Tony Monfret
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Tony Monfret has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- Eric Calais
Département de Géosciences, École normale supérieure, Université PSL, Paris, France
5 shared publications - Françoise Courboulex
Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS, IRD, Géoazur, Valbonne, France
4 shared publications - Steeve Julien Symithe
URGéo, Faculté des Sciences, Université d’État d’Haïti, Port-au-Prince, Haïti
4 shared publications - Sylvert Paul
Observatoire de la Côte d’Azur, Université Côte d’Azur, IRD, CNRS, Geoazur, Sophia Antipolis, Valbonne, France
4 shared publications - David Ambrois
Observatoire de la Côte d’Azur, Université Côte d’Azur, IRD, CNRS, Geoazur, Sophia Antipolis, Valbonne, France
3 shared publications - Dominique Boisson
LMI CARIBACT, Urgéo, Faculté des Sciences, Université d’État d’Haïti, Port-au-Prince HT 6110, Haiti
3 shared publications
Latest Publications
- Open Access
- 2 July 2026
- Published by Seismological Society of America (SSA) in Seismological Research Letters
A Detailed Picture of Haiti’s Seismicity Given by Deep Learning and Template Matching
Haiti regularly experiences destructive earthquakes, but seismic monitoring in the region has historically been limited. Recent deployments of citizen-hosted RaspberryShake seismometers and temporary seismic deployment following the 2021 Mw 7.2 earthquake provide new data to study the region’s seismotectonics. However, high noise levels at many stations, in particular the RaspberryShake ones, limit detection, hence the fault imaging capability of these instruments. This study explores the use of a Deep Learning denoising algorithm, DeepDenoiser, to improve their seismic signal and earthquake detection capabilities. We find that DeepDenoiser raises the average signal-to-noise ratio of seismic signals by 4.7 dB and increases earthquake detections, but also raises false detections when using short-term average/long-term average and deep learning detection methods. Template matching, however, when combined with DeepDenoiser, yields more true detections and fewer false detections than traditional band-pass-filtered waveforms. This suggests that DeepDenoiser is better suited for retrospective studies than for real-time applications. Using DeepDenoiser and template matching, we compile a 2 yr, high-resolution earthquake catalog for Haiti containing about three times the number of events of the original catalog. The improved catalog furthers our understanding of the 2021 Mw 7.2 earthquake sequence, highlighting particularly clearly the segmented nature of the aftershock distribution with a generally northeast-dipping cluster in the east that coincides with the hypocenter and first reverse phase of the rupture, and a series of aftershocks farther west that coincide with the mostly strike-slip phase of the rupture. The improved catalog also reveals potentially fluid-induced offshore seismic swarms in the Jérémie basin and active seismicity below Lake Enriquillo in the Dominican Republic. This catalog advances our understanding of the region’s seismicity and provides further opportunities to study the larger regional tectonic context.
- 2 July 2026
- Published by Seismological Society of America (SSA) in Seismological Research Letters
Open AccessA Detailed Picture of Haiti’s Seismicity Given by Deep Learning and Template Matching
Haiti regularly experiences destructive earthquakes, but seismic monitoring in the region has historically been limited. Recent deployments of citizen-hosted RaspberryShake seismometers and temporary seismic deployment following the 2021 Mw 7.2 earthquake provide new data to study the region’s seismotectonics. However, high noise levels at many stations, in particular the RaspberryShake ones, limit detection, hence the fault imaging capability of these instruments. This study explores the use of a Deep Learning denoising algorithm, DeepDenoiser, to improve their seismic signal and earthquake detection capabilities. We find that DeepDenoiser raises the average signal-to-noise ratio of seismic signals by 4.7 dB and increases earthquake detections, but also raises false detections when using short-term average/long-term average and deep learning detection methods. Template matching, however, when combined with DeepDenoiser, yields more true detections and fewer false detections than traditional band-pass-filtered waveforms. This suggests that DeepDenoiser is better suited for retrospective studies than for real-time applications. Using DeepDenoiser and template matching, we compile a 2 yr, high-resolution earthquake catalog for Haiti containing about three times the number of events of the original catalog. The improved catalog furthers our understanding of the 2021 Mw 7.2 earthquake sequence, highlighting particularly clearly the segmented nature of the aftershock distribution with a generally northeast-dipping cluster in the east that coincides with the hypocenter and first reverse phase of the rupture, and a series of aftershocks farther west that coincide with the mostly strike-slip phase of the rupture. The improved catalog also reveals potentially fluid-induced offshore seismic swarms in the Jérémie basin and active seismicity below Lake Enriquillo in the Dominican Republic. This catalog advances our understanding of the region’s seismicity and provides further opportunities to study the larger regional tectonic context.
- 23 March 2026
- Published by Seismological Society of America (SSA) in Seismological Research Letters
- Vol. 97 (5) , 3275-3288
Integrating Submarine DAS into a Regional Seismic Network for Enhanced Offshore Earthquake Location: A Case Study from the Canary Islands
Repurposing a submarine telecommunication cable as a distributed acoustic sensor (DAS) significantly enhances earthquake monitoring in the Canary Archipelago, where much of the seismicity occurs offshore. We interrogated two ∼60 km segments of a dark-fiber cable between Tenerife and Gran Canaria from 26 July to 29 September 2020. P and S waves of local earthquakes were first detected using a deep learning model for phase picking and then merged with arrival-time readings from the Instituto Geográfico Nacional seismic network. To prevent overweighting the dense DAS array relative to the sparser land-based stations, we designed a selection strategy that uses a machine-learning-based channel quality index algorithm to automatically retain a limited number of high-signal-to-noise DAS channels per event. The selected channels are spaced according to the actual interstation distances of the National Seismic Network on the islands. We relocated 31 earthquakes (1.1≤mbLg≤3.2) using a nonlinear, probabilistic earthquake location method. The addition of DAS data narrowed the median azimuthal gap by ∼12° and reduced the median depth uncertainty from over 12 to 3.4 km. Within the Enmedio submarine volcano corridor, the hybrid DAS-land network detected 66% of events with mbLg≥1.5, and all events above mbLg=2.0 within 50 km of the cable. These results demonstrate that even a modest number of well-selected DAS channels can significantly improve the precision of earthquake location in submarine environments, providing a scalable and operationally compatible path for enhancing offshore seismic and volcanic monitoring. The reduced volume of data also ensures that DAS integration does not overwhelm existing processing pipelines, making it feasible for use in real-time alert systems.
- 23 March 2026
- Published by Seismological Society of America (SSA) in Seismological Research Letters
- Vol. 97 (5) , 3275-3288
Integrating Submarine DAS into a Regional Seismic Network for Enhanced Offshore Earthquake Location: A Case Study from the Canary Islands
Repurposing a submarine telecommunication cable as a distributed acoustic sensor (DAS) significantly enhances earthquake monitoring in the Canary Archipelago, where much of the seismicity occurs offshore. We interrogated two ∼60 km segments of a dark-fiber cable between Tenerife and Gran Canaria from 26 July to 29 September 2020. P and S waves of local earthquakes were first detected using a deep learning model for phase picking and then merged with arrival-time readings from the Instituto Geográfico Nacional seismic network. To prevent overweighting the dense DAS array relative to the sparser land-based stations, we designed a selection strategy that uses a machine-learning-based channel quality index algorithm to automatically retain a limited number of high-signal-to-noise DAS channels per event. The selected channels are spaced according to the actual interstation distances of the National Seismic Network on the islands. We relocated 31 earthquakes (1.1≤mbLg≤3.2) using a nonlinear, probabilistic earthquake location method. The addition of DAS data narrowed the median azimuthal gap by ∼12° and reduced the median depth uncertainty from over 12 to 3.4 km. Within the Enmedio submarine volcano corridor, the hybrid DAS-land network detected 66% of events with mbLg≥1.5, and all events above mbLg=2.0 within 50 km of the cable. These results demonstrate that even a modest number of well-selected DAS channels can significantly improve the precision of earthquake location in submarine environments, providing a scalable and operationally compatible path for enhancing offshore seismic and volcanic monitoring. The reduced volume of data also ensures that DAS integration does not overwhelm existing processing pipelines, making it feasible for use in real-time alert systems.
- Open Access
- 16 November 2025
- Published by Elsevier in Tectonophysics
- Vol. 918
Precise relocation of the 14 August 2021 M w 7.2 Nippes, Haiti, earthquake sequence using broadband and citizen-hosted short-period seismometers
- 16 November 2025
- Published by Elsevier in Tectonophysics
- Vol. 918
Open AccessPrecise relocation of the 14 August 2021 M w 7.2 Nippes, Haiti, earthquake sequence using broadband and citizen-hosted short-period seismometers
- Open Access
- 28 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 19869-19883
Automatic Identification of High-Quality Channels in Distributed Acoustic Sensing Through Implementation of a Channel Quality Index
Using fiber-optic cables, distributed acoustic sensing (DAS) has emerged as a powerful technology for seismic monitoring, particularly in oceanic regions, but the massive amount of data generated presents challenges for its integration into operational seismic networks. However, since the quality of DAS recordings varies along the fiber due to physical and environmental factors, not all data collected from fiber-optic cables contribute meaningfully to earthquake detection and characterization. Identifying and filtering out low-quality channels reduces unnecessary data processing and optimizes resource use. To address this, we developed a channel quality index (CQI), a machine-learning-based approach that automatically identifies high-quality DAS channels based on their ability to record earthquake signals. Trained on data from four offshore DAS experiments in the Spanish Mediterranean and Canary Islands, our model evaluates temporal, frequency, and energy-based features to assign quality scores to individual channels, reliably identifying high-quality channels with high precision (0.92–0.96) and strong performance in terms of area under the receiver operating characteristic curve (ROC–AUC, 0.92–0.98). By prioritizing high-quality data, CQI enhances the efficiency of data transmission and storage in large-scale DAS deployments. It enables the selective retrieval of relevant seismic data from repositories, reducing storage demands and minimizing data transmission time, which is especially critical for large datasets. It also provides a mechanism to quantify information loss in data compression strategies, ensuring critical seismic signals are preserved. These capabilities of CQI facilitate the practical adoption of DAS for both real-time and long-term seismic monitoring, optimizing data quality and resource management.
- 28 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 19869-19883
Open AccessAutomatic Identification of High-Quality Channels in Distributed Acoustic Sensing Through Implementation of a Channel Quality Index
Using fiber-optic cables, distributed acoustic sensing (DAS) has emerged as a powerful technology for seismic monitoring, particularly in oceanic regions, but the massive amount of data generated presents challenges for its integration into operational seismic networks. However, since the quality of DAS recordings varies along the fiber due to physical and environmental factors, not all data collected from fiber-optic cables contribute meaningfully to earthquake detection and characterization. Identifying and filtering out low-quality channels reduces unnecessary data processing and optimizes resource use. To address this, we developed a channel quality index (CQI), a machine-learning-based approach that automatically identifies high-quality DAS channels based on their ability to record earthquake signals. Trained on data from four offshore DAS experiments in the Spanish Mediterranean and Canary Islands, our model evaluates temporal, frequency, and energy-based features to assign quality scores to individual channels, reliably identifying high-quality channels with high precision (0.92–0.96) and strong performance in terms of area under the receiver operating characteristic curve (ROC–AUC, 0.92–0.98). By prioritizing high-quality data, CQI enhances the efficiency of data transmission and storage in large-scale DAS deployments. It enables the selective retrieval of relevant seismic data from repositories, reducing storage demands and minimizing data transmission time, which is especially critical for large datasets. It also provides a mechanism to quantify information loss in data compression strategies, ensuring critical seismic signals are preserved. These capabilities of CQI facilitate the practical adoption of DAS for both real-time and long-term seismic monitoring, optimizing data quality and resource management.
- 6 September 2023
- Published by Seismological Society of America (SSA) in Seismological Research Letters
- Vol. 94 (6) , 2725-2739
Monitoring of Local Earthquakes in Haiti Using Low-Cost, Citizen-Hosted Seismometers and Regional Broadband Stations
Seismic monitoring in Haiti is currently provided by a mixed network of low‐cost Raspberry Shake (RS) seismic stations hosted by citizens, and short‐period and broadband stations located mainly in neighboring countries. The level of earthquake detection is constantly improving for a better spatio‐temporal distribution of seismicity as the number of RS increases. In this article, we analyze the impact of the quality of the signals recorded by the RS—low‐cost seismometers with the smallest magnitude that the network can detect by studying the ambient noise level at these stations. Because the RS stations are installed as part of a citizen‐science project, their ambient noise estimated by the power spectral density (PSD) method often shows a high‐noise level at frequencies above 1 Hz. In the near field (<50 km), we show that the network detects seismic events of local magnitude on the order of 2.2 with signal‐to‐noise ratios (SNRs) greater than 4. Improving the network detection threshold requires densifying the network with more RS stations in locations that are less noisy, if possible. In spite of these limitations, this mixed network has provided near‐field data essential to rapidly understand the mechanism of the mainshock of the 14 August 2021 7.2 earthquake, to monitor its sequence of aftershocks in near‐real time, and to monitor background seismicity in Haiti on a routine basis.- 6 September 2023
- Published by Seismological Society of America (SSA) in Seismological Research Letters
- Vol. 94 (6) , 2725-2739
Monitoring of Local Earthquakes in Haiti Using Low-Cost, Citizen-Hosted Seismometers and Regional Broadband Stations
Seismic monitoring in Haiti is currently provided by a mixed network of low‐cost Raspberry Shake (RS) seismic stations hosted by citizens, and short‐period and broadband stations located mainly in neighboring countries. The level of earthquake detection is constantly improving for a better spatio‐temporal distribution of seismicity as the number of RS increases. In this article, we analyze the impact of the quality of the signals recorded by the RS—low‐cost seismometers with the smallest magnitude that the network can detect by studying the ambient noise level at these stations. Because the RS stations are installed as part of a citizen‐science project, their ambient noise estimated by the power spectral density (PSD) method often shows a high‐noise level at frequencies above 1 Hz. In the near field (<50 km), we show that the network detects seismic events of local magnitude on the order of 2.2 with signal‐to‐noise ratios (SNRs) greater than 4. Improving the network detection threshold requires densifying the network with more RS stations in locations that are less noisy, if possible. In spite of these limitations, this mixed network has provided near‐field data essential to rapidly understand the mechanism of the mainshock of the 14 August 2021 7.2 earthquake, to monitor its sequence of aftershocks in near‐real time, and to monitor background seismicity in Haiti on a routine basis.- 13 October 2022
- Published by Seismological Society of America (SSA) in Bulletin of the Seismological Society of America
- Vol. 113 (1) , 58-72
Rupture Segmentation of the 14 August 2021 Mw 7.2 Nippes, Haiti, Earthquake Using Aftershock Relocation from a Local Seismic Deployment
The 14 August 2021 7.2 Haiti earthquake struck 11 yr after the devastating 2010 event within the Enriquillo Plantain Garden (EPG) fault zone in the Southern peninsula of Haiti. Space geodetic results show that the rupture is composed of both left‐lateral strike‐slip and thrust motion, similar to the 2010 rupture; but aftershock locations from a local short‐period network are too diffuse to precisely delineate the segments that participated in this rupture. A few days after the mainshocks, we installed 12 broadband stations in the epicentral area. Here, we use data from those stations in combination with four local Raspberry Shakes stations that were already in place as part of a citizen seismology experiment to precisely relocate 2528 aftershocks from August to December 2021, and derive 1D P‐ and S‐crustal velocity models for this region. We show that the aftershocks delineate three north‐dipping structures with different strikes, located to the north of the EPG fault. In addition, two smaller aftershock clusters occurred on the EPG fault near the hypocenter area, indicative of triggered seismicity. Focal mechanisms are in agreement with coseismic slip inversion from Interferometric Synthetic Aperture Radar data with nodal planes that are consistent with the transpressional structures illustrated by the aftershock zones.- 13 October 2022
- Published by Seismological Society of America (SSA) in Bulletin of the Seismological Society of America
- Vol. 113 (1) , 58-72
Rupture Segmentation of the 14 August 2021 Mw 7.2 Nippes, Haiti, Earthquake Using Aftershock Relocation from a Local Seismic Deployment
The 14 August 2021 7.2 Haiti earthquake struck 11 yr after the devastating 2010 event within the Enriquillo Plantain Garden (EPG) fault zone in the Southern peninsula of Haiti. Space geodetic results show that the rupture is composed of both left‐lateral strike‐slip and thrust motion, similar to the 2010 rupture; but aftershock locations from a local short‐period network are too diffuse to precisely delineate the segments that participated in this rupture. A few days after the mainshocks, we installed 12 broadband stations in the epicentral area. Here, we use data from those stations in combination with four local Raspberry Shakes stations that were already in place as part of a citizen seismology experiment to precisely relocate 2528 aftershocks from August to December 2021, and derive 1D P‐ and S‐crustal velocity models for this region. We show that the aftershocks delineate three north‐dipping structures with different strikes, located to the north of the EPG fault. In addition, two smaller aftershock clusters occurred on the EPG fault near the hypocenter area, indicative of triggered seismicity. Focal mechanisms are in agreement with coseismic slip inversion from Interferometric Synthetic Aperture Radar data with nodal planes that are consistent with the transpressional structures illustrated by the aftershock zones.- 6 October 2022
- Published by Seismological Society of America (SSA) in Bulletin of the Seismological Society of America
- Vol. 113 (1) , 252-267
Simulation of Tsunami Inundation for the Island of Martinique to Nearby Large Earthquakes
In this article, we estimate the tsunami hazard in Martinique due to tsunamis generated by earthquakes associated with the Lesser Antilles subduction zone. Using a deterministic approach based on reliable earthquake scenarios, we use high‐resolution bathymetric and topographic data to model tsunami propagation and inundation with Cornell Multi‐grid Coupled Tsunami model. An extreme earthquake subduction scenario of magnitude 8.0 is tested, and a further realistic scenario of lower magnitude 7.5, thus of different tsunami frequency content, is also processed to test the possible appearance of bay resonances. We find that the western coast of the island is relatively sheltered, because it represents a shadow area to diffraction, in particular, for the major city of Fort de France. Because of its very gentle slope, the eastern coast is prone to numerous floodings with meter scale wave amplitudes; most of the inundated zones consist of mangroves and geological depressions t are naturally regularly flooded by tides or storm surges. Hence such areas are often not exploited, the mangroves being let in their natural state, enhancing the protection of the surrounding communities. However, some strategic inhabited areas are subject to severe inundation. Finally, comparing our results with studies of the 1755 Lisbon transoceanic tsunami reveals a tsunami hazard close to our local 7.5 scenario. It suggests the possibility to generalize our local tsunami hazard assessment in Martinique to other tsunami contexts and enlarge its validity. This issue is crucial for minimizing the efforts and increasing the efficiency of tsunami preparedness.- 6 October 2022
- Published by Seismological Society of America (SSA) in Bulletin of the Seismological Society of America
- Vol. 113 (1) , 252-267
Simulation of Tsunami Inundation for the Island of Martinique to Nearby Large Earthquakes
In this article, we estimate the tsunami hazard in Martinique due to tsunamis generated by earthquakes associated with the Lesser Antilles subduction zone. Using a deterministic approach based on reliable earthquake scenarios, we use high‐resolution bathymetric and topographic data to model tsunami propagation and inundation with Cornell Multi‐grid Coupled Tsunami model. An extreme earthquake subduction scenario of magnitude 8.0 is tested, and a further realistic scenario of lower magnitude 7.5, thus of different tsunami frequency content, is also processed to test the possible appearance of bay resonances. We find that the western coast of the island is relatively sheltered, because it represents a shadow area to diffraction, in particular, for the major city of Fort de France. Because of its very gentle slope, the eastern coast is prone to numerous floodings with meter scale wave amplitudes; most of the inundated zones consist of mangroves and geological depressions t are naturally regularly flooded by tides or storm surges. Hence such areas are often not exploited, the mangroves being let in their natural state, enhancing the protection of the surrounding communities. However, some strategic inhabited areas are subject to severe inundation. Finally, comparing our results with studies of the 1755 Lisbon transoceanic tsunami reveals a tsunami hazard close to our local 7.5 scenario. It suggests the possibility to generalize our local tsunami hazard assessment in Martinique to other tsunami contexts and enlarge its validity. This issue is crucial for minimizing the efforts and increasing the efficiency of tsunami preparedness.- 15 April 2022
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 376 (6590) , 283-287
Citizen seismology helps decipher the 2021 Haiti earthquake
On 14 August 2021, the moment magnitude (Mw) 7.2 Nippes earthquake in Haiti occurred within the same fault zone as its devastating 2010Mw7.0 predecessor, but struck the country when field access was limited by insecurity and conventional seismometers from the national network were inoperative. A network of citizen seismometers installed in 2019 provided near-field data critical to rapidly understand the mechanism of the mainshock and monitor its aftershock sequence. Their real-time data defined two aftershock clusters that coincide with two areas of coseismic slip derived from inversions of conventional seismological and geodetic data. Machine learning applied to data from the citizen seismometer closest to the mainshock allows us to forecast aftershocks as accurately as with the network-derived catalog. This shows the utility of citizen science contributing to our understanding of a major earthquake.
- 15 April 2022
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 376 (6590) , 283-287
Citizen seismology helps decipher the 2021 Haiti earthquake
On 14 August 2021, the moment magnitude (Mw) 7.2 Nippes earthquake in Haiti occurred within the same fault zone as its devastating 2010Mw7.0 predecessor, but struck the country when field access was limited by insecurity and conventional seismometers from the national network were inoperative. A network of citizen seismometers installed in 2019 provided near-field data critical to rapidly understand the mechanism of the mainshock and monitor its aftershock sequence. Their real-time data defined two aftershock clusters that coincide with two areas of coseismic slip derived from inversions of conventional seismological and geodetic data. Machine learning applied to data from the citizen seismometer closest to the mainshock allows us to forecast aftershocks as accurately as with the network-derived catalog. This shows the utility of citizen science contributing to our understanding of a major earthquake.