‘Triggering’ = a perturbation in the loading deformation that leads to a change in the probability of failure. How do we know it happens? ‘Triggering’ = a perturbation in the loading deformation that leads to a change in the probability of failure. How do we know it happens? Measure or infer a loading perturbation, & observe a change in seismicity rate (fault population or single fault recurrence), possibly its spatial variation too. The ‘Reference’ State Central California Ambient Seismicity The Perturbation Coyote Lake Mainshock & Ambient Seismicity The Perturbation & Response Coyote Lake Mainshock & Aftershocks Dynamic loads: • Seismic waves (oscillatory, transient) Dynamic loads: • Seismic waves (oscillatory, transient) • Aseismic slip (not oscillatory, may be permanent) Dynamic loads: • Seismic waves (oscillatory, transient) • Aseismic slip (not oscillatory, permanent) • Solid earth tides and ocean loading (oscillatory, ongoing) Dynamic loads: • Seismic waves (oscillatory, transient) • Aseismic slip (not oscillatory, permanent) • Tides (oscillatory, ongoing) • Surface/shallow: snow and ice, reservoir filling/draining, mining, ground water, fluid injection or withdrawal (localized) • Magma movement (temperature, pressure, and chemical changes too) What’s unique about dynamic loads? What’s unique about dynamic loads? They’re transient! Static Load Change Dt shear stress failure threshold time Static Load Change Dt Dt shear stress failure threshold time Dynamic Triggering shear stress failure threshold time Dynamic Triggering Dt shear stress failure threshold time What’s unique about dynamic loads? They’re transient; the failure conditions must change! Dt shear stress failure threshold time What’s unique about dynamic loads? They’re transient; the failure conditions must change! They’re oscillatory, but they only enhance failure probability (ASSUMPTION); no stress shadows. What’s unique about dynamic loads? They’re transient; the failure conditions must change! They only enhance failure probability (ASSUMPTION); no stress shadows. Slower distance decay than static stress changes. Dynamic Triggering Observations (by load type) Seismic waves (transient, oscillatory) Remote (many source dimensions) Near-field (few source dimensions) Distance-independent Quasi-seismic responses Laboratory Aseismic slip (slow, permanent) Tides (oscillatory, ongoing) Surface/Shallow: snow and ice, reservoir filling/draining, mining, ground water, fluid injection or withdrawal (localized) Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Seismicity rate increases following large earthquakes. Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Seismicity rate increases following large earthquakes. Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Seismicity rate increases following large earthquakes. Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Seismicity rate increases following large earthquakes. Missing? rate increases. Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Correlation of spatial rate increase with directivity. Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Correlation of spatial rate increase with directivity. Correlation of (no) rate change with co-located seismic & aseismic events. Pollitz & Johnston, 2007 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Correlation of spatial rate increase with directivity. Correlation of (no) rate change with co-located seismic & aseismic events. Pollitz & Johnston, 2007 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Correlation of spatial rate increase with directivity. Correlation of (no) rate change with co-located seismic & aseismic events. Early excess of aftershocks. Rate increases in stress shadows. Chi-Chi earthquake shadows start with 3-month rate increases. Ma et al., 2005 1998 1999 2000 2001 2002 1998 1999 2000 2001 2002 “Observed seismicity rate decreases in the Santa Monica Bay and along parts of the San Andreas fault are correlated with the calculated stress decrease.” Stein, 1999 “Observed seismicity rate decreases in the Santa Monica Bay and along parts of the San Andreas fault are correlated with the calculated stress decrease.” Stein, 1999 Time history of seismicity from Santa Monica Bay (Marsan, 2003). “The [Stein, 1999] interpretation is made difficult by the fact that the transient activity modulation by the 1989 M5 Malibu earthquake was still ongoing….the quiescence observed after 1994 can be tracked back several months before Northridge, the latter main shock actually triggering seismicity in the region at the very short (i.e. days) timescale. Marsan, 2003 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Measured Linear Aftershock Densities Felzer & Brodsky, 2006 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Modeled Linear Aftershock Densities (r,D) C10 Mmin N(r, D) [ number of aftershocks at distance r 2 constant! Dr Dr ]P(r, D) number of probability of potential nucleation nucleation sites per unit distance Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view D ‘Linear density’ = number of aftershocks within a volume defined by surface S everywhere at distance r and width Dr Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Modeled Linear Aftershock Densities N(r, D) (r, D) [ Dr ]P(r, D) N(r, D) [ F(r)ds] Dr S [4 A{1 ( D ) ( 1 )( D )2}r (d 1) ] Dr r 2 r Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Modeled Linear Aftershock Densities N(r, D) (r, D) [ P(r, D) D P(r) 1 2 [ D r ] 2 [ r ] 2 2 or Dr ]P(r, D) or D 1 2 [ D r] 2 [ r] 2 r r D Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Modeled Linear Aftershock Densities N(r, D) (r, D) [ P(r, D) D P(r) 1 2 [ D r ] 2 [ r ] 2 2 or Dr ]P(r, D) or D 1 2 [ D r] 2 [ r] 2 r r D Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Are dynamic deformations consistent with these probabilities? Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Are dynamic deformations consistent with these probabilities? Peak Velocities vs r, M5.5-7.0 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Are dynamic deformations consistent with these probabilities? Peak Velocities vs r, M5.5-7.0 perhaps! Peak Velocities vs r/D, M5.5-7.0 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses ‘Low-frequency’ events Sumatra surface waves in Japan High-passed Sumatra surface waves in Japan Correlation with Rayleigh waves - Dilatation & Fluids Miyazawa & Mori, 2006 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses ‘Low-frequency’ events Sumatra surface waves in Japan Denali surface waves in Japan, Correlation with Love waves - Shear Load! High-passed Sumatra surface waves in Japan Correlation with Rayleigh waves - Dilatation & Fluids Miyazawa & Mori, 2006 Rubinstein et al., 2007 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses ‘Low-frequency’ events Creep and tilt Response to Hector Mine waves on Imperial Fault (260 km) H H Glowacka et al., 2002 Dynamic Triggering Observations (by loading type) Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory Granular surface quasi-static experiments. “Our results predict that a transient dynamic normal load during creep can strengthen a fault…gouge particles become compacted into a lower energy configuration.” Richardson and Marone, 1999 Dynamic Triggering Observations delayed failure Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory delayed failure Granite surface stick-slip experiments. Sobolev et al., 1996 Dynamic Triggering Observations delayed failure Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory Vibration Clock-advances Failure delayed failure Granite surface, shear vibration, stick-slip experiments. Sobolev et al., 1996 Dynamic Triggering Observations Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory Granular surface, acoustic vibration, stick-slip experiments. Dynamic Triggering Observations Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory Granular surface, acoustic vibration, stick-slip experiments. triggered ‘new’ seismic events triggered ‘new’ seismic events clock-delayed failure Dynamic Triggering Observations Seismic waves Remote (many source dimensions) Near-field (few source dimensions) Distance-independent view Quasi-seismic responses Laboratory Granular surface, acoustic vibration, stick-slip experiments. triggered ‘new’ seismic events memory clock-delayed failure Dynamic Triggering Observations (by loading type) Seismic waves Aseismic slip Earthquakes Number of earthquakes & displacement Hawaii Slow Slip & Earthquakes Dynamic Triggering Observations (by loading type) Seismic waves Aseismic slip Earthquakes Tremor Geodetic Displacement (mm east) Cascadia Slow Slip & Tremor Dragert et al., 2002 General features: •apparent more commonly in areas of •geothermal & Quaternary to recent volcanism, •extensional regimes, •high strain rates, •seismic strains required ~mstrains, •sometimes instantaneous but also delayed. Models Coulomb-Navier failure: no delays Frictional: traditional clock-advance models can’t explain long delays, require high (near lithostatic) pressures or critical conditions, changing frictional properties or stability regime. Subcritical crack growth: same behavior as rate-state friction. Dynamic nonlinear softening. Fluid and pore pressure mechanisms: decrease effective normal stress, local, fluid-driven deformation disruption of clogged fractures and hydraulic fracturing bubbles rectified diffusion (volatiles selectively pumped into bubbles during the dilatation) advective overpressure (rising of loosened bubbles within magma body ) Models Coulomb-Navier failure: no delays Frictional: traditional clock-advance models can’t explain long delays, require high (near lithostatic) pressures or critical conditions, changing frictional properties or stability regime. Subcritical crack growth: same behavior as rate-state friction. Dynamic nonlinear softening. Fluid and pore pressure mechanisms: decrease effective normal stress, local, fluid-driven deformation disruption of clogged fractures and hydraulic fracturing bubbles rectified diffusion (volatiles selectively pumped into bubbles during the dilatation) advective overpressure (rising of loosened bubbles within magma body ) Dynamically reduced contact area (i.e. critical slip distance) Power-law distribution of contact areas. Parsons, 2005 Dynamically reduced contact area (i.e. critical slip distance) Power-law distribution of contact areas. Number of ‘events’ vs clock-advance for 10% reduction in critical slip distance. Parsons, 2005 Dynamically reduced contact area (i.e. critical slip distance) Power-law distribution of contact areas. Number of ‘events’ vs clock-advance for 10% reduction in critical slip distance. Perturbed failure rate. Parsons, 2005 Models Coulomb-Navier failure: no delays Frictional: traditional clock-advance models can’t explain long delays, require high (near lithostatic) pressures or critical conditions, changing frictional properties or stability regime. Subcritical crack growth: same behavior as rate-state friction. Dynamic nonlinear softening. Fluid and pore pressure mechanisms: decrease effective normal stress, local, fluid-driven deformation disruption of clogged fractures and hydraulic fracturing bubbles rectified diffusion (volatiles selectively pumped into bubbles during the dilatation) advective overpressure (rising of loosened bubbles within magma body ) Models Coulomb-Navier failure: no delays Frictional: traditional clock-advance models can’t explain long delays, require high (near lithostatic) pressures or critical conditions, changing frictional properties or stability regime. Subcritical crack growth: same behavior as rate-state friction. Dynamic nonlinear softening. Fluid and pore pressure mechanisms: decrease effective normal stress, local, fluid-driven deformation, disruption of clogged fractures and hydraulic fracturing, bubbles rectified diffusion (volatiles selectively pumped into bubbles during the dilatation) advective overpressure (rising of loosened bubbles within magma body). Elastic moduli decrease (soften) with increasing dynamic load amplitude -> weakening mechanism? Pulse Experiments, Glass Beads Elastic moduli decrease (soften) with increasing dynamic load amplitude -> weakening mechanism? sinusoid amplitude (strain) Sinusoid Experiments, Rocks Pulse Experiments, Glass Beads Models Coulomb-Navier failure: no delays Frictional: traditional clock-advance models can’t explain long delays, require high (near lithostatic) pressures or critical conditions, changing frictional properties or stability regime. Subcritical crack growth: same behavior as rate-state friction. Dynamic nonlinear softening. Fluid and pore pressure mechanisms: decrease effective normal stress, local, fluid-driven deformation, disruption of clogged fractures and hydraulic fracturing, bubbles rectified diffusion (volatiles pumped into bubbles during the dilatation), advective overpressure (rising of loosened bubbles within magma body), liquefaction. -Outstanding QuestionsIs our sampling biased (e.g., best monitoring in high strain rate and/or geothermal areas)? -Outstanding QuestionsIs our sampling biased (e.g., best monitoring in high strain rate and/or geothermal areas)? How important are local conditions; are multiple mechanisms at work? -Outstanding QuestionsIs our sampling biased (e.g., best monitoring in high strain rate and/or geothermal areas)? How important are local conditions; are multiple mechanisms at work? What are the important characteristics of the dynamic field (frequency/rate, duration, max. value)? Strain Rate (acceleration) Strain (velocity) Displacement Velocity Strengthening, Slip Weakening Friction Theoretical Frequency Sensitivity Non-Linear, Slip Weakening Friction Dynamically Induced Pore Pressure Change -Outstanding QuestionsIs our sampling biased (e.g., best monitoring in high strain rate and/or geothermal areas)? How important are local conditions; are multiple mechanisms at work? What are the important characteristics of the dynamic field (frequency/rate, duration, max. value)? How does delayed failure happen? Thanks! Comments?