Thailand Training Program in Seismology and Tsunami Warnings, May 2006 Forecasting Earthquakes

・ Difference between Predictions and Forecasts ・ Early Efforts in Earthquake Prediction ・ Long-term Forecasting

Earthquake Predictions versus Forecasts Predictions have specific times, locations, and magnitudes for future earthquakes. Forecasts are more long-term estimates of earthquake occurrences with probability information.

Short-term (minutes to weeks) Intermediate (months to 10 years ) Long-term (> 10 years) Forecast Loading Cycle Interseismic Preseismic Coseismic Postseismic

1. Time 2. Location 3. Magnitude What is needed in an earthquake prediction ? Allen, 1996 Frequency of Occurrence of Earthquakes Descriptor Magnitude Average Annually Great8 and higher 1 Major Strong Moderate Light ,000 (estimated) Minor ,000 (estimated) Very Minor ,300,000 (estimated)

1. Time 2. Location 3. Magnitude 4. Indication of confidence (window) 5. Prediction must be presented in accessible form for later evaluation 6. Chance earthquake occurs anyway as a random event What is needed in an earthquake prediction ? Allen, 1996

‘Now, when will earthquake prediction be possible and an efficient Forewarning service available ? …if we start the project presented here we should be able to answer the question with sufficient certainty within ten years.’ The Japanese Blueprint (Tsuboi et al, 1962) ‘Based on an assessment of worldwide observations and findings over the past few years, it is the panel’s unanimous opinion that the development of an effective earthquake prediction capability is an achievable goal. …with appropriate commitment and level of effort, the routine announcement of reliable predictions may be possible within ten years…’ Panel of the US National Research Council (Allen et al., 1976) Optimism in the 1960 ’ s and 1970 ’ s In actuality: Can only forecast earthquakes and mitigate hazard

‘Predicted’ ・ 1973 Blue Mountain Lake ・ 1975 Haicheng, China ・ 1978 Oaxaca, Mexico ・ 1978 Izu, Japan ‘ Not Predicted’ ・ 1976 Tangshan, China (M ,000 est. casualties) ・ 2004 Parkfield, California Successes and Failures

Blue Mountain Lake, NY

‘Dilatancy’

Scholz et al., 1973 Dilatancy hardening

Prediction of the 1975 Haicheng, China Earthquake (M7.3) Prediction based on foreshocks and animal behavior saved many lives

Short-Term Crustal Deformation Precursor (M 8.1 Tonankai Earthquake 1944) Mogi, 1984

Short-Term Electromagnetic Precursor 1989 Loma Prieta Earthquake Fraser-Smith, et al., 1990

1978 Oaxaca, Mexico (Ohtake et al., 1981) Intermediate-Term: Seismicity Patterns ‘Mogi Doughnut’ Dilatancy Hardening

Mogi, 1985 Intermediate-Term: Crustal Deformation 1964 Niigata earthquake

Wakita, 1981 Intermediate-Term: Water and Chemical M 7.0 Izu earthquake 1978

‘Successfully’ Predicted ・ 1973 Blue Mountain Lake ・ 1975 Haicheng, China ・ 1978 Oaxaca, Mexico ・ 1978 Izu, Japan

Parkfield Earthquake Prediction Experiment 95% probability that M5.5 to 6 will occur Based on Recurrence pattern Characteristic waveforms

Bakun and McEvilly, Repeating Characteristic Earthquakes - M5 foreshock 17 minutes before 1934 and 1966 events

Not Predicted ・ 1976 Tangshan, China (M ,000 est. casualties) ・ 2004 Parkfield, California For short-term prediction, there have been more negative results than positive results: Other precursors have not been consistently seen

Why has there been lack of continued success in observing precursors? Animal experiments have failed Original observations were hindsight Instrumentation Geology dependent (including lights and piezoelectric effects) Predictive algorithms have not worked (at least short-term) Earthquakes may be non-characteristic No connection to tides or planetary alignment No psychics have made consistent predictions

Short-term minutes to weeks Intermediate-term months to 10 years Long-term > 10 years Earthquake Forecasting

Recurrence times of earthquakes to forecast future earthquakes Using historic data Using geologic data

Shimazaki and Nakata, 1980 Earthquake Cycle Periodic Time-predictable Slip-predictable

History of Nankai Earthquakes

PALEOSEISMOLOGY Trenching faults to find geological evidence of past earthquakes Pallet Creek site on the San Andreas fault

Periodic Time-predictable Slip-predictable Shimazaki and Nakata, 1980 Earthquake Cycle

100 years Conditional Probability Probability

Well defined recurrence interval Wide range of recurrence intervals (Large variability) (Small variability) Variability in Repeating Earthquakes

Probabilistic Earthquake Forecasting for California Hector Mines Landers 1992 Parkfield 2004 Loma Prieta 1989

USGS National Hazard Maps

Earthquake Hazards Ground Shaking Surface Faulting Landslides and Liquefaction Avalanches and soil (mud flows)

NEIC Catalog Seismicity (M>4.5) What about Thailand? ・ Low Seismicity ・ No Historical Earthquakes ・ Active Faults (?) ・ Probability is low for future earthquakes But earthquakes can happen in low probability regions and offshore

Will we be able to predict earthquakes in the future ? Long-term: Probably Intermediate term: Maybe Short-term: Maybe Future Outlook

Don ’ t forget that the precursor to a tsunami is the earthquake.

Parkfield Earthquake Prediction Experiment

Loading or deformation cycle –Four phases Interseismic Preseismic Coseismic Postseismic

For short-term precursors, there are currently more ‘negative’ results than ‘positive’results. Kanamori et al., 1996 Johnston and Linde, 2002

Distance Measurements on the San Andreas fault Matsushiro strainmeter Earthquake Prediction Research in the 1970 ’ s

Laser Distance Measurements

M 6.8 Nisqually 2001 Ground Shaking

Surface faulting Bei-Fung Bridge near Fung-Yan city, 1999 Chi-Chi, Taiwan earthquake

1964 Niigata Earthquake Liquefaction