Modeling swarms: A path toward determining short- term probabilities Andrea Llenos USGS Menlo Park Workshop on Time-Dependent Models in UCERF3 8 June 2011.

Slides:



Advertisements
Similar presentations
Ilya Zaliapin Department of Mathematics and Statistics University of Nevada, Reno SAMSI workshop Dynamics of Seismicity Thursday, October 10, 2013 Yehuda.
Advertisements

The rate of aftershock density decay with distance Karen Felzer 1 and Emily Brodsky 2 1. U.S. Geological Survey 2. University of California, Los Angeles.
Review of Catalogs and Rate Determination in UCERF2 and Plans for UCERF3 Andy Michael.
State Estimation and Kalman Filtering CS B659 Spring 2013 Kris Hauser.
Contributions of Prof. Tokuji Utsu to Statistical Seismology and Recent Developments Ogata, Yosihiko The Institute of Statistical Mathematics , Tokyo and.
The Community Geodetic Model (CGM): What is it and how does it relate to studies of lithospheric rheology? Jessica Murray, David Sandwell, and Rowena Lohman.
Chambery, 17 octobre, 2008 Seismic monitoring of volcano processes Paola Traversa, Jean-Robert Grasso.
Detecting Aseismic Fault Slip and Magmatic Intrusion From Seismicity Data A. L. Llenos 1, J. J. McGuire 2 1 MIT/WHOI Joint Program in Oceanography 2 Woods.
Stress- and State-Dependence of Earthquake Occurrence: Tutorial 2 Jim Dieterich University of California, Riverside.
Earthquake swarms Ge 277, 2012 Thomas Ader. Outline Presentation of swarms Analysis of the 2000 swarm in Vogtland/NW Bohemia: Indications for a successively.
16/9/2011UCERF3 / EQ Simulators Workshop RSQSim Jim Dieterich Keith Richards-Dinger UC Riverside Funding: USGS NEHRP SCEC.
Evidence from the A.D Izu islands earthquake swarm that stressing rate governs seismicity By Toda, S., Stein, R.S. and Sagiya, T. In Nature(2002),
Numerical simulation of seismic cycles at a subduction zone with a laboratory-derived friction law Naoyuki Kato (1), Kazuro Hirahara (2), and Mikio Iizuka.
The long precursory phase of many large interplate earthquakes Michel Bouchon, Hayrullah Karabulut, Virginie Durand, David Marsan With: Mustafa Aktar,
Importance of OEF to the USGS Earthquake Hazards Program Michael L. Blanpied USGS Earthquake Hazards Program For Powell Center March.
Yan Y. Kagan Dept. Earth and Space Sciences, UCLA, Los Angeles, CA , EARTHQUAKE.
Yan Y. Kagan Dept. Earth and Space Sciences, UCLA, Los Angeles, CA , EARTHQUAKE PREDICTABILITY.
Yan Y. Kagan Dept. Earth and Space Sciences, UCLA, Los Angeles, CA , Full.
Epidemic Type Earthquake Sequence (ETES) model  Seismicity rate = "background" + "aftershocks":  Magnitude distribution: uniform G.R. law with b=1 (Fig.
Chapter 5: Calculating Earthquake Probabilities for the SFBR Mei Xue EQW March 16.
Omori law Students present their assignments The modified Omori law Omori law for foreshocks Aftershocks of aftershocks Physical aspects of temporal clustering.
CONTRASTING SEISMIC RATES BETWEEN THE NEW MADRID AND WABASH VALLEY SEISMIC ZONES: STRESS TRANSFER OR AFTERSHOCKS? Miguel Merino, Seth Stein & Emile Okal.
Yan Y. Kagan, David D. Jackson Dept. Earth and Space Sciences, UCLA, Los Angeles, CA ,
Network Strain Filter and its applications on GPS data Matt Wei, Jeff McGuire WHOI September 10, 2011.
The Empirical Model Karen Felzer USGS Pasadena. A low modern/historical seismicity rate has long been recognized in the San Francisco Bay Area Stein 1999.
Omori law The modified Omori law Omori law for foreshocks Aftershocks of aftershocks Physical aspects of temporal clustering.
 ss=  * +(a-b) ln(V/V * ) a-b > 0 stable sliding a-b < 0 slip is potentially unstable Correspond to T~300 °C For Quartzo- Feldspathic rocks Stationary.
Brainstorm: How to assess an Earthquake: Stroked off B.C. coast? Rapid Earthquake Risk Assessment Source Parameters USGS World Shake Maps USGS Shake Aftershocks.
If we build an ETAS model based primarily on information from smaller earthquakes, will it work for forecasting the larger (M≥6.5) potentially damaging.
Statistics of Seismicity and Uncertainties in Earthquake Catalogs Forecasting Based on Data Assimilation Maximilian J. Werner Swiss Seismological Service.
The use of earthquake rate changes as a stress meter at Kilauea volcano Nature, V. 408, 2000 By J. Dietrich, V. Cayol, and P. Okubo Presented by Celia.
The Evolution of Regional Seismicity Between Large Earthquakes David D. Bowman California State University, Fullerton Geoffrey C. P. King Institut de Physique.
Analysis of complex seismicity pattern generated by fluid diffusion and aftershock triggering Sebastian Hainzl Toni Kraft System Statsei4.
Searching for Long Duration Aftershocks in Continental Interiors Miguel Merino, Seth Stein Northwestern University.
Agnès Helmstetter 1 and Bruce Shaw 2 1,2 LDEO, Columbia University 1 now at LGIT, Univ Grenoble, France Relation between stress heterogeneity and aftershock.
Toward urgent forecasting of aftershock hazard: Simultaneous estimation of b-value of the Gutenberg-Richter ’ s law of the magnitude frequency and changing.
A functional form for the spatial distribution of aftershocks Karen Felzer USGS Pasadena.
Application of the RI model to forecasting future large earthquakes in Japan Kazu Z. Nanjo (ERI, Univ. of Tokyo) International symposium “Toward constructing.
Southern California Earthquake Center Triggering Models vs. Smoothed Seismicity PG = 1.35/eqk PG = 10/eqk Information gain per earthquake Reference forecast.
Forward-Scan Sonar Tomographic Reconstruction PHD Filter Multiple Target Tracking Bayesian Multiple Target Tracking in Forward Scan Sonar.
Borehole Strainmeters: Instruments for Measuring Aseismic Deformation in Subduction Zones Evelyn Roeloffs U.S. Geological Survey, Vancouver, WA.
CRUSTAL DEFORMATION BREAKOUT Key Scientific Questions  How do magmatic systems evolve and how can we improve eruption forecasting?  How can we quantify.
Karen Felzer & Emily Brodsky Testing Stress Shadows.
Coulomb Stress Changes and the Triggering of Earthquakes
Some Properties of “aftershocks” Some properties of Aftershocks Dave Jackson UCLA Oct 25, 2011 UC BERKELEY.
Forecasting Magnitude from Fault Geometry Bill Ellsworth, USGS Menlo Park, CA.
2. MOTIVATION The distribution of interevent times of aftershocks suggests that they obey a Self Organized process (Bak et al, 2002). Numerical models.
Relative quiescence reported before the occurrence of the largest aftershock (M5.8) with likely scenarios of precursory slips considered for the stress-shadow.
1 Producing Omori’s law from stochastic stress transfer and release Mark Bebbington, Massey University (joint work with Kostya Borovkov, University of.
David Schmidt Ray Weldon Reed Burgette Randy Krogstad Haiying Gao
A proposed triggering/clustering model for the current WGCEP Karen Felzer USGS, Pasadena Seismogram from Peng et al., in press.
California Earthquake Rupture Model Satisfying Accepted Scaling Laws (SCEC 2010, 1-129) David Jackson, Yan Kagan and Qi Wang Department of Earth and Space.
2002/05/07ACES Workshop Spatio-temporal slip distribution around the Japanese Islands deduced from Geodetic Data Takeshi Sagiya Geographical Survey Institute.
GNS Science Testing by hybridization – a practical approach to testing earthquake forecasting models David Rhoades, Annemarie Christophersen & Matt Gerstenberger.
A new prior distribution of a Bayesian forecast model for small repeating earthquakes in the subduction zone along the Japan Trench Masami Okada (MRI,
Zhaoxia Fu, Yan Han Measurement Volume 45, Issue 4, May 2012, Pages 650–655 Reporter: Jing-Siang, Chen.
Jiancang Zhuang Inst. Statist. Math. Detecting spatial variations of earthquake clustering parameters via maximum weighted likelihood.
Abstract The space-time epidemic-type aftershock sequence (ETAS) model is a stochastic process in which seismicity is classified into background and clustering.
Future Directions and Capabilities of Simulators Jim Dieterich
Comments on physical simulator models
Seismicity shadows: observations and modelling
Some issues/limitations with current UCERF approach
Yan Y. Kagan Dept. Earth, Planetary, and Space Sciences, UCLA, Los Angeles, CA , NEGATIVE.
Creager, Wech, Vidale, Melbourne
R. Console, M. Murru, F. Catalli
K. Maeda and F. Hirose (MRI)
Creager, Wech, Vidale, Melbourne
Creager, Wech, Vidale, Melbourne
by Naoki Uchida, Takeshi Iinuma, Robert M
Nome Sobrenome. Time time time time time time..
Presentation transcript:

Modeling swarms: A path toward determining short- term probabilities Andrea Llenos USGS Menlo Park Workshop on Time-Dependent Models in UCERF3 8 June 2011

Outline Motivation: Why are swarms important for UCERF? Where things stand now – Characteristics of swarms – Detecting swarms (retrospectively) What needs to be done – Detecting swarms (prospectively) – Implementation As ETAS add-on? As a data assimilation application? Background seismicity rate Observed seismicity rate Aftershock sequences

Time-dependent background rates are needed to account for rate changes due to external (aseismic) processes Daniel et al. (2011) Ubaye swarm (fluid-flow) Lombardi et al. (2006) 2000 Izu Islands swarm (magma/fluids) 2000 Vogtland/Bohemia swarm (fluids) Hainzl and Ogata (2005)

Salton Trough Time-dependent background rate matches observed seismicity better than stationary ETAS model Llenos and McGuire (2011) Transformed Time

Characteristics of swarms Increase in seismicity rate above background without clear mainshock Don’t follow empirical aftershock laws – Bath’s Law – Omori’s Law These characteristics make them appear anomalous to ETAS Holtkamp and Brudzinski (2011)

Detecting swarms in an earthquake catalog Swarms associated with aseismic transients 2005 Obsidian Buttes, CA ( , SCEDC) 2005 Kilauea, HI ( , ANSS) 2002, 2007 Boso, Japan ( , JMA) Ozawa et al. (2007) Lohman & McGuire (2007) Wolfe et al. (2007) Slow slip events on the subduction plate interface off of Boso, Japan observed by cGPS, tiltmeter Shallow aseismic slip on a strike-slip fault in southern CA observed by InSAR and GPS Slow slip events on southern flank of Kilauea volcano in HI observed by GPS

Data analysis: ETAS model optimization Optimize ETAS model to fit catalog prior to swarm and extrapolate fit through remainder of catalog Calculate transformed times (~ ETAS predicted number of events in a time interval) – Cumulative number of events vs. transformed time should be linear if seismicity behaving as a point process – Positive deviations occur when more seismicity is being triggered in a time interval than ETAS can explain Swarms associated with aseismic transients 2005 Obsidian Buttes, CA ( , SCEDC) 2005 Kilauea, HI ( , ANSS) 2002, 2007 Boso, Japan ( , JMA) 2005 Kilauea

Swarms appear as anomalies relative to ETAS 2002, 2007 Boso, Japan 2005 Obsidian Buttes 2005 Kilauea SwarmPre-swarm MLE (K, , , p, c) Swarm MLE (K, , , p, c) 2002 Boso 0.13, 0.022, 0.56, 1.11, , 2.09, 0.09, 1.0, Kilauea 0.28, 0.16, 1.24, 1.21, , 0.89, 0.61, 0.92, Obs Buttes 0.61, 0.031, 0.88, 1.1, , 225, 1.05, 1.0, Boso 0.20, 0.013, 0.55, 0.88, , 2.4, 1.37, 1.0,

A path toward determining short-term probabilities Build off of ETAS-based forecasts – Detect that a swarm is occurring Has been done retrospectively Prospectively? – During the swarm Re-estimate the background rate (and other parameters?) Re-calculate short-term probabilities How often? 1x? 2x? Every 5 days? 10 days? – Identify when the swarm is over Return to pre-swarm background rate? More sophisticated approaches (e.g., data assimilation)?

Data Assimilation Algorithms Combines dynamic model with noisy data (e.g. seismicity rates) to estimate the temporal evolution of underlying physical variables (states) Examples: Kalman filters, particle filters Applications in navigation, tracking, hydrology Welch & Bishop (2001)

Data Assimilation Example State-space model based on rate-state equations States: stressing rate, rate-state state variable  Algorithm: Extended Kalman Filter Approach: Optimize ETAS for the catalog, subtract ETAS predicted aftershock rate to obtain time-dependent background rate, use data assimilation algorithm to estimate stressing rate and detect transients that trigger swarms Llenos and McGuire (2011)

A path toward determining short-term probabilities Build off of ETAS-based forecasts – Detect that a swarm is occurring Has been done retrospectively Prospectively? – During the swarm Re-estimate the background rate (and other parameters?) Re-calculate short-term probabilities How often? 1x? 2x? Every 5 days? 10 days? – Identify when the swarm is over Return to pre-swarm background rate? More sophisticated approaches (e.g., data assimilation)?

Outline Why are swarms important for UCERF? – Need time-dependent background rate (mu) to model earthquake rates observed in catalogs accurately Salton Trough Ubaye France Campei Flagrei Vogtland Bohemia – Swarms prevalent in Salton Trough, volcanic regions like Long Valley, places where M>6 events have occurred Characteristics of swarms – Don’t fit empirical models of aftershock clustering, appear anomalous ETAS parameters change during swarms (primarily stationary background rate) How to implement this to calculate short-term probabilities? – Where we are now Detection (retrospective) How they affect ETAS parameters – Outstanding issues that need to be addressed – Data assimilation?