DEEPtrigger

Slow slip detection with deep learning in multi-station raw geodetic time series validated against tremors in Cascadia

Angelique Carrara — 2023-04-25T10:11:09Z

Reference: G. Costantino, S. Giffard-Roisin, M. Radiguet, M. Dalla Mura, D. Marsan and A. Socquet (2023). Slow slip detection with deep learning in multi-station raw geodetic time series validated against tremors in Cascadia, Nature Communications Earth & Environment, submitted.

Slow slip events (SSEs) originate from a slow slippage on faults that lasts from a few days to years. A systematic and complete mapping of SSEs is key to characterizing the slip spectrum and understanding its link with coeval seismological signals. Yet, SSE catalogues are sparse and usually remain limited to the largest events, because the deformation transients are often concealed in the noise of the geodetic data. Here we present the first multi-station deep learning SSE detector applied blindly to multiple raw geodetic time series. Its power lies in an ultra-realistic synthetic training set, and in the combination of convolutional and attention-based neural networks. Applied to real data in Cascadia over the period 2007-2022, it detects 78 SSEs, that compare well to existing independent benchmarks: 87.5% of previously catalogued SSEs are retrieved, each detection falling within a peak of tremor activity. Our method also provides useful proxies on the SSE duration and may help illuminate relationships between tremor chatter and the nucleation of the slow rupture. We find an average day-long time lag between the slow deformation and the tremor chatter both at a global- and local-temporal scale, suggesting that slow slip may drive the rupture of nearby small asperities.

Overview of the performance of SSEdetector on real raw GNSS time series. The blue curves show the probability of detecting a slow slip event (output by SSEdetector) in 60-day sliding windows centered on a given date. Grey bars represent the number of tremors per day, smoothed (gaussian smoothing, σ = 2 days) in the grey curve. Red horizontal segments represent the known events catalogued by Michel et al., 2019. The (a) panel shows the global performance of SSEdetector over 2007-2022. The red arrow indicates the time window analyzed by Michel et al., while the green arrows describe the two periods from which the synthetic training samples have been derived. The grey rectangle indicates the period which was not covered by the PNSN catalog. In this period, data from Ide, 2012 has been used. The (b), (c) and (d) panels show zooms on 2016-2017, 2019-2021 and 2017 (January to July), respectively.

Interseismic Coupling and Seismic Potential on the South Peru Subduction Megathrust

Angelique Carrara — 2023-04-03T11:54:00Z

Reference: B. Lovery, M. Chlieh, E. Norabuena, J.C. Villegas, M. Radiguet, N. Cotte, M. Langlais, A. Tsapong-Tsague, W. Quiroz, Ciro Sierra Farfán, M. Simons, J.M. Nocquet, H. Tavera, and A. Socquet, (2023). Interseismic Coupling and Seismic Potential on the South Peru Subduction Megathrust, Journal of Geophysical Reasearch, submitted.

The Central Andes subduction zone has been the theater of numerous large megathrust earthquakes since the beginning of the 21th Century, starting with the 2001 Mw=8.4 Arequipa, 2007 Mw=8.0 Pisco and 2014 Mw=8.1 Iquique earthquakes. Here, we present an analysis of 47 permanent GNSS time series and 26 survey GNSS measurements acquired in Central-South Peru between 2007 and 2022 to better understand the frictional properties of the megathrust interface. Using a trajectory model that mimics the different phases of the cycle, we extract a coherent interseismic GNSS field at the scale of the Central Andes from Lima to Arica (12°S - 18.5°S). GNSS-derived interseismic models on a 3D slab geometry indicate that the level of locking is relatively high and concentrated between 20 and 40 km depth. Locking distributions indicate a high spatial variability of the coupling along the trench axis, with the presence of many locked patches that spatially correlate with the seismic segmentation of that subduction. Our study confirms the presence of a creeping segment where the Nazca Ridge enters in subduction; we also observe a more tenuous apparent decrease of coupling related to the Nazca Fracture Zone (NFZ). However, since the Nazca Ridge appears to behave as a strong barrier, the NFZ is relatively weak and less efficient to arrest seismic rupture propagation. Considering various uncertainties on the data and models, we discuss the implication of our coupling estimates with the size and timing of large and great megathrust earthquakes considering both deterministic and probabilistic approaches. We estimate that the South Peru segment could host a maximum magnitude Mmax 9.0 earthquake depending principally on the considered seismic catalog and the seismic/aseismic slip ratio.

Interseismic coupling distribution from our inversion of interseismic velocities in the PS frame, with three different smoothing (λ = 15, 30 and 200 km). Slab contours (dashed lines) are reported every 20 km depth, from the Slab2 model (Hayes et al. 2018). The interseismic coupling is highly heterogeneous reflecting strong variations of the frictional properties of the slab interface. Thus, there are four highly-coupled areas: a very-highly coupled area close to Lima, a less highly-coupled area near Ica, a large band of high-coupling between the Nazca Ridge and the Nazca Fracture Zone, and a highly-coupled area between the Nazca Fracture Zone and the Arica bend. Discontinuities can be observed where the Nazca Ridge and the Nazca Fracture Zone are subducting below the South America plate. The approximate rupture area of the Mw=8.4 2001 Arequipa has been estimated in this study, while the Mw=8.0 2007 Pisco earthquake rupture is reported from Chlieh et al. (2011). On the right panel are displayed the moment deficit rate by latitude for models in the PS and SSA frame, and with various smoothing (λ = 15, 30 and 200 km), computed over 10-km wide segments. Patches with gray hachure pattern depict areas that are not well resolved by the model. Rupture extents from major seismic events are reported as blue segments.

Largest aftershock nucleation driven by afterslip during the 2014 Iquique sequence

Angelique Carrara — 2023-04-03T09:37:00Z

Reference: Itoh, Y., Socquet, A., & Radiguet, M. (2023). Largest aftershock nucleation driven by afterslip during the 2014 Iquique sequence. EarthArXiv, DOI

Various earthquake source models predict that aseismic slip modulates the seismic rupture process. However, observations of aseismic slip associated with earthquakes are scarce, which has left the earthquake source model controversial. Here, we characterise seismic and aseismic processes for 3 days during the 2014 Iquique earthquake sequence in northern Chile by analysing seismicity and crustal deformation time series measured by high-rate Global Positioning System (GPS). We demonstrate that the early afterslip started immediately after the M 8.1 mainshock and led to the largest M 7.6 aftershock 27 hours later, located 120 km to the south. At the mainshock latitude, the interevent early afterslip is located downdip of the mainshock rupture, and is associated with aftershocks. These afterslip and aftershocks exhibit a rapid temporal decay. In contrast, south of the mainshock slip patch, a peak of afterslip separates the mainshock rupture from the largest aftershock, suggesting that this area acted as a barrier to the southward propagation of the mainshock rupture. Seismicity count and moment accelerate in this southern area during the interevent stage. We conclude that the largest aftershock nucleation was driven by the interevent afterslip. The mechanical connection between sequential great earthquakes can therefore be mediated by aseismic slip.

Figure caption: (a) Cleaned GPS time series (coloured dots) with a trajectory model fit (black curve). Red and purple vertical lines indicate timing of the 2014 Iquique mainshock and the largest aftershock. The site location is shown as a red solid square in (b). (b) Intervevent GPS displacements between the two earthquakes (black vectors) and model displacements (blue vectors) from afterslip model shown in (c). Red, purple and orange stars indicate the epicentres of the mainshock, the largest aftershock and an M 6.1 aftershock, respectively (Soto+2019JGR). Black dots indicate moderate aftershocks (McBrearty+2019). (c) Slip distribution of the mainshock (red), interevent afterslip (blue), the largest aftershock (purple) and 2-day afterslip following the largest aftershock (green). Black dots indicate GPS sites used for the slip inversion. (d-e) afterslip distribution for the first and second halves of the interevent period (blue contours) with outlines of the mainshock and the largest aftershock slip area (red and purple curves). Blue, green and black dots indicate moderate aftershocks during the corresponding windows (see (f) for meaning of colours). (f) Latitude-time evolution of interevent moderate aftershocks. (g) Temporal evolution of seismicity count (top) and geodetic and seismicity moments (bottom) in the two regions north and south of 20.2°S as labelled. Each curve is normalised with respect to their cumulative value. This figure is modified from Itoh+2023EarthArXiv.

Cargese 2023 school

Angelique Carrara — 2023-03-21T10:37:43Z

- Workshop

Kick-off meeting

Angelique Carrara — 2023-03-21T10:08:36Z

Slow, Deep, Large-scale trigger?

Webmaster — 2021-04-28T10:12:00Z

Subduction zones host the world's largest earthquakes and tsunamis. Understanding how such earthquakes initiate and interact is a first-order challenge in earth sciences.

The objective of the DEEPtrigger project is to study subduction zones recently affected by megathrust events or earthquakes sequences, considering the mechanisms of deformation in the larger subduction system and their potential role in pushing the megathrust to failure.

View online : Read more

Scientific collaborations

Angelique Carrara — 2021-04-26T09:41:46Z

Instituto Geofisico del Péru

Edmundo Norabuena
Hernando Tavera
Juan Carlos Villegas

Pontificia Universidad Católica de Chile

Marcos Moreno

Universidad de Chile

Juan Carlos Baez
Francisco Ortega Culaciati

Engineers

Angelique Carrara — 2021-04-26T09:41:23Z