Postdoctoral in passive imaging of the crust M/F

Job Description

Missions

Crustal tomography of the Betics (outh Spain) with passive imaging approaches(local earthquake tomography and ambient noise tomography)

Activity

Objective: Obtain high-resolution seismic images of the Ronda and Beni Bousera peridotite bodies and their host terranes using two complementary methods conducted in parallel:
Surface wave tomography based on seismic ambient noise (ANT).
This approach will yield a 3D S-wave velocity model with relatively homogeneous lateral resolution across the study region. Although the resolution is lower than local earthquake tomography, especially in depth, ANT is particularly robust in mapping large-scale structures and provides more realistic amplitude estimates of seismic velocity anomalies. Importantly, it is well suited to detecting the high-velocity anomalies expected for the peridotite massifs, and can serve as a consistent background model for more detailed tomographic studies.
Local earthquake tomography (LET) using P- and S-wave arrival times.
LET provides higher-resolution images than ANT, particularly in depth, making it valuable for resolving dipping structures and detailed crustal features. However, the method is highly dependent on ray coverage: regions with dense seismicity and station coverage are imaged in great detail, while areas with sparse illumination remain poorly resolved. Additionally, the amplitude of velocity anomalies tends to be underestimated relative to their true magnitude.
Together, ANT and LET are complementary methods: ANT provides a broad, amplitude-reliable framework, while LET offers fine-scale detail where ray coverage is sufficient. Their integration maximizes resolution and reliability, producing a balanced view of the peridotite bodies' geometry and their tectonic context.
Program structure: The work proceeds in two stages, with ANT and LET running simultaneously in each stage. Stage 1 exploits existing datasets to deliver baseline models quickly.
Stage 2 augments both methods with additional data (TT component correlations for Love waves in ANT, and new phase picks from temporary arrays in LET) to deliver refined updates. In parallel, Mantle8 may deploy a passive seismic acquisition campaign in the region. These data will be generated independently of the collaboration contract. However, if they are collected within the project timeframe, they will, after internal processing, be shared and integrated into the study to further improve the results.

1. Surface Wave Tomography Using Seismic Ambient Noise (ANT)

Stage 1 (Months 1–5): Baseline ANT from existing ZZ cross correlations
Measure group and phase velocities of fundamental-mode Rayleigh waves from ZZ cross correlations of ambient noise.
Using existing continuous seismic data, we have computed ambient noise cross correlations on the vertical–vertical (ZZ) component. From these cross correlations we will extract dispersion curves (group and phase velocities) for the fundamental mode Rayleigh wave. These measurements will be made using an automatic implementation of FTAN (frequency-time analysis).
Quality control of measurements.
Because the dataset includes a very large number of cross correlations, the dispersion measurements will be made automatically. To ensure reliability, the results will undergo a careful quality control process to detect and remove bad measurements, inconsistent picks, and statistical outliers. This step is crucial to avoid biasing the tomographic inversion.
Obtain 2D maps of Rayleigh wave group and phase velocities. 
The validated dispersion data will be inverted to produce two-dimensional tomographic maps of Rayleigh-wave group and phase velocities at discrete periods between 4 and 40 s. These period bands are sensitive to crustal and uppermost mantle depths, allowing us to resolve the lateral structure of the Ronda and Beni Bousera peridotite massifs as well as their surrounding terranes. The 2D maps will be obtained using both the classical (linearized) and transdimensional approaches.
1D transdimensional inversion for shear-wave velocity on a grid, resulting in a 3D model.
The dispersion maps will then be inverted using a transdimensional Bayesian framework, applied to a regular grid of 1D profiles. This non-parametric method allows the model complexity to adapt to the data, providing robust estimates of shear-wave velocity with quantified uncertainties. The ensemble of 1D profiles will be assembled into a 3D shear-wave velocity model.

Stage 2 (Months 5–11): ANT refinement with TT correlations and Love+Rayleigh measurements and transdimensional 2D maps
Compute new TT (transverse–transverse) cross correlations to extract Love-wave dispersion (group and phase velocities).
In addition to the ZZ correlations used in Stage 1, we will now compute cross correlations of the transverse component (TT). These provide sensitivity to Love waves, allowing us to measure both group and phase velocity dispersion curves and thereby expand the dataset to include complementary shear-wave information.
Obtain 2D maps of Rayleigh and Love wave velocities using transdimensional tomography.
The dispersion measurements from both Rayleigh and Love waves will be inverted with a transdimensional Bayesian tomography approach. Unlike conventional linearized inversions, this method avoids imposing arbitrary parameters such as smoothing and instead allows model complexity to be guided directly by the data. This is particularly important for this study, since the target anomalies are relatively small and can be spread before its real extent when using smoothing. An additional advantage is that it provides realistic, data-driven estimates of uncertainties, which are essential for quantifying the robustness of velocity anomalies.
Jointly invert Rayleigh and Love dispersion to update the 3D Vs structure.
By combining Rayleigh and Love wave datasets in a joint inversion, we will update the 3D shear-wave velocity model. This approach improves sensitivity to shallow shear properties and provides constraints on potential anisotropy (transverse isotropy) associated with deformation structures in and around the peridotite massifs. The refined model will thus capture both the overall geometry and subtle fabric of the lithosphere with greater reliability than Stage 1.
Integrate the new dataset acquire and processed internally by Mantle8 to improve locally the resolution of the ANT

2. Local Earthquake Tomography (LET)

Stage 1 (Months 1–5): Baseline LET from existing catalogs
Integrate earthquake catalogs from IGN (Spain), CNRST (Morocco), and IPMA (Portugal); address duplicate readings.
We will compile and merge existing earthquake catalogs from the main national seismic networks in Spain, Morocco, and Portugal. During this process, particular care will be taken to identify and remove duplicate readings — cases where the same station reports the same event in more than one catalog. Cleaning these redundancies ensures that each arrival time is represented only once, preventing bias and over-weighting in the tomographic inversion.
Data selection for tomography.
From the combined catalog, we will select a subset of well-recorded earthquakes to be used in the tomography. The selection criteria will include events with a sufficient number of reliable P- and S-phase picks, adequate azimuthal coverage of stations around the epicenter, and robust depth control. This curated dataset will maximize the quality of the inversion and the resolution of the resulting seismic velocity models.
Invert for 3D Vp and Vs models and generate baseline resolution diagnostics.
Using the selected events, we will apply an inversion method that simultaneously solves for both the 3D velocity structure and earthquake locations. This approach minimizes trade-offs between hypocentral errors and velocity anomalies. The inversion will produce baseline three-dimensional models of P- and S-wave velocities for the study region. Resolution diagnostics (checkerboard and spike tests, hit count maps) will be carried out to evaluate the robustness of the models and identify well-resolved versus poorly constrained areas.
Stage 2 (Months 5–11): LET refinement with temporary experiments
Gather waveform archives from major temporary deployments (primarily TOPO-IBERIA and PICASSO).
We will collect continuous waveform data from large-scale temporary experiments, with a focus on TOPO-IBERIA and PICASSO, which together provide excellent coverage across the Betic–Rif–Alboran region. These deployments significantly increase the density of stations compared to the permanent networks, enhancing the potential resolution of tomographic imaging beneath the Ronda and Beni Bousera peridotites.
Use a deep-learning phase picker (PhaseNet) to (re)pick P and S arrivals.
PhaseNet will be applied systematically to the waveforms from both temporary and permanent stations. This dual strategy ensures not only the extraction of high-quality phase arrivals from the temporary arrays, but also the recovery of additional picks from permanent stations that may not have been included in the regional catalogs. Deep learning algorithms are particularly effective for identifying S-wave arrivals, which are sometimes missed or mis-picked in routine catalogs.
Address duplicate and overlapping picks.
As the re-picking process involves both temporary and permanent stations, special care will be taken to identify and eliminate duplicate picks (instances where the same arrival has been registered multiple times across different processing streams). This step will also allow us to directly compare the new deep-learning picks with the original catalog picks, offering a systematic evaluation of their relative quality and potential improvements.
Re-invert to obtain updated Vp and Vs models with enhanced resolution.
The new, combined set of high-quality P and S arrivals will be used to perform a new tomographic inversion that simultaneously solves for earthquake locations and velocity structure. The updated 3D models of P- and S-wave velocities are expected to significantly improve resolution beneath the peridotite bodies and surrounding terranes, particularly at depth.
Provide resolution maps and uncertainty estimates; compute difference volumes (Stage 2 minus Stage 1).
To evaluate the improvements gained in Stage 2, resolution and uncertainty analyses will be carried out (checkerboard and spike tests, hit count maps). Difference volumes between the Stage-2 and Stage-1 models will highlight new features resolved by the enhanced dataset, thereby quantifying the added value of re-picking and temporary deployments.
Integrate the new dataset acquire and processed internally by Mantle8 to improve locally the resolution of the LET

Your Profil

Skills

Passive imaging methods

Your Work Environment

The Gibraltar Arc hosts two major peridotite outcrops: the Ronda peridotite massif in southern Spain and the Beni Bousera peridotite in northern Morocco. These peridotites are of high interest for natural hydrogen exploration, as serpentinization processes within ultramafic rocks are considered a key source of H₂ generation. However, the geometry of these bodies remains poorly constrained, including their lateral extent beyond the surface outcrops, their depth continuation, and their relationship with surrounding geological units. This project aims to use advanced seismic imaging methods to characterize the three-dimensional structure of these peridotite bodies, thereby improving our understanding of their potential for natural hydrogen resources.

Compensation and benefits

Compensation

041,58 to 4216,70 euros depending on the experience of the candidate

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions