A Detailed Picture of Haiti’s Seismicity Given by Deep Learning and Template Matching

Abstract 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.

Authors

Institutions

Publication Details

Journal
Seismological Research Letters
Published
2026-07-02
DOI
https://doi.org/10.1785/0220250412
Primary Topic
Seismology and Earthquake Studies
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

A Detailed Picture of Haiti’s Seismicity Given by Deep Learning and Template Matching

Miguel Neves, Jérôme Chèze, Quentin Blétery, Sylvert Paul et al.
Seismological Research Letters
Seismology and Earthquake Studies
article

A Detailed Picture of Haiti’s Seismicity Given by Deep Learning and Template Matching

Miguel Neves, Jérôme Chèze, Quentin Blétery, Sylvert Paul, David Ambrois, Bryan Raimbault, Tony Monfret, E. Calais, Steeve Symithe, Françoise Courboulex, Louis De Barros
article en

Abstract

Abstract 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.

Seismological Research Letters
Centre National de la Recherche Scientifique (FR), University of Haiti (HT), École Normale Supérieure (BI), Institut Català de Ciències del Clima (ES), Institut des Sciences de la Terre (FR), Observatoire de la Côte d’Azur (FR), Université Gustave Eiffel (FR), Laboratoire de Géologie de l’École Normale Supérieure (FR), Géoazur (FR), Institut de Recherche pour le Développement (FR), University of Bergen (NO), Université Savoie Mont Blanc (FR), Université Grenoble Alpes (FR)
Openalex Percentile: Top 5%
Seismology and Earthquake Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.