How to cite: Kheirdast, N., Almakari, M., Villafuerte, C., Thomas, M. Y., Cheng, J., Gupta, A., and Bhat, H. S.: Fault volume digital twin to reproduce the full slip spectrum, scaling and statistical laws, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11136, https://doi.org/10.5194/egusphere-egu26-11136, 2026.

Compared to continental strike-slip faults, oceanic transform faults (OTFs) are thought to mainly slip aseismically and host significantly more foreshocks likely triggered by precursory aseismic slip which enhance the mainshocks' short-term predictability. However, long-term high-resolution observational constraints remain limited. In December 2024, one of the largest ever OTF earthquakes occurred offshore California on the Mendocino OTF. Here we show that compared to similar-magnitude continental strike-slip earthquakes, this moment magnitude (Mw) 7.0 earthquake has an order of magnitude fewer aftershocks which suggests limited inter-event stress triggering. Nevertheless, the aftershock zone expanded with logarithmic time substantially beyond the mainshock's co-seismic rupture zone, hence likely reflects propagating aseismic slip transients. However, foreshock activity within the mainshock's rupture zone is limited and does not indicate any accelerating aseismic slip in the preceding 30 days. The 2016 Mw 6.6 and 1994 Mw 7.0 Mendocino OTF earthquakes share similar aftershock and foreshock characteristics. The 15 historical Mw>5.5 mainshocks also have few foreshocks on average. Our results demonstrate that low-seismic-coupling OTF segments can host aseismic slip transients triggered by earthquakes on neighboring segments while inhibiting these seismic ruptures’ propagation, and enhanced foreshock activity is not a general characteristic of OTFs despite prevalent aseismic slip.

How to cite: Liu, M., Liu, H., and Tan, Y. J.: Large Mendocino transform fault earthquakes’ foreshock and aftershock characteristics, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3364, https://doi.org/10.5194/egusphere-egu26-3364, 2026.

Foreshocks are occasionally detected prior to earthquakes, but their influence on rupture nucleation is still poorly understood. Standard nucleation models generally attribute earthquake initiation to slow, quasi-static slip driven by fault weakening, and often disregard impulsive precursory slip events. In contrast, we demonstrate through laboratory experiments combined with a rate-and-state, Griffith-type rupture framework that foreshocks can exert a first-order control on earthquake initiation when they occur at, or during, the nucleation phase. Our results show that foreshock-induced slip bursts impose a transient sliding velocity, denoted Vmin​, whose amplitude depends on foreshock size and systematically governs both the duration and spatial extent of nucleation. Larger foreshocks produce higher Vmin​ values and promote a rapid progression toward dynamic rupture, whereas smaller foreshocks lead to prolonged quasi-static nucleation, and sufficiently weak perturbations result in complete rupture arrest. When extrapolated to tectonic fault conditions, the framework predicts that foreshock sequences and accompanying slow slip preceding natural earthquakes obey similar scaling relationships. These findings constrain characteristic nucleation slip distances to approximately 0.3–3 mm, substantially smaller than those typically inferred for dynamic rupture. Overall, our study indicates that transient slip induced by foreshocks controls the timing, evolution, and observability of earthquake nucleation.

How to cite: Fryer, B., Garagash, D., Lebihain, M., and Passelègue, F.: From Foreshocks to Rupture: Transient Slip Controls on Nucleation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7390, https://doi.org/10.5194/egusphere-egu26-7390, 2026.

Earthquake prediction remains one of the most challenging problems in Earth science. Recent advances in physics-based fault modelling, high-resolution laboratory observations, and deep-learning frameworks have opened new opportunities to assess how predictable seismic processes may be in controlled environments.

Here, we present the analysis of more than 1,000 laboratory earthquakes produced in a biaxial apparatus hosting a 400 × 100 mm PMMA fault interface, allowing two-dimensional rupture propagation analogous to natural faults. The experimental setup is instrumented with 38 strain gauges distributed within the fault interface, 20 accelerometers located along both fault surfaces, and 14 acoustic emission (AE) sensors positioned at varying distances from the fault. The experiments were conducted under constant loading rate and normal stresses ranging from 50 to 250 bar. This dense instrumentation enables us to reconstruct, for each laboratory earthquake, the nucleation location, initiation time, rupture evolution, and final event magnitude. The resulting catalog spans nearly three orders of magnitude in seismic moment (from M_w=-6.5 for small ruptures to  M_w=-3.8 for the largest events).

Building on this comprehensive dataset, we explore the potential of Graph Neural Networks to predict the spatial and temporal occurrence of laboratory seismicity. The models are trained on a subset of experiments and tested on independent experiments not included in the training phase. We focus in particular on identifying the minimal set of observational features required for successful prediction, and on assessing the level of physical complexity that machine-learning algorithms trained on homogeneous laboratory faults can capture.

How to cite: Passelegue, F., Paglialunga, F., Bletery, Q., Fryer, B., and Arzu, F.: How predictable are laboratory earthquakes? Insights from dense fault instrumentation and graph neural networks, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11501, https://doi.org/10.5194/egusphere-egu26-11501, 2026.

Repeating earthquakes are pairs or families of events that rupture an identical patch of a fault repeatedly, having recurrence times from days to several years. 
They are to be distinguished from quasi-repeaters whose source areas are non or only partly overlapping. 
We report on a particular quasi-repeater phenomenon:  repeated burst repeaters, which we call raspberry repeaters. In contrast to regular repeater series, each rupture phase consists of multiple events (the burst) rather than just one event. 
In a burst phase, multiple events occur in a cascade over a short time period followed by a notably longer waiting time. In our case the waiting time between bursts range from weeks to years, while the activity during a burst is usually smaller than six hours. We identified over 20 of such raspberry repeater series in northern Chile. 
For multiple series we relocate the events and analyze their source properties as well as their inter-event interactions in detail. We find that they neither obey a classical mainshock-aftershock pattern, nor the diffuse pattern of earthquake swarms. Interestingly, several groups show remarkably consistent repeating pattern, i.e., their sub-cluster rupture order remains similar. 
The identification and description of raspberry repeater series can improve our understanding of subduction related failure and earthquake generation mechanisms, of stress transfer and triggering processes between earthquakes. 

How to cite: Folesky, J., Kummerow, J., and Chen, K.: Repeated Burst Repeating Earthquakes in North Chile.        From long waiting times for short bursts., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18211, https://doi.org/10.5194/egusphere-egu26-18211, 2026.

Repeating earthquakes provide constraints on fault slip and loading processes along subduction zones. We analyze a comprehensive repeating earthquake catalog from northern Chile spanning more than two decades (Folesky et al., 2025), consisting of 3153 repeating earthquake sequences with magnitudes ranging from −0.3 to 4.7. These sequences cluster at two depth intervals: a shallow group (<70 km) and an intermediate-depth group (70–210 km), spanning from the plate interface to within the subducting slab.

We compare the recurrence behavior and slip-rate response of shallow and intermediate-depth repeating earthquakes. Shallow repeaters show strong sensitivity to large megathrust earthquakes. Following the 2014 Mw 8.1 Iquique earthquake, inferred slip rates accelerated to peak values of up to 51.96 cm/yr, then decayed over approximately five years to a quasi-steady level of 3.3449 cm/yr, and eventually returned toward a background rate of 0.549 cm/yr. In contrast, intermediate-depth repeating earthquakes exhibit little systematic response to large megathrust events.

Despite these contrasting responses, both shallow and intermediate-depth repeaters record comparable background slip rates of ~0.5–1.0 cm/yr. Along-strike and space–time analyses further indicate that north–south variability at intermediate depth is expressed primarily in recurrence patterns rather than in slip-rate amplitude. These results demonstrate pronounced depth-dependent differences in repeating earthquake behavior and provide new observational constraints on fault slip processes from the shallow megathrust to intermediate depths.

How to cite: Carpio, A. J., Chen, K. H., Peng, W., and Folesky, J.: Depth-Dependent Recurrence and Slip-Rate Behavior of Repeating Earthquakes in Northern Chile, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18482, https://doi.org/10.5194/egusphere-egu26-18482, 2026.

The origin of tectonic tremors, low-amplitude and long-duration seismic signals, observed at depth greater than 20-30km at some of the world’s tectonic plate boundaries, remains enigmatic. Here, olivine + antigorite mineral assemblages, containing up to 75 vol.% of the hydrous bearing phase, thought as analogues for dry and water-rich subducting lithologies, were experimentally compressed along pressure-temperature (P-T) paths typical of hot subduction zones. During all experiments, ultrasonic acoustic monitoring of the compression was performed. At PT conditions below 1GPa and 500°C, earthquake-like signals were recorded, with a peak activity at 0.5GPa and 250°, ie. conditions coresponding to the base of the megathrust . Above these PT conditions, spectral analysis (corner frequency, stress drop, duration vs. moment) revealed that the recorded acoustic emissions (AE) signals shared striking similarities with natural tectonic tremors. In particular, stress drops of few kPa and linear moment release vs. duration scalings were observed. While nominally dry experiments confirmed that these tremor-like AEs originated from the viscous deformation of the dry matrix, hydrous mineral bearing experiments demonstrated that experimental tremors could be triggered in bursts at the onset of dehydration reactions, probaly via a mechanism compatible with dehydration stress transfer.

How to cite: Schubnel, A., Zverev, P., Yano, S., Gasc, J., John, T., Kummerow, J., Labrousse, L., and Ide, S.: Experimental tectonic tremors triggered at subduction zone conditions , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12296, https://doi.org/10.5194/egusphere-egu26-12296, 2026.

Tectonic tremor swarms are commonly observed at depths near the brittle-ductile transition at convergent plate boundaries. Composed of many temporally overlapping low frequency earthquakes (LFEs), these swarms extend over distances of 5 to 500 km and persist over times ranging from 1 to 1000 hours. The largest swarms have been correlated with slow slip earthquakes and we assume here that smaller swarms also serve as proxies for slow slip events. Swarms are characterized by their area A, their duration T, their scalar seismic moment 𝑀₀ (and corresponding moment magnitude m), the number of their constituent LFEs 𝑁_𝑒, and their along-strike propagation velocity 𝑣. These parameters have been linked in the literature by the following five scaling relations: 1) the scalar moment of a swarm is proportional to its duration, 𝑀₀ ~ 𝑇,  2) the number of swarms 𝑁_𝑠 follows the Gutenberg-Richter (G-R) frequency-magnitude relation, 𝑁_𝑠 = 10^{𝑎−𝑏𝑚} with b =1, 3) the number of swarms is a power law function of their duration, 𝑁_𝑠 ~ 𝑇^−2/3, 4) the number of swarms is a power law function of the number of events in a swarm, 𝑁_𝑠 ~ 𝑁_𝑒^−2/3 , and 5) the along-strike velocity of a swarm scales with its duration 𝑣 ~ 𝑇^−0.8. We demonstrate here that if scaling law (1) is correct then scaling law (3) is equivalent to the G-R distribution (2) with b = 1. If the moment is proportional the number of events in the swarm, 𝑀₀ ~ 𝑁𝑒, then scaling law (4) is also equivalent to the G-R distribution (2) with b = 1. Further, if 𝑑̅ ~ 𝑀₀^1/6, as observed for repeating earthquakes on the San Andreas Fault, then scaling law (5) can be written as 𝑑̅ ~ 𝐿 where 𝑑̅ is the average displacement and L is the along-strike fault length. The relation 𝑑̅ ~ 𝐿 implies that a slow earthquake behaves more like a crack than like a self-healing slip pulse often used to describe normal earthquakes, a result that is consistent with the observation of rapid tremor reversals. Finally, the emergent relation 𝑀₀ ~ 𝑁_𝑒 provides a possible explanation for scaling law (1) 𝑀₀ ~ 𝑇, and a fractal distribution of swarm sizes with dimension D = 1.6 leads to the observed G-R relation with b = 1. This fractal dimension characterizes the early stages of fragmentation, consistent with the idea that tremor is the seismic signature of the breakup and underplating of subducting oceanic crust.

How to cite: Sammis, C. and Bostock, M.: A Note on Empirical Scaling Laws for Tremor Swarms in Subduction Zones, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5817, https://doi.org/10.5194/egusphere-egu26-5817, 2026.

The Ecuadorian subduction zone is one of the few subduction zones where aseismic slip occurs in the shallow segment of the megathrust fault. This aseismic slip appears to be characterized by seismic swarms. So far, no non-volcanic tremor has been detected using classical methods. This may be partially attributed to the fact that the previous deployments were mainly on land and only sparsely offshore, away from the expected locus of potential tremors.

During the HIPER marine campaign (2022, 15/03-12/04), we deployed around 40 OBS on a 3D grid to  image the structure of the Ecuadorian subduction zone in the region of the 2016 Mw 7.8 Pedernales earthquake. 

An automatic machine-learning CNN model was developed, relying on modulation spectrum representations of the seismic signals acquired from the OBS network. This approach is rooted in the detection of typical/atypical patterns in animal vocalizations or human speech, as it has been demonstrated to be highly effective in profiling and detecting the "natural" variations from noise - how the modulation patterns (the “timbre” and “prosody”) evolve around the carrier frequency (the “pitch"). 

Thus, the representation dataset in our approach consists of streams of time-varying images 2D+t (carrier vs modulation frequencies) computed for each unidimensional directional seismic time series. This approach was tested and proved to be both discriminatory and efficient in validating the detection of tremors obtained on OBS seismic signals extracted from the SEIS-PNSN tremors dataset from the Cascadia subduction zone.

For the first time, we have recorded seismic activity on a dense offshore network over a one-month-long period, which will reveal whether tremors occurred in the region of the Pedernales earthquake, a region which is prone to aseismic and seismic slip.

How to cite: Gillert, A., Lebrun, J., Galve, A., Font, Y., and Laigle, M.: Automatic detection of atypical seismic events through machine learning models trained on modulation spectrum representations of OBS datasets, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18113, https://doi.org/10.5194/egusphere-egu26-18113, 2026.

Shallow tectonic tremors along the northeastern Japan subduction zone show regional differences in their spatiotemporal evolution, raising the question of whether their response to tidal stressing also varies along strike. We analyze the tremor catalogue obtained by Sagae et al. (JGR, 2025, e2025JB031348)  for the period from August 2016 to August 2024. Based on their spatial distribution, tremor activity can be divided into three major regions: the southern end of the Kuril Trench (40.8–42°N; northern region), the northern Japan Trench (38.8–40.5°N; central region), and the southern Japan Trench (35–36.8°N; southern region). Here, we investigate the tidal sensitivity of tectonic tremors in these three regions. Our statistical analysis shows that tidal sensitivity is highest in the northern area, where tremors are clustered and occur in recurrent along-strike propagating bursts. Cluster-scale analyses in this northern region indicate that tidal sensitivity increases during the later stages of tremor clusters, consistent with the characteristics reported for deep tectonic tremors. Tidal sensitivity is intermediate in the southern area, where tremors appear more scattered. In the central region, where tremor activity has declined gradually since 2011 Mw 9.0 Tohoku-Oki earthquake, tidal sensitivity is lowest. In this region however, tremors and fast earthquakes occur in close spatial proximity. There, we further examine the relationship between tremor activity, fast earthquakes and tidal stress to explore potential interactions between slow and fast earthquakes.

How to cite: Zhou, Y., Aochi, H., Schubnel, A., Ide, S., Bhat, H., Lu, W., Yano, S., and Gupta, A.: Tidal sensitivity of shallow tectonic tremors in northeastern Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9848, https://doi.org/10.5194/egusphere-egu26-9848, 2026.

Slow-slip events (SSEs) release an important part of the accumulated strain at plate boundaries and can interact with large earthquakes. It is thus crucial to analyse in detail their temporal dynamics. While GNSS observations robustly capture the cumulative, static displacements associated with SSEs, their noise level limits the temporal resolution of transient, short-timescale potential variations in slip rate. In contrast, co-occurring tectonic tremor, sampled at much finer temporal resolution, reveals pronounced short-term intermittency within SSEs. Despite this, tremor-derived temporal variability has not yet been incorporated into kinematic SSE models, leaving the short-timescale dynamics largely unresolved.

In the Mexican subduction zone, large SSEs persist for several months,recur every few years, and GNSS-based kinematic models resolve only first-order spatiotemporal evolution at these long timescales. Here we investigate the spatio-temporal evolution of the 2009–2010 SSE sequence in Guerrero, Mexico. This sequence is of particular interest because it consists of two distinct sub-events, with the onset of the second coinciding with the occurrence of the distant Maule earthquake. A detailed kinematic analysis of this SSE, combining geodetic observations and tremor activity, therefore provides a unique opportunity to assess the potential role of dynamic stress perturbations during large SSEs.

We construct a tremor catalog covering the SSE sequence, using the temporary mini-array seismic network GGAP and a beamforming method at the tremor frequency band.

In parallel, we use GNSS time series from the local network, to jointly analyze the SSE crustal displacement signal with the resulting tremor catalog to observe the finer dynamics of the SSE sequence.

We develop two kinematic slip modeling schemes based on a least-squares formulation with regularization. In the first scheme, GNSS positions on fixed time windows are inverted sequentially as independent time steps. In the second scheme, the full GNSS time series are inverted simultaneously, which improves the recovery of displacement amplitudes and allows the incorporation of tremor-derived temporal constraints. Tremor burst timings are defined based on events clustering properties, and introduced as prior information in the kinematic inversion, allowing larger slip rates during tremor dense periods.

Our results show that the studied 2009-2010 SSE sequence includes 7 major tremor bursts that are accompanied by a slip acceleration. In 2009, two major episodes of tremor and slow-slip occurred in the westernmost part of Guerrero. Immediately following the 2010 Maule earthquake, a persistent and energetic tremor and slip episode was triggered, extending the slipping region eastward along strike, where multiple additional tremor and slip episodes were subsequently observed.

Although aseismic slip releases the largest moment, the accompanying tremor provides a high-resolution temporal proxy for fault slip. This enables improved temporal resolution in kinematic SSE models, and allows the identification of short-term slip accelerations that coincide with tremor timings. The complex 2009–2010 Guerrero slow-slip and tremor sequence analyzed here highlights the sensitivity of SSE slip rates and migration to far-field dynamic stress perturbations.

How to cite: El Yousfi, Z., Radiguet, M., Rousset, B., and Zigone, D.: Tremor-informed kinematic slip modeling of the 2009-2010 slow-slip event doublet in the Mexican subduction zone, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19567, https://doi.org/10.5194/egusphere-egu26-19567, 2026.