Louise Cordrie

Earthquake magnitude‐frequency distributions exhibit significant space‐time variations, which can provide critical insights into the physical processes driving seismicity. Understanding these variations is crucial for assessing seismic hazards and uncovering the physical processes driving earthquakes. One key parameter, the b ‐value, describes the relative proportion of small to large earthquakes and is thought to reflect factors such as stress conditions and fault properties. However, empirical evidence linking b ‐value variations to physical processes in real tectonic settings is still limited. Here, we show that b ‐value is systematically higher in regions with elevated heat flow, consistently across different tectonic settings and faulting style. This suggests that thermal conditions play a fundamental role in controlling earthquake size distributions, controlling the likelihood of large earthquakes in the different areas.

We examined sedimentary records in a coastal lagoon on Saint Martin Island in the Lesser Antilles to identify and characterize extreme-wave events (EWEs), such as hurricanes and tsunamis. Employing a comprehensive approach involving sedimentological, geochemical, and radiocarbon dating analyses, complemented by X-ray computed microtomography (micro-CT) for examining sediment fabrics, we applied this multiproxy method to three oriented short sediment cores along a transect. This allowed us to identify sediment layers linked to both tsunami- and hurricane-induced EWEs. Five out of the seven EWEs were identified as paleo-tsunamis through their geochemical, sedimentary, and structural signatures. These five paleotsunamis were successfully dated over the last 3500 years, including the well-documented Pre-Columbian tsunami at approximately 1400 yrs CE and the transatlantic Lisbon tsunami at 1755 CE. This suggests a tentative local tsunami chronology of five well-documented events over the last 3500 years with a recurrence interval of 400 to 500 years.However, the most recent EWE corresponded to the powerful Category 5 Hurricane Irma in 2017. Over the last 150 years, another 14 less intense hurricanes impacted the island, leaving no sediment imprints in the lagoon. This finding is in line with most recent publications showing that tropical storm intensification rates have already changed as anthropogenic greenhouse gas emissions have warmed the globe (Garner, 2023).Furthermore, micro-CT-based sediment analysis provided a deeper understanding of the relationship between sediment fabric and tsunami wave dynamics. Therefore, we used the deposits of the Pre-Columbian tsunami and compared paleo-flow directions from micro-CT-derived fabric patterns to those from numerical tsunami models. The results that best explain the Pre-Columbian tsunami deposit emplacement are in line with a Mw 8.5–8.7 megathrust earthquake source located on the subduction interface at the Puerto Rico Trench, north of Anegada Island (Cordrie et al., 2022). Ultimately, integrating deposits of EWEs with numerical models is pivotal for devising effective hazard mitigation strategies tailored to vulnerable coastal communities.Cordrie, L., Feuillet, N., Gailler, A., Biguenet, M., Chaumillon, E., Sabatier, P., 2022. A Megathrust earthquake as source of a Pre-Colombian tsunami in Lesser Antilles: Insight from sediment deposits and tsunami modeling. Earth Sci Rev 228, 104018. https://doi.org/10.1016/j.earscirev.2022.104018Garner, A.J. Observed increases in North Atlantic tropical cyclone peak intensification rates. Sci Rep 13, 16299 (2023). https://doi.org/10.1038/s41598-023-42669-y

The Urgent Tsunami Computing procedures discussed herein are designed to quantify potential hazards resulting from seismically-induced tsunamis following an earthquake, with a temporal scope ranging from minutes to a few hours. The presented workflow employs comprehensive simulations, encompassing the entire tsunami propagation process, while accounting for uncertainties associated with source parameters, tsunamigenesis and wave propagation dynamics. Within the EuroHPC eFlows4HPC project, we present a High-Performance Computing (HPC) workflow tailored for urgent tsunami computation in which the Probabilistic Tsunami Forecast (PTF) code has been restructured and adapted for seamless integration into a PyCOMPSs framework. This framework enables parallel execution of tasks and includes simulations from Tsunami-HySEA numerical model within a unified computational environment. Of particular significance is the workflow's capability to incorporate new datasets, such as focal mechanism data, seismic records, or real-time tsunami observations. This functionality facilitates an "on-the-fly" update of the PTF, ensuring that the forecasting model remains responsive to the latest information. The development of this workflow involves a systematic exploration of diverse scenarios, realistic simulations, and the assimilation of incoming data. The overarching goal is to rigorously diminish uncertainties, thereby producing updated probabilistic forecasts without compromising precision and enhancing risk mitigation efforts far from the seismic source. Improved risk management, achieved by informing decision-making in emergency situations, underscores the importance of this development. We will showcase the technical advancements undertaken to tailor the workflow for HPC environments, spanning from the developers' perspective to that of the end user. Additionally, we will highlight the scientific enhancements implemented to leverage the full potential of HPC capabilities, aiming to significantly reduce result delivery times while concurrently enhancing the accuracy and precision of our forecasts.

Tsunami urgent computing procedures quantify the potential hazard due to a seismically-induced tsunami right after an earthquake, that is from minutes to a few hours. The hazard is quantified by simulating the tsunami from source to shore, taking into account the uncertainty in the source parameters and the uncertainty associated with the wave generation, propagation, and inundation.In the European eFlows4HPC project, an HPC workflow for urgent computing of tsunami hazard assessment is currently being developed, consisting of the following steps: 1) retrieval of parameters for the tsunamigenic earthquake (magnitude, hypocentre and associated uncertainties), 2) definition of a seismic source ensemble, 3) simulation of the tsunami generated by each scenario in the ensemble, 4) aggregation of the results to produce an estimate of tsunami hazard, which also incorporates a basic treatment of uncertainty modelling and 5) update of the ensemble based on incoming data.Initially implemented on the Power-9 machine at BSC, the workflow has been fully embedded into a PyCOMPSs framework that enables parallel task execution and integrates full tsunami simulations for the first time. The tsunami numerical model (Tsunami-HySEA) computes the tsunami from the source to coastal impact using nested grids with resolution from kilometres to meters.To limit the number of simulations and converge faster towards stable hazard estimates, new methods for defining the seismic source ensembles have been developed. When applied to several past earthquakes and tsunamis (e.g., the 2003 Boumerdes and the 2017 Kos-Bodrum earthquakes), our new sampling strategy yielded a reduction of 1 or 2 orders of magnitude for ensemble size, allowing a drastic reduction in the computational effort. This reduction may be exploited to improve tsunami simulation accuracy, increasing the computational effort available for each simulation for the same overall cost. The workflow also allows the integration of new incoming data (focal mechanism, seismic or tsunami records) for an “on the fly” update of the PTF based on this new information.The improvement of the workflow through a well-defined ensemble of scenarios, realistic simulations and integration of incoming data, strongly reduces the uncertainty and yields to an update of the probabilistic forecasts without compromising theiraccuracy. This can be crucial in mitigating the risk far from the seismic source, and in improving risk management by better informing decision-making in an emergency framework.

Tsunami urgent computing procedures quantify the potential hazard due to an earthquake right after its occurrence, that is within a few hours. The hazard is quantified by simulating the propagation of the tsunami waves in the sea, accounting for the uncertainty due to the scarce knowledge of the source parameters and wave modelling uncertainty.In the context of the European project eflows4HPC, a workflow is currently in development for tsunamis hazard urgent computing, which consists of the following steps: 1) Retrieval of information about the tsunamigenic seismic event (magnitude, hypocentre and their uncertainties); 2) Definition of an ensemble of seismic sources; 3) Simulation of seismic/tsunamigenic waves propagation for each scenario in the ensemble; 4) Results aggregation to produce an estimate of seismic and tsunami hazard, which also incorporates a basic treatment of modelling uncertainty. The ensembles cover the uncertainty on source characteristics and may consequently be very large (generally 10,000 to 100,000 of scenarios; Selva et al., Nat. Comm.), requiring very high computational resources for the urgent computing context. It is thus necessary to reduce the size of these ensembles to limit the number of simulations and to converge faster towards stable results of hazard calculation.We developed and tested several sampling procedures aiming to reduce the number of scenarios in the ensemble and, at the same time, to integrate the new incoming information as they become available (e.g. solutions for focal mechanisms, seismic or tsunami records). When applied to several past earthquakes and tsunamis (e.g., the 2003 Boumerdes and the 2017 Kos-Bodrum earthquakes), our novel sampling strategies yielded a reduction of 1 or 2 order of magnitudes of the ensemble size, allowing a drastic reduction of the computational effort. Also, the update of the ensemble based on the incoming of new data, which strongly reduce the uncertainty, yields to an update of the probabilistic forecasts without compromising its accuracy. This may result very important for mitigating the risk far from the seismic source, as well as improving the risk management by better informing decision making in a frame of urgency.

The arc of the Lesser Antilles is one of the most quiet subduction zone in the world. In this region, the convergence of the Atlantic and the Caribbean plates is low (few mm/year) and most of the seismicity is an intraplate and crustal seismicity. Among the Mw>7 earthquakes recorded in the historical catalog (1690 near Barbuda, 1843 near Guadeloupe, 1867 near the Virgin Islands, 1839 offshore Martinica, 1969 offshore Dominica, 1974 near Antigua), only the 1839 and 1843 events are suspected to be interplate earthquakes. The 1867 Virgin Island earthquake generated an important tsunami with waves of 10m that devastated the closest islands. A tsunami followed the 1843 earthquake but without much damage. These two events are the only known damaging tsunami in this region, but another older one might be added to the list. Indeed, an increasing number of tsunami deposits have been identified in the recent years on several islands of the arc, all of them being around 500 years old (~1450 AD). These deposits are all located in the northern segment of the arc, between Antigua and Puerto-Rico, in Anegada, St-Thomas (Virgin Islands), Anguilla and Scrub islands. There is unfortunately no record and no testimonies of an extreme event at that time.The northern segment of the arc is particularly complex because located at the transition between the Greater Antilles and the Lesser Antilles. It is crossed by the Anegada Passage, a series of faults and basins cutting through the arc, which defines the limit between the Puerto-Rico micro-plate and the Caribbean plate. This passage and the numerous intra-arc fault systems present between the islands are active and likely compensate for the plates motion. The very low slip deficit detected with GPS measurements at the subduction contacts of Puerto-Rico and the Lesser Antilles indicates that the interface from Guadeloupe to Puerto-Rico can be considered as totally uncoupled or holding the characteristics of a very long seismic cycle. A tsunami generated by an extreme event 500 years ago in this region could be related to intra-arc, outer-rise, intraplate or interface fault rupture. The identification of the source would enable a better understanding of the seismic cycle and the dynamic of this part of the arc.This study lists and set models of all the potential faults that could trigger an earthquake in the area encompassing the three islands : Anguilla, Anegada and StThomas. We have created high-resolution bathymetric grids and performed tsunami simulations for each fault model. We uses run-up models to compare the simulated wave heights and run-up distance to all the deposits heights and positions. The magnitudes of our fault models range between 7 and 9, but very few of them generate a strong enough tsunami to match the observed deposits.