Introduction
Slow earthquakes (or silent earthquakes) are fault-slip events that release seismic energy over durations of hours to months, in contrast to ordinary earthquakes whose main energy discharge occurs in seconds to minutes. Initially recognized through longâterm strain measurements, many slow events are now known to produce measurable seismic signalsâmost notably tremorâwhen continuous seismometer records are processed appropriately. Tremor associated with slow slip commonly emerges after bandâfiltering in the 1â5 Hz range, enabling approximate localization of sources despite the protracted slip history.
Field observations often link slow slip to fluid movement and attendant seismic tremor, suggesting that poreâfluid processes and slow fault slip are coupled in the regions where these phenomena occur. Although sometimes described as âsilent,â slow earthquakes are therefore not instrumentally invisible; they occupy a distinct observational niche as comparatively lowârate, longâduration slip episodes that nevertheless generate strain and tremor signals. Slow earthquakes are distinct from tsunami earthquakes, which produce large tsunamis because of unusually slow rupture propagation yet still concentrate most energy release on the conventional secondsâtoâminutes timescale.
Within seismic taxonomy slow earthquakes constitute one class among many phenomena differentiated by source location, rupture behavior, or temporal sequencing (e.g., mainshocks, foreshocks, aftershocks, interplate/intraplate events, swarms, tsunami or supershear ruptures). Their occurrence can be driven by the same broad causal agents that produce other seismicity types: slip on faults, volcanic processes, or humanâinduced changes such as fluid injection and reservoir impoundment. Describing and interpreting slow events relies on standard seismological vocabularyâepicenter, hypocenter, seismic waves (P and S), epicentral distance, and shadow zonesâand on instruments such as seismometers and strainmeters. However, the extended duration of slow slip complicates traditional magnitude assignments and intensity assessments.
Research coordination and probabilistic forecasting efforts (for example through organized committees and international initiatives) aim to integrate slow slip into broader earthquake forecasting frameworks, while related disciplines and techniquesâshearâwave splitting, elasticâproperty relations (e.g., the AdamsâWilliamson equation), regionalization schemes, earthquake engineering, and sedimentary seismitesâcontribute to mechanistic and hazard understanding.
Earthquakes begin with the gradual build-up of tectonic stress until stress exceeds the strength of the host rocks, producing a brittle rupture and a drop in shear stress that radiates seismic energy as waves through the Earth. Slow earthquakes are most often investigated with mathematical fracture models that represent the event as the growth and propagation of shear cracks and explicitly account for spatial variations in initial stress, frictional resistance on the fault, and the fracture energy required to propagate slip. Model results show that recurring slow events are favored when the margin between initial stress and sliding (frictional) stress is small while the specific fracture energyâa measure of the materialâs resistance to crack propagation relative to the applied loadâis relatively large.
Mechanically, slow earthquakes occupy an intermediate regime between classic asperity-controlled brittle failure and ductile, aseismic flow; they commonly reflect mixtures of transient stickâslip and sustained creep rather than a single failure mode. Small-scale asperities on fault surfaces focus stress and can produce localized brittle behaviour that coexists with slower slip processes, thereby influencing seismic coupling and rupture characteristics. Such asperity-governed intermediate slip is well documented on shallowly dipping subduction megathrusts (for example, SW Japan, Cascadia, and Chile) but also occurs on other fault typesâmost notably major strike-slip boundaries like the San Andreas and large normal faults associated with volcanic flanksâwhere topographic relief and material contrasts promote heterogeneous, complex slip behavior.
Locations
Slow earthquakes occur along faults worldwide but are most abundant at plate boundaries, where convergent (including subduction), divergent, and transform faults create distinct kinematic and deformation regimes. Investigations often emphasize plate-margin subduction zonesâexemplified by studies of the Cascadia subduction system and its crossâsectional geometryâbecause their vertical and lateral fault architecture strongly influences slow-slip behavior. Since the early 21st century, focused research has documented slow-slip events and associated tremor in regions such as Cascadia, California, Japan, New Zealand, Mexico, and Alaska, demonstrating that these phenomena are distributed across diverse plateâboundary settings. Systematic seismic and geodetic monitoring of tremor linked to slow-slip permits mapping of the lateral and depth extent of slipping patches on faults; such spatial delineation yields insights into the mechanics of ordinary (fast) earthquakes and helps constrain the probabilities and characteristic parameters of future seismic events in the studied areas.
Types of slow earthquakes form a continuum of faultâslip behavior that releases tectonic moment with minimal highâfrequency radiation and is concentrated along plate boundaries and major crustal faults, particularly subduction megathrusts and some transform segments. At the small end of this spectrum are lowâfrequency earthquakes (LFEs): brief, discrete seismic bursts embedded within tremor bands that radiate relatively more energy at periods of ~0.1â1 s, have very small seismic magnitudes (commonly near M 0â3), occur episodically or in swarms, and are interpreted as highly localized slip patches on or near the plate interface. At lower frequencies and typically greater depths are very lowâfrequency (VLF) and deep lowâfrequency events, which exhibit longer durations (tens of seconds or more), lower corner frequencies, and often larger seismic moments per event than individual LFEs; these are spatially associated with slowâslip regions on the megathrust or within the subducting slab and are distinguished from volcanic lowâfrequency signals by their tectonic setting.
Aseismic or weakly seismic slow slip events (SSEs) occupy longer timescales, from days to years, and produce measurable surface or nearâsurface deformation detected by continuous GPS, tiltmeters, and strainmeters. Individual SSEs typically accommodate millimetres to tens of centimetres of fault displacement and, in large subduction zones, the integrated moment of some events can be equivalent to an earthquake of about Mw 7. Episodic tremor and slip (ETS) describes coupled occurrences in which nonâvolcanic tremor (largely composed of LFEs and related lowâfrequency signals) coincides spatially and temporally with slow slip; ETS episodes commonly last days to weeks, may migrate along strike, and recur on characteristic intervals in wellâstudied regions such as Cascadia, southwestern Japan, Guerrero, Hikurangi and parts of Alaska.
Distinguishing and mapping these types requires joint analysis of highâsensitivity seismic data (e.g., borehole arrays, spectrogram and envelope detection to resolve LFEs and VLF signals) and geodetic time series (continuous GPS and strain/tilt records to detect SSEs), together with spatial correlation to known plate interfaces. Characterizing their depth distribution, recurrence, slip magnitude and migration behavior is fundamental for understanding transient stress transfer, the loading and unloading of adjacent locked seismogenic patches, and the complex role of aseismic slip in modulating or potentially triggering large, damaging earthquakes.
Low-frequency earthquakes (LFEs) are small (typically M < 3) seismic events whose energy is concentrated at much lower frequencies than ordinary earthquakes, with spectral peaks commonly between about 1 and 3 Hz. Although LFEs occur in volcanic and semiâvolcanic settings, tectonic LFEs associated with subduction plate interfaces are the principal constituentâand in some regions the sole componentâof nonâvolcanic tremor. These events cluster temporally and spatially with slow slip events (SSEs): LFE hypocenters are repeatedly active during long-lived, weeksâtoâmonthâscale slip episodes and tend to migrate along strike in step with the propagating shear slip front of the SSE.
Mechanically, tectonic LFEs are interpreted as small thrustâsense slips on transitional segments of the plate interface located downâdip of the shallow locked seismogenic zone. In many subduction systems they occur at depths of roughly 20â45 km (with shallower occurrences on some strikeâslip faults, such as in California), a depth range that in thermally âwarmâ margins corresponds to a transitional or transient slip zone between the locked and deeper stable slip intervals. This transitional zone often lies near the continental Moho, so LFE clusters effectively map the deep plate contact at or near the crustâmantle boundary. Observations from Cascadia, for example, show LFE hypocenters forming a plane parallel to upperâcrust seismicity but systematically displaced 5â10 km downâdip, supporting an interface origin rather than generation within the overriding crust.
Conceptually, the interplate region relevant to LFEs comprises three downdip domains: a shallow locked zone capable of producing megathrusts, an intermediate transient slip zone hosting SSEs and LFEs, and a deeper stable slip zone accommodating steady aseismic creep. LFEs are highly sensitive to small stress perturbationsâresponding to tidal loading and transient waves from distant earthquakesâand because SSEs change stress on the locked interval some studies infer that the probability of a large earthquake during an SSE may be tens to a hundred times greater than background; nevertheless, no recorded SSE with its associated LFEs has directly coincided with a megathrust rupture. Beyond their potential as precursory signals, LFEs provide valuable constraints on subduction geometry: the repeatable locations and spatial patterns of LFE activity furnish highâresolution markers of the deep plate contact and the position of the Moho, thereby improving tomographic imaging of the plate interface.
History
Lowâfrequency earthquakes entered the seismological record as a distinct class in 1999, when the Japan Meteorological Agency began cataloguing a characteristic lowâfrequency seismic signature. Until that time, tremorâlike signals were largely interpreted within a volcanic frameworkâattributed to partially coupled flow of magmatic fluidsâbecause most recognized lowâfrequency energy had been observed at volcanoes. The identification of similar signals well away from volcanic centers prompted a reassessment of nonâvolcanic origins, especially in subduction environments.
A key development occurred in 2002, when Japanese researchers reported persistent lowâfrequency tremor located near the upper surface of the subducting Philippine Sea plate, directly linking such signals to the plate interface. Those observations were initially explained as dehydrationâdriven tremor, but a 2007 reanalysis revealed that the record contained numerous discrete lowâfrequency earthquake waveforms and concentrated swarms. This reinterpretation led to a conceptual shift: tectonic tremor is largely composed of many small, discrete LFEs. LFEs are now commonly observed in association with slow slip events and have been documented across multiple subduction zones (western North America, Japan, Mexico, Costa Rica, New Zealand), with analogous lowâfrequency activity also reported on shallow strikeâslip faults in California.
Low-frequency earthquakes (LFEs) are seismically distinct from ordinary earthquakes: their P-wave onsets are very weak or indistinct and S-wave arrivals are typically emergent rather than impulsive. This diffuse character â historically noted by agencies such as the JMA â renders LFEs difficult to recognize with conventional earthquake-detection algorithms that depend on clear, highâamplitude body waves; in many cases LFE signals are masked by background noise and fall below the detection capabilities of classical methods.
Contemporary detection therefore relies on waveform-matching techniques that tolerate low signalâtoânoise ratios. Continuous seismic records are scanned by crossâcorrelating them against highâSNR template waveforms created by stacking multiple confirmed LFE traces; stacking suppresses both random and coherent noise and yields a representative template that can be correlated with long time series. Events are identified when correlation coefficients exceed predefined thresholds, allowing systematic recovery of weak, repeating LFE signals that would otherwise be missed.
Resolving source properties of LFEs depends critically on extracting usable Pâwave observations. Firstâmotion polarity analysis of sufficiently clear P arrivals is used to infer fault slip sense, and where measurable these polarities commonly indicate compressional first motions consistent with thrust (reverse) slip. Historically, poor Pâwave picks limited precise hypocentral depths and robust focalâmechanism determinations, but the recent proliferation of sensitive seismic networks has increased the number of reliable Pâwave detections and thereby improved firstâmotion analyses and depth estimates.
LFE depth has been constrained by two complementary approaches: direct hypocenter location using P arrivals and spatial projection of LFE epicenters onto the geometry of the subducting plate. Epicenter-toâslab mapping alone cannot unambiguously discriminate between nucleation on the plate interface versus within the downâgoing slab interior, so robust interpretation requires combining Pâwave locations with slab geometry and other geophysical constraints. When such combined analyses are available, they consistently place LFEs at or very near the plate contact, implicating plateâboundary processes â for example slow or episodic slip on the interface â in their generation.
Low-frequency earthquakes in the Cascadia subduction zone
The Cascadia margin, where the Juan de Fuca, Explorer and Gorda plates subduct beneath North America from northern California to midâVancouver Island, hosts a discrete down-dip band of slow seismicity and transient slip beneath the locked seismogenic zone. Low-frequency earthquakes (LFEs) and associated tremor concentrate on this transition or transient-slip section of the plate interface, typically located beneath the shallow megathrust (ââ¤25 km) and above the deeper stable-slip region. Along-strike variability is pronounced: in southern Cascadia LFEs occur between ~28 and 47 km depth, whereas beneath Vancouver Island the active depth window narrows to roughly 25â37 km.
Seismological and teleseismic observations identify the transient-slip zone as a Low Velocity Zone with elevated Vp/Vs and Poissonâs ratios, consistent with an overpressured portion of the downgoing slab and high pore-fluid pressures. Fluid-related processesâincluding the presence of interfacial water and possible hydrolytic weakening of contact rocksâare therefore widely invoked to explain LFE generation and episodic slip, although the precise causal mechanisms remain unresolved. Large megathrust earthquakes (M>8) are primarily confined to the shallower locked interface, while LFEs and slow slip occupy the down-dip transition region.
Geodetic detection of slow slip began with a 1999 GPS-documented aseismic event in British Columbia in which the overriding plate moved ~2 cm over a ~50 Ă 300 km area during ~35 days; the inferred moment was equivalent to Mw ~6â7 despite little conventional seismic radiation. Subsequent GPS studies have shown that these reverse, down-dip slow-slip events recur episodically every ~13â16 months, with transient surface displacements at individual stations lasting ~2â4 weeks. Seismic analyses established that the geodetically observed slow-slip episodes are accompanied by tremor composed of numerous LFEsâa coupled phenomenon termed episodic tremor and slip (ETS)âand that LFEs are a ubiquitous feature of the plate interface down-dip of the seismogenic zone.
LFEs in Cascadia are highly sensitive to short-term changes in loading: their occurrence is modulated by tidal forcing, with studies finding phase relationships alternating between peak tidal shear stress and peak shear-stress rate. This sensitivity implies responsiveness to pressure perturbations on the order of a few kilopascals, a magnitude compatible with ocean-tidal loading and sea-levelârelated stress variations, and supports the inference that near-interface fluids play a key role in controlling slow-slip and low-frequency seismicity.
Low-frequency earthquakes (LFEs) in Japan were first recognized in the subduction environment where the Philippine Sea plate descends beneath southwest Japan at the Nankai Trough. Their detection was enabled by the national expansion of seismological networks after the 1995 Kobe earthquake, and early studies interpreted the persistent, low-frequency tremor as a product of dehydration/metamorphic reactions within the subducting lithosphere.
Subsequent mapping placed the principal tremor sources at an average depth near 30 km and distributed them along roughly 600 km of the plate interface, revealing a broad along-strike band of tremor generation. LFEs and the emergent low-frequency tremor are now known to occur widely across Japanese subduction zones and to be tightly coupled to slow slip events (SSEs); in Japan these tremorâSSE episodes commonly recur on the order of six months, a periodicity comparable to tremorâSSE coupling described in Cascadia.
The spatial pattern of Japanese LFEs is concentrated where the Philippine Sea plate subducts (Nankai/Tokai region) rather than beneath the Japan Trench, a contrast attributable to subduction geometry: the Philippine Sea plate enters at a shallower dip than the Pacific plate, creating conditions more favorable for SSEs and associated tremor. Hypocenters of LFEs lie down-dip from the locked seismogenic zone, near the base of the transition region; thermal models for Tokai place the seismogenic zone between roughly 8 and 22 km depth, with LFEs occurring deeper than this rupture-prone layer.
Thermal constraints strongly implicate temperature-dependent processes in LFE generation. In Tokai, LFEs are localized within a narrow thermal window (approximately 450â500 °C), consistent with rheological transitions and fluid-releasing metamorphic reactions that alter fault zone strength and promote slow slip and tremor. Taken together, the depth, thermal regime, and subduction geometry indicate that LFEs in Japan result from coupled mechanical and metamorphic processes concentrated at the deep, down-dip margin of the seismogenic plate interface.
Very low frequency earthquakes
Very low frequency earthquakes (VLFs) are a distinct class of slow seismic events marked by unusually long source durations and energy concentrated at periods much lower than typical lowâfrequency earthquakes (LFEs). Typical VLFs have magnitudes near 3.0â3.5, source durations on the order of ~20 s, and dominant spectral energy in the ~0.02â0.03 Hz band. Seismological analyses indicate VLFs arise from reverse/thrust faulting on or near the plate interfaceâmechanics that closely resemble those inferred for LFEsâthough their longer duration produces the enhanced veryâlowâfrequency content.
VLFs display a pronounced but asymmetric relationship with LFEs: they most often accompany LFE activity, yet many LFEs occur without an associated VLF, suggesting VLFs represent a less common, perhaps more spatially or temporally concentrated expression of the same slowâslip processes. Two principal structural settings host VLFs: shallow events within the offshore accretionary prism and deeper events at the plate interface below the seismogenic zone; this depth dichotomy motivates the shallow/deep classification. Observations from subduction zones such as Cascadia and segments of the Japanese margin (Nankai trough, Ryukyu trench) show that VLFs participate in episodic tremor and slip (ETS), including alongâstrike migration patterns analogous to those of LFEs, implying VLFs are integrated components of largeâscale slowâslip and tremor migration along subduction interfaces.
Slow slip events
Slow slip events (SSEs) are prolonged episodes of shear on subduction-zone plate interfaces in which the overriding and subducting plates accommodate displacement by slow, thrust-directed slip rather than by the rapid rupture characteristic of ordinary earthquakes. Individual SSEs typically unfold over periods of weeks, although some persist much longerâinto months or yearsâso that the phenomenon spans a continuum of durations and slip rates distinct from regular seismic rupture.
When slip episodes recur with observable periodicity and are temporally linked to tectonic tremor, the coupled phenomenon is termed episodic tremor and slip (ETS). Recurrence intervals exhibit systematic along-strike variability: for example, crustal deformation studies indicate a typical return period of roughly 14.5 months in Cascadia (with local variations along the margin) whereas in Shikoku, southwest Japan, SSEs recur at about six-month intervals as revealed by crustal tilt measurements. Long-duration occurrences illustrate the upper end of the spectrumâmost notably the Tokai slow slip event, which persisted from mid-2000 into 2003.
Geodetic techniques and high-sensitivity tiltmeters are effective tools for detecting and quantifying SSEs, resolving both timing and amplitude of slip. The spatial evolution of SSEs commonly involves alongâstrike propagation of a slip front; in Cascadia this migration has been observed at rates of order 5â10 km per day. This migrating slip front provides a physical mechanism for the contemporaneous alongâstrike migration of lowâfrequency earthquakes (LFEs) and tectonic tremor, which are seismic expressions that are closely linked in space and time to SSEs.
Episodic tremor and slip (ETS)
Episodic tremor and slip (ETS) refers to recurring, plateâboundaryârelated episodes of slow, aseismic fault slip accompanied by sustained nonâimpulsive seismic tremor. Unlike conventional earthquakes, which concentrate strain release into seconds of rapid rupture, ETS events discharge strain gradually over days to weeks (and in some cases months to a year), producing prolonged, lowâamplitude seismic signals and measurable deep fault slip. Because ETS is tied to steady plate motion, it often recurs on quasiâregular intervals and can therefore be anticipated more readily than isolated fast ruptures.
Field observations and seismic geodesy indicate that ETS can interact mechanically with shallower, brittle faulting: some slowâslip episodes have temporally preceded or coincided with the initiation of major, damaging shallow earthquakes (notable examples include the 2001 Nisqually and the 1995 Antofagasta sequences), suggesting that deep aseismic slip can modulate stress sufficiently to promote fast rupture. The causal relationship is not unidirectional; large, rapid earthquakes commonly produce postâseismic aseismic deformationâcontinued creep and viscous relaxation in the deeper crust and upper mantleâdemonstrating twoâway feedbacks between instantaneous seismic failure and longerâtimescale, depthâdependent relaxation processes.
Regional case studies illustrate the spectrum of ETS behaviour. For example, beneath Wellington, New Zealand, a yearâlong slowâslip episode recurs roughly every five years (observed in 2003, 2008 and 2013), with each event releasing total moment equivalent to an approximately magnitudeâ7 earthquake while unfolding over months rather than seconds. Such examples emphasize that ETS is a significant component of plateâboundary deformation with important implications for seismic hazard and the mechanics of earthquake triggering.