The 2016 Bombay Beach Swarm Triggering Probabilities for Larger Earthquakes (on the San Andreas and Elsewhere)
Morgan Page USGS Pasadena The Basics of Earthquake Forecasting The Statistical Seismologist’s Approach
M 104 ≥ Gutenberg-Richter Magnitude Scaling
102 Earthquakes in California, 1984-2011
b=1
100
-2
Number of earthquakes per year 10 345678 Magnitude (M) The Basics of Earthquake Forecasting The Statistical Seismologist’s Approach
M 104 ≥ Gutenberg-Richter Magnitude Scaling
102 Earthquakes in California, 1984-2011
b=1 103 100 Omori Decay of 102
-2
Number of earthquakes per year 10 345678 Aftershock 101 Magnitude (M)
Rate rate aftershock Daily 100
10-1 10-2 10-1 100 101 Time since Mainshock (days) The Basics of Earthquake Forecasting The Statistical Seismologist’s Approach
M 104 ≥ Gutenberg-Richter Magnitude Scaling
102 Earthquakes in California, 1984-2011
b=1 103 100 Omori Decay of 102
-2
Number of earthquakes per year 10 345678 Aftershock 101 Magnitude (M)
Rate rate aftershock Daily 100
10-1 10-2 10-1 100 101 Big earthquakes trigger more aftershocks Time since Mainshock (days) The Basics of Earthquake Forecasting The Statistical Seismologist’s Approach
M 104 ≥ Gutenberg-Richter Magnitude Scaling
102 Earthquakes in California, 1984-2011
b=1 103 100 Omori Decay of 102
-2
Number of earthquakes per year 10 345678 Aftershock 101 Magnitude (M)
Rate rate aftershock Daily 100
10-1 10-2 10-1 100 101 Big earthquakes trigger more aftershocks Time since Mainshock (days)
Aftershock rates decay with distance from the mainshock The Basics of Earthquake Forecasting The Statistical Seismologist’s Approach
M 104 ≥ Gutenberg-Richter Magnitude Scaling
102 Earthquakes in California, 1984-2011
b=1 103 100 Omori Decay of 102
-2
Number of earthquakes per year 10 345678 Aftershock 101 Magnitude (M)
Rate rate aftershock Daily 100
10-1 10-2 10-1 100 101 Big earthquakes trigger more aftershocks Time since Mainshock (days)
Aftershock rates decay with distance from the mainshock
These scaling laws are used in short-term forecasting models like ETAS (Ogata, 1988) and STEP (Gerstenberger et al., 2005) ETAS Predictions
The foreshock rate and the aftershock rate follow the same trend
Given the aftershock rate, you can predict the foreshock rate, suggesting they represent the same process
California data
Felzer et al. (2004) ETAS Predictions
No significant correlation between number of foreshocks and mainshock size
size of mainshocks
dashed lines - aftershock rates for different mainshock sizes
solid lines - foreshock rates for different mainshock sizes Southern California data
Helmstetter and Sornette (2003) Inverse Omori Acceleration
ETAS models can match the acceleration seen in Bouchon et al. (2014) dataset
A) Normalized 150-Day Stack B) Normalized 5-Day Stack 1 1 ETAS Mean & 95% Confidence ETAS Mean & 95% Confidence 0.9 0.9
0.8 Mainshock Data 0.8 Mainshock Data
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1 Cumulative number of earthquakes (normalized) Cumulative number of earthquakes (normalized) 150 100 50 0 5 4 3 2 1 0 Time before mainshocks (days) Time before mainshocks (days)
Felzer, Page and Michael (2015) Foreshocks are not predictive of mainshock size
Moment rate functions all start out the same, which suggests the earthquake doesn’t “know” its final size What do changing earthquake probabilities look like in time?
M ≥ 8 M ≥ 7 M ≥ 6 M ≥6 Event times
M ≥ 2.5 rate = proxy for large event probability 104
103 2.5 rate ≥
M
102
Monthly ETAS Rate Average Rate
101 0 5 10 15 20 2 5 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Time (years) Field and Milner (2018)
Even though foreshocks are not predictive of earthquake size, foreshock/ aftershock statistics can give orders of magnitude changes in the probabilities for future earthquakes of all sizes. The 2016 Bombay Beach Swarm
McBride, Llenos, Page, and van der Elst (submitted) Major forecast uncertainties
What is the appropriate magnitude How long will increased rate of distribution for the Southern San earthquakes last? Andreas? Major forecast uncertainties
Swarm duration in Salton trough are well-fit by a Poisson model
1 1 (a) 0.8 0.8
0.6 0.6 ending today ending today / / 0.4 0.4
Swarm catalog Swarm catalog 0.2 Poisson model 0.2 Poisson model Swarm catalog Swarm catalog Poisson model Poisson model 0 0 Probability over Probability over 051015202530 05101520253035 Duration (days) Duration (days)
15% chance of terminating each day Average length of 7 days
How long will increased rate of earthquakes last? Llenos and van der Elst, 2019 Major forecast uncertainties
Schwartz and Coppersmith (1984) What is the appropriate magnitude 20 km from South-Central SAF distribution for the Southern San (Aftershocks of 1952 Kern country and 1971 San Fernando earthquakes removed) Andreas? Major forecast uncertainties
Page and Felzer (2015) Schwartz and Coppersmith (1984) What is the appropriate magnitude 20 km from South-Central SAF distribution for the Southern San (Aftershocks of 1952 Kern country and 1971 San Fernando earthquakes removed) Andreas? Major forecast uncertainties effect of proximity to San Andreas distance from CFM fault
In the instrumental catalog, we do not see that earthquakes near major faults are larger.
We also do not see that earthquakes near faults produce more aftershocks or are more likely to be a foreshock to a larger event.
Page and van der Elst, 2018 Major forecast uncertainties lead to factor of 30 difference in hazard estimate
0.03% to 1% chance of M≥7 earthquake next 7 days
Probabilities decay very quickly UCERF3 Forecast following Bombay Swarm What faults are likely to be triggered?
Aftershock forecast following Bombay Beach swarm Average of 100,000 simulations UCERF3 Forecast following Bombay Swarm How are probabilities elevated compared to long-term rate?
San Andreas (Coachella section) Rupture Probabilities r r k rs rs y day
1 wee 10 y 1 yea 1 hou 100 1 1 0-1 1 month
1 0-2 bability
1 0-3
1 0-4
1 0-5 M≥6.7 Participation Pro UCERF3-TI UCERF3-TD UCERF3-ETAS 1 0-6
-3 -2 -1 0 1 2 1 0 1 0 1 0 1 0 1 0 1 0 Forecast Timespan (years) Conclusions
During seismic swarms, the chance of a large earthquake can change San Andreas (Coachella section) Rupture Probabilities r r k rs rs y
day by orders of magnitude
1 wee 10 y 1 yea 1 hou 100 1 1 0-1 1 month
1 0-2 bability These elevated -3 1 0 probabilities decay
1 0-4 quickly
1 0-5 M≥6.7 Participation Pro UCERF3-TI UCERF3-TD UCERF3-ETAS 1 0-6 Proximity to major faults is -3 -2 -1 0 1 2 1 0 1 0 1 0 1 0 1 0 1 0 Forecast Timespan (years) a concern (but it is not proven that proximity to major faults elevates foreshock probability) Conclusions
During seismic swarms, the chance of a large earthquake can change by orders of magnitude
These elevated probabilities decay quickly
Proximity to major faults is a concern (but it is not proven that Questions? proximity to major faults elevates foreshock probability)