Research topics

My research concerns the study of acoustic and elastic waves in complex media, at scales ranging from the mm (ultrasonic waves) to the km (seismic waves). In particular, I am interested in the development of methods for high-sensitivity measurements, and high-resolution imaging, using arrays of sensors.

Ultrasonic or seismic waves, Physics are the same, thus these methods can be transferred from one scale to another. However, investigations are generally more convenient with ultrasounds, mostly because active sources can be used easily, and because the objects to study are simpler, in terms of geometry and elastic (acoustic) properties.

1 - Imaging complex media with ultrasound
At the ultrasound scale, my research is focused mainly on wave propagation in waveguides and in multiple scattering media. The main fields of application are non-destructive evaluation (NDE) and structural health monitoring (SHM). The multiplicity of sources and receivers (for example an ultrasonic phased array), allows many new perspectives in the measurement and processing of wave propagation. For example, principal component analysis can be combined with antenna processing methods for significant improvements in the sensitivity and signal-to-noise of the measurements.

Imaging of the propagation medium can be achieved in many ways, from classical delay-and-sum beamforming to advanced model-based algorithms. On the one hand, the former approach is easy to implement and computationally efficient, but it suffers from an inherent lack of accuracy due to ambiguous or badly interpreted travel times between the different waves. This results in unwanted artefacts in the image that may lead to a wrong interpretation of the medium structure. On the other hand, despite heavier computations the latter approach does not suffer from such ambiguities and allow super-resolution images to be achieved. I focus mainly of model-based methods such as the Full Waveform Inversion or Bayesian approaches. the benefit of model-based Bayesian imaging is presented in figure 1 for diffuse waves and in figure 2 for guided waves.

Figure 1 - a) A 15 cm large x 15 cm long x 30 cm high block of concrete with 9 sensors and a drilled hole. b) A measured diffuse waveform with the theoretical envelope. Application of the LOCADIFF imaging method with c) Bayesian inversion and d) linear inversion. In collaboration with Eric Larose
figure 2 - 3D profile of a corrosion patch in a 5mm steel plate. a) Laser scan of the true profile and b) Super-resolution Bayesian reconstruction of profile after inspection with the S0 Lamb wave mode at 150 kHz (wavelenght = 35 mm)

2 - Imaging with ambient seismic noise

At the seismic scale, the use of active sources is difficult and quite restrictive. However, this difficulty can be overcome be replacing the sources with receivers that play the role of virtual sources, via ambient noise correlation techniques (M. Campillo and A. Paul, Science, 2003). Moreover, with current technology it is possible to deploy thousands of seismometers on a given region, which was impossible only 10 years ago due to very high costs. The combination of these two factors allows the above-mentioned investigation methods to be applied at the seismic scale as well. For example, figure 3 shows the velocity of the surface wave on the San Jacinto Fault, measured with a dense array of geophones.

Figure 3 - a) An array of 1108 sensors deployed on the San Jacinto Fault around the Clark branch, South California (from Ben Zion et & al. GJI 2015) and b) Surface wave velocity profile obtained after post-processing the ambient noise data

3 - Seismic guided waves

Seismic waves are like any elastic wave in the sense that they are sensitive to the elastic properties (velocity, density) of the propagation medium. In the Earth’s crust, at scales varying from a few meters to a few kilometres, these waves may be trapped in thin layers of the structure. Typically this phenomenon occurs as soon as the thickness of the layers becomes of the order of the wavelength. In that case these thin layers behave like waveguides. In addition to being sensitive to the elastic properties of the medium, guided waves are also sensitive to the geometry of the waveguide. Potentially, guided waves can therefore be used to characterize thin geophysical and geotechnical structures, such as Fault zones, sea ice, glaciers urban soils etc.

However, contrary to the mm scale, at the km scale measuring guided waves poses a practical difficulty, because spatial sampling is required. A decade ago this difficulty made guided waves almost impossible to detect, because the arrangement of seismic stations was essentially limited to sparse configurations for the measurement of surface or body waves only. However, the new generations of seismometers allow deployments of large and dense arrays that are well adapted to measure seismic guided waves.

My main long-term research project is to develop innovative strategies that allow the measurement of guided waves at the Earth scale; and to solve the inverse problem in order to characterize the mechanical and geometrical properties of geophysical waveguides. These strategies will first be developed on experimental data, obtained via laboratory-scale waveguides, and then applied to actual seismic data.

a) The 247 stations of the array, including the main central array, and the four linear arrays to the north, east, south and west. Arrows indicate the positions of ground penetrating radar profiles. b) Aerial view of the main array, with station numbering and a photo of one geophone. The two crosses indicate stations that were installed but failed to record: 125-542 and 113- 509. Station 133-517 was originally a 1C station, but was replaced by a spare 3C station due to technical problems. Red circles are for 1C stations and blue circles for 3C stations. The large arrowheads indicate the positions of ice thickness measurements.

During March 2019, I conducted a seismic experiment on sea ice in Svalbard (Norway). An array of geophones was deployed on sea ice and left to record the seismic noise for four weeks. The goal was to demonstrate that sea ice thickness and elastic properties can be monitored with seismic noise. Figure 4 shows the coordinates of the deployment in Arctic, and an aerial view of the main array.

An example of icequake recording and its propagation through the array is shown in figure 5:

Top: Waveform of an icequake recorded at the center of the main array. Bottom: representation of the vertical displacement component through the main array at the times marked with a red cross in the signal. Higher frequencies arrive before the lower frequencies. This dispersion is typical of the quasi-Scholte mode, a mode propagating at the interface of the sea ice layer and water.

To learn more about this expedition and the results, please visit the Icewaveguide project webpage