V polarization can provide a complementary observable for characterizing laterally heterogeneous structures.

Skills and profile
Scientific and technical skills :

• signal processing and analysis ;
• scientific programming ;
• basic knowledge of seismology and/or geophysics ;
• interest in experimental data analysis.

Personal qualities :

• autonomy ;
• rigour in data analysis and management ;
• ability to explore and interpret numerical and experimental results ;
• interest in scientific research.

Previous experience in signal processing, programming, or seismological data analysis will be appreciated.

Scientific interest and strengths of the internship
This internship offers the opportunity to work with an original real-world seismological dataset acquired specifically to address questions raised by numerical simulations.
It will combine :

• analysis of real data from a two-month campaign using an array of 30 sensors ;
• use of 3D numerical simulations of ambient seismic noise ;
• signal processing and ambient-noise analysis at both single-station and array scales ;
• joint analysis of H/V amplitude and polarization ;
• contribution to an active research topic concerning the use of H/V curves to constrain subsurface structure and better quantify site effects and strong ground-motion amplification ;
• exploration of a new observable, H/V polarization, which may provide complementary information on laterally heterogeneous structures.

Beyond the analysis of the Volvi site, the internship will aim to determine to what extent variations induced by noise sources can be identified and accounted for, in order to make better use of H/V measurements for the study of site effects.

References :
David Gregor, Emmanuel Chaljub, Marc Wathelet, Fabrice Hollender, Vincent Perron, Florent De Martin, Three-dimensional numerical simulation of ambient seismic noise for arbitrary surface source distributions : application to the study of horizontal-to-vertical spectral ratios and their variability, Geophysical Journal International, Volume 247, Issue 2, November 2026, ggag348, https://doi.org/10.1093/gji/ggag348