Sophie Giffard-Roisin

Sophie Giffard-Roisin

Auto-generated profile, based on publicly available information

Sophie Giffard-Roisin

Univ. Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, IRD, Univ. Gustave Eiffel, ISTerre, Grenoble, France

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 May 2018 in IEEE Transactions on Biomedical Engineering
Verified
Journal Article
Closed access
SGSophie Giffard-Roisin
Sophie Giffard-Roisin
HDHerve Delingette
Herve Delingette
TJThomas Jackson
Thomas Jackson
JWJessica Webb
Jessica Webb
LFLauren Fovargue
Lauren Fovargue
MSMaxime Sermesant
Maxime Sermesant
Goal: Noninvasive cardiac electrophysiology (EP) model personalisation has raised interest for instance in the scope of predicting EP cardiac resynchronization therapy (CRT) response. However, the restricted clinical applicability of current methods is due in particular to the limitation to simple situations and the important computational cost. Methods: We propose in this manuscript an approach to tackle these two issues. First, we analyze more complex propagation patterns (multiple onsets and scar tissue) using relevance vector regression and shape dimensionality reduction on a large simulated database. Second, this learning is performed offline on a reference anatomy and transferred onto patient-specific anatomies in order to achieve fast personalized predictions online. Results: We evaluated our method on a dataset composed of 20 dyssynchrony patients with a total of 120 different cardiac cycles. The comparison with a commercially available electrocardiographic imaging (ECGI) method shows a good identification of the cardiac activation pattern. From the cardiac parameters estimated in sinus rhythm, we predicted five different paced patterns for each patient. The comparison with the body surface potential mappings (BSPM) measured during pacing and the ECGI method indicates a good predictive power. Conclusion: We showed that learning offline from a large simulated database on a reference anatomy was able to capture the main cardiac EP characteristics from noninvasive measurements for fast patient-specific predictions. Significance: The fast CRT pacing predictions are a step forward to a noninvasive CRT patient selection and therapy optimisation, to help clinicians in these difficult tasks.

10 July 2021 in Journal of Geophysical Research: Solid Earth
Verified
Journal Article
Open access
PDF file
JMJosipa Majstorović
Josipa Majstorović
SGSophie Giffard‐Roisin
Sophie Giffard‐Roisin
PPPiero Poli
Piero Poli
In this study, we developed an end‐to‐end two‐stage pipeline using 1D convolutional neural networks (CNNs) to detect, localize, and characterize earthquakes from single‐station three‐component waveforms. We are presenting here the insights of what makes the difference in developing a deep learning algorithm by performing an extensive hyperparameter grid search for model training, tackling the question of the optimal number of classes, the importance of training data sets as well as the CNN architecture design in terms of optimal length of the CNN model. Moreover, our pipeline is robust and does not need any preprocessing of the seismograms (e.g., filtering) or any prior knowledge of the region. Training, validation, and evaluation of the CNN models is performed on data recorded at the AQU station placed in the city of L’Aquila in the Abruzzo region (Central Italy). Before MW 6.3 2009 L’Aquila earthquake that occurred near the city of L’Aquila the local catalog of events is sparse. Therefore, we provide a methodological pipeline on how one can extend near‐fault local catalog of earthquakes by applying our two‐stage pipeline on unseen continuous data. Our results show that we are able to design a CNN model that is detecting the earthquake events among random noise waveforms with 97% accuracy (first stage: detection). Additionally, we are able to determine the events that are close to the station (<10 km) with a 94% accuracy as well as identify their belonging to four magnitude classes with a 68% accuracy (second stage: characterization).

19 September 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
GCGiuseppe Costantino
Giuseppe Costantino
SGSophie Giffard-Roisin
Sophie Giffard-Roisin
Mauro Dalla MuraMauro Dalla Mura
Mauro Dalla Mura
ASAnne Socquet
Anne Socquet
Geospatial data has been transformative for the monitoring of the Earth, yet, as in the case of (geo)physical monitoring, the measurements can have variable spatial and temporal sampling and may be associated with a significant level of perturbations degrading the signal quality. Denoising geospatial data is, therefore, essential, yet often challenging because the observations may comprise noise coming from different sources, including both environmental signals and instrumental artifacts, which can be spatially and temporally correlated, thus hard to disentangle. This study addresses the denoising of multivariate time series acquired by irregularly distributed networks of sensors, requiring specific methods to handle the spatiotemporal correlation of the noise and the signal of interest. Specifically, our method focuses on the denoising of geodetic position time series, used to monitor ground displacement worldwide with centimeter-to-millimeter precision. Among the signals affecting GNSS data, slow slip events (SSEs) are of interest to seismologists. These are transients of deformation that are weakly emerging compared to other signals. Here, we design SSEdenoiser, a multi-station spatiotemporal graph-based attentive denoiser that learns latent characteristics of GNSS noise to reveal SSE-related displacement with sub-millimeter precision. It is based on the key combination of graph recurrent networks and spatiotemporal Transformers. The proposed method is applied to the Cascadia subduction zone, where SSEs occur along with bursts of tectonic tremors, a seismic rumbling identified from independent seismic recordings. The extracted events match the spatiotemporal evolution of tremors. This good space-time correlation of the denoised GNSS signals with the tremors validates the proposed denoising procedure.

17 November 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
TMTristan Montagnon
Tristan Montagnon
JHJames Hollingsworth
James Hollingsworth
EPErwan Pathier
Erwan Pathier
MMMathilde Marchandon
Mathilde Marchandon
Mauro Dalla MuraMauro Dalla Mura
Mauro Dalla Mura
SGSophie Giffard-Roisin
Sophie Giffard-Roisin
Optical satellite imagery is widely used for estimating ground movement in the aftermath of natural disasters such as earthquakes. This type of imagery enables detailed analysis of the factors and mechanisms that drive or influence these events. By using sub-pixel correlation algorithms, it provides precise displacement measurements (in the meter-to-centimeter range) and high spatial resolution (decimeter-to-centimeter level) by comparing images taken before and after the event. In this study, we present a deep neural network approach, trained on our new specific realistic dataset FaultDeform, to retrieve full-scale seismic ground motion displacement fields from optical satellite images with sub-pixel precision. The FaultDeform dataset, available at https://doi.org/10.57745/G02ZXZ, is the first satellite synthetic dataset tailored for ground motion estimation. We introduce the GeoFlowNet pipeline, utilizing a U-net architecture to solve the displacement estimation problem, delivering high-speed performance through GPU implementation, and outperforming current correlators in speed and precision. Comprehensive comparisons with state-of-the-art methods such as COSI-Corr, MicMac and CNN-DIS, and validation on real-world data from the 2019 Ridgecrest and 2013 Balochistan earthquakes showcases the robustness of our method. Codes are freely available: gricad-gitlab.univ-grenoble-alpes.fr/montagtr/GeoFlowNet.

9 September 2025 in Journal of Geophysical Research: Solid Earth
Verified
Journal Article
Open access
JMJannes Münchmeyer
Jannes Münchmeyer
WFWilliam B. Frank
William B. Frank
SGSophie Giffard‐Roisin
Sophie Giffard‐Roisin
DMDavid Marsan
David Marsan
FTFrederik Tilmann
Frederik Tilmann
ASAnne Socquet
Anne Socquet
Subduction megathrusts release stress not only seismically through earthquakes, but also through creep and transient slow deformation, called slow slip events (SSEs). Understanding the interplay between fast and slow slip is essential for illuminating the deformation processes on the subduction interface. The Chilean subduction margin, while one of the most seismically active regions worldwide, has few reports of SSEs. Furthermore, there are no comprehensive reports of tectonic tremors or low‐frequency earthquakes (LFEs), seismic signals typically accompanying SSEs, tracking deformation at small spatial and temporal scales. Here, we perform a systematic search for tectonic tremors and LFEs in the Atacama segment in Northern Chile, a region hosting both shallow and deep SSEs. Using dense seismic networks, we investigate 3.3 years between November 2020 and February 2024. Due to the network geometry, we focus on deep tremor and LFEs. We apply two complementary methods, envelope correlation for tremor search and deep learning detection for LFEs, to generate initial catalogs. To validate the potential detections, we use clustering, matched filtering, heuristics, and extensive manual inspection. While our initial search provides numerous candidates, after verification, we find no evidence for tectonic tremor or LFEs in the region. In contrast, our approaches successfully recover tremors and LFEs in two reference regions outside Chile with known tremor and LFE activity. Our observations show that tremors and LFEs in Northern Chile are either of lower moment rate than in other regions, have substantially longer recurrence rates, or are absent altogether, potentially due to the cold subduction.

8 July 2025 in Seismica
Verified
Journal Article
Open access
PDF file
LPLéa Pousse-Beltran
Léa Pousse-Beltran
TLTheo Lallemand
Theo Lallemand
LALaurence Audin
Laurence Audin
PLPierre Lacan
Pierre Lacan
ANAndres David Nunez-Meneses
Andres David Nunez-Meneses
SGSophie Giffard-Roisin
Sophie Giffard-Roisin
Geomorphic markers such as displaced surfaces, offset rivers or scarps are witnesses to the neotectonic activity of the faults.The characterization (such as fault detailed surface trace, the scarp height, etc.) of these geomorphological markers is currently a time-consuming step with expert-dependent results, often qualitative and with uncertainties that are difficult to estimate. To overcome those issues, we present a proof of concept study for the use of deep learning in morphotectonics, specifically on fault markers. We developed a Bayesian supervised machine learning method using one-dimentional (1D) convolutional neural networks (CNN) trained on a database of simulated topographic profiles across normal fault scarps, called ScarpLearn. From a topographic profile, ScarpLearn is able to automatically give the cumulative scarp height with an uncertainty. We have developed two versions: one designed for more generalized cases involving profiles with multiple fault scarp (ScarpLearn), and another specifically trained to handle profiles featuring a single fault scarp (ScarpLearn_1F). We apply ScarpLearn for the characterization of active normal faults in extensional settings such as the Trans-Mexican Volcanic Belt and Malawi Rift system. From those specific case studies, we explore the progress (computation time, accuracy, uncertainties) that machine learning methods bring to the field of morphotectonics, as well as the current limits (such as bias). Our results show that we are able to develop a CNN model that is estimating scarp heights on topographic profiles from 5m resolution digital elevation model. We compared the results obtained with ScarpLearn and other non deep-learning methods. ScarpLearn achieves similar accuracy while being much faster and having smaller uncertainties. We invite readers to use and to extend our study: codes to build the synthetic scarp database and for the CNN model ScarpLearn are available at: https://gricad-gitlab.univ-grenoble-alpes.fr/poussel/scarplearn.