Earthquake Predictibility, Forcasting and Early Warning Bill Menke October 18, 2005
Summary Are long-term predictions of earthquakes possible? Are short-term (or intermediate term) predictions of earthquakes possible? Can specific earthquakes be forecast? Are a few seconds or minutes of Early Warning useful. Is rapid assessment useful?
Why might long-term predictions be possible ? 1. Most earthquakes are on plate boundaries 2. Plate motions are very constant over long periods of time 3. Faults at plate-boundaries are long term features 4. Long-term fault slip rate of faults are fairly constant 5. Segments of faults seem to rupture time and time again in similar earthquakes 6. Earthquake occurs when loading exceeds strength
Long-term Predictability Most earthquakes are on plate boundaries
Long-term Predictability Plate motions are very constant Here are the current-day plate motions as measured by GPS satellite geodesy
But present-day rates (black) agree very well for averages for the last million years, as determined from geological features (red)
Long-term Predictability Faults at plate-boundaries are long term features San Andreas Fault System Pacific-North America plate motion: 4. 8 cm/year Part of slip accommodated by San Andreas: 2. 9 cm/year
Long-term Predictability Faults at plate-boundaries are long term features 23. 5 Ma Volcanic rocks offset by 315 km, rate of 1. 3 cm/year
Almost no new faults Faults grow slowly A big earthquake on a fault tends to increase the length of the fault The bigger the fault, the bigger the earthquake that can occur
Strategy: map the faults to determine where earthquakes will occur (but look for evidence of recent motion, make sure it’s a recently active fault)
Problem: deeply buried faults, such as blind thrusts (especially if they have few small earthquakes) (example fault that caused 1994 Northridge Earthquake) (But now we know it’s there!)
Long-term Predictability Faults segmentation: characteristic large earthquakes
Example – Parkfield Segment of San Andreas Fault
History of 7 large earthquake that broke the segment 2004
Segmentation in Japan
Long-term Predictability Earthquake occurs when loading exceeds strength Loading rate correlates With plate-tectonic motions loading Maximum load Eq time, years Eq Now: where Are we in the Loading cycle ?
Earthquakes on neighboring faults change loading but can be accounted for Northern Anatolian Fault (strike-slip) in Turkey Portions of fault that recently experience large earthquakes are far from failure But neighboring portions are driven closer to failure
Loading in California since 1812, incl. plate motions and large eqs. Note most small earthquakes (circular symbols) occurred in high-load areas
Results of this kind of analysis Is a prediction of likelihoof of a large earthquake on each Of a region’s fault Assuming: long-term loading rates determined by GPS and/or geological studies and closeness to failure based on when last large event occured
Why might short-term or intermediateterm predictions be possible ? 1. Detectable changes in fault behavior as it approaches failure 2. Examples: 3. Foreshocks – small earthquakes that occur before the big one – short term 4. 5. Seismicity rate changes – increase in rate of moderate earthquakes prior to the big one – intermediate term
Foreshock little one before the Big One • In California, foreshocks occur less than 5 days before about half of the large earthquakes. For these reasons, the California Office of Emergency Services issues an advisory of an increased likelihood of a major earthquake within the next 5 days following moderate-sized earthquakes.
Northern California: Rate and pattern of moderate earthquakes not constant with time
What can you do with a prediction of an earthquake ? Especially if it has low skill
Short term heightened emergency preparedness curtain endangered activities evacuate people Intermediate term redirect preparedness funds re-site future construction
Early Warning - or every second counts -
How long do you have ? city Strong ground motion sensors 10 km fault 50 km distant At 2 km/s shear wave velocity Is 25 seconds Minus 10 seconds to Detect strong motion at a Few stations near fault Is …
15 seconds
But say the damaging effects extend to 100 km … 50 km 100 km < 15 second warning > 15 second warning … There may be a lot of people & structure in the >15 second warning region area
For this to have any hope of working you must plaster the earth with sensors capable of detecting strong ground motion and immediately sending that information to a processing and distribution center
Seismic Intensity Stations in Japan
So little time is available that both the announcement of impending strong shaking and the response must be fully automated
What can you do in 15 seconds ? Shut down delicate or dangerous equipment Have people dive for shelter (? )
Just knowing where the strong shaking occurred can help in formulating an emergency response Strong ground motions after the 1995 Kobe, Japan earthquake
Скачать презентацию Earthquake Predictibility Forcasting and Early Warning Bill Menke
a3b35433ba7baee233b0200a9e72d5db.ppt