Early Warning System

We have been reminded repeatedly over the past few years about the potentially catastrophic impact to life and property from large earthquakes. The Magnitude-7 earthquake in Haiti in 2010 became the fifth most deadly earthquake on record, killing more than 200,000 people and resulting in $8 billion in direct damages. The economic impact from the Magnitude-8.8 earthquake that struck Chile in 2010 is estimated at as much as $30 billion. The Magnitude-9 earthquake and tsunami that struck Japan in 2011 caused at least $200 billion in estimated losses. As a result of this potential regional and national impact, research into earthquake prediction has existed on some level for almost 100 years.

Over the past 40 years, however, the increasing density of seismic networks in active tectonic regions around the world has resulted in the availability of a wealth of seismicity data at unprecedented small magnitude levels. It has long been recognized that temporal and spatial clustering is evident in seismicity data, but past research associated with these patterns tended to focus on a relatively small fraction of the events, primarily at the larger magnitudes. The availability of these new, larger data sets and the computational advancements that facilitate simulations, rigourous statistical tests and innovative filtering techniques provided a new impetus for earthquake forecasting and earthquake early warning (EEW) systems.

Earthquake Forecasting

Time-dependent earthquake forecasting

Over the years, a suite of empirical data has been the subject of studies into precursory phenomena, including seismicity patterns, tilt and strain precursors, electromagnetic signals, hydrologic phenomena and chemical emissions. None have been successful in providing reliable and repeatable earthquake predictions. However, the first prospective forecast using small-magnitude seismic data was published in 2002. This initial publication was followed by a renewed interest in seismicity-based methodologies and prompted a renewed effort to better define and test these techniques. The physical motivation is that small-magnitude seismicity acts as a sensor for the underlying stress changes in the earthquake fault system. Changes in that seismicity provide an indicator of earthquake locations in either the short-term (days to weeks) or intermediate-term (years to decades). Landmark initiatives in earthquake forecasting validation and testing followed, including the working group on Regional Earthquake Likelihood Models (RELM), as well as the Collaboratory on the Study of Earthquake Predictability (CSEP), both founded after 2000.

Improvements in the accuracy and evaluation of these seismicity-based earthquake forecasts over the past 10 years suggest they provide the most promising avenue for actionable, operational earthquake forecasts. Current techniques include binary forecast techniques, with alarm regions of various sizes and time periods, as well as those that automatically generate probability fields. Significant work remains related to evaluating the various methods, determining optimal forecast time periods and accuracies and identifying reasonable levels of probability gain. Finally, a gap clearly still exists between forecasts that perform well on the order of days to weeks (foreshock and aftershock forecast models) and those performing well over a five- to 10-year period.

Both types of seismicity-based forecasts are limited by the quality of the data and the relatively short time length of the instrumental catalog. Errors or lack of information in these catalogs can result in large errors in the resulting forecasts, particularly for large events that have sparse statistics. In addition, the small number of large events that occur either regionally or worldwide makes definitive statistical evaluation of all these techniques extremely difficult at the present time. This serves to emphasize the importance of understanding the various time scales associated with the natural earthquake process and the fact that longer testing periods will be necessary to properly evaluate the viability and efficacy of forecast models.

One notable benefit of these methods is that they are generally formulated so that they can be updated to account for the changing nature of the earthquake fault system. These seismicity-based forecasting techniques allow for both regular updating of their forecasts as well as revised forecasts after the occurrence of large events, theoretically capturing the dynamics of the system.

Operational earthquake forecasting

The term operational earthquake forecasting refers to the capability to provide actionable information on the short-term time dependence of local or regional seismic hazards. As discussed above, seismic hazard changes dynamically in time: stress conditions change and alter the conditions that lead to new earthquakes. For example, the short-term stress and seismicity changes associated with aftershock sequences are well understood; they can be characterized relatively reliably in areas with longer seismic records. For example, the short-term earthquake probability (STEP) model that has provided aftershock forecasts for the United States Geological Survey (USGS).

The combination of statistical and physical analysis of earthquake interactions has made progress in understanding seismicity clustering such as aftershocks and swarms, leading to a considerable, ongoing effort to quantify the probability gain in these short-term forecasts — including attempts to incorporate it into the next version of the Uniform California Earthquake Rupture Forecast (UCERF). It is anticipated these seismicity-based operational forecasts will display order-of-magnitude improvements in probability gain over time periods of days to weeks. These developments suggest it is time to consider how best to prepare emergency hazard providers, businesses and the insurance industry to respond to short-term operational earthquake forecasts.

Earthquake early warning (EEW) systems

The goal of EEW is to reduce the damaging effects of earthquakes by providing a warning of between a few seconds and a few tens of seconds before the arrival of damaging ground motion. EEW systems are a new, promising tool to provide rapid information about upcoming ground shaking before the arrival of the large seismic waves.

The physical basis for EEW systems is the fact that destructive secondary and surface waves travel at about half of the speed of the first, smaller seismic waves. These are much slower than signals transmitted by telephone, Internet or radio. EEW systems use the capability of real-time monitoring and communications systems to process and transmit information faster than the slower seismic waves propagate, once the initial wave registers at a seismometer. The EEW system then provides information that an earthquake is occurring and potentially damaging ground motion is approaching the user site. Possible warning times can reach up to 70 seconds, depending on the distance between seismic source, sensor and user site.

A real-time EEW system consists of (1) sensors deployed in the field (a network of seismic stations); (2) some form of telemetry for communication between the seismic stations and a central receiving station; (3) a central receiving station for real-time data acquisition and processing; and (4) a broadcast system to send the information to the user sites.

EEW warnings can be used to evacuate buildings, shut down gas pipelines, slow or stop rapid-transit vehicles and high-speed trains, shut down manufacturing operations to decrease potential damage, save vital computer information and prevent data loss, shut down critical systems such as nuclear reactors and generate ground-shaking maps for emergency response and disaster relief operations.

Today, a number of countries are testing and implementing EEW, including the United States, Japan, Mexico, Turkey, Taiwan, Romania, Italy, Switzerland, and China. Integration of the resulting EEW notifications into emergency operation, business planning and government preparedness will provide an unprecedented opportunity to reduce damage and loss of life from these large, damaging events.

• Cavallo, E., Powell, A., Becerra, O., 2010. Estimating the direct economic damage of the earthquake in Haiti. Inter-American Development Bank working paper series, No. IDB-WP-163.

• Daniell, J.E., Wenzel, F., Vervaeck, A. [2011] “The Socio-economic effects of the 2011 Tohoku earthquake”, Geophysical Research Abstracts Vol. 13, EGU2011-14270.

• Field, E.H., 2007. Overview of the Working Group for the Development of Regional Earthquake Likelihood Models (RELM), Seis. Res. Letts. 78, 7-16.

• Jordan, T.H., 2006. Earthquake predictability, brick by brick, Seis. Res. Letts. 77, 3-6.

• Jordan, T.H. Jones, L.M, 2010. Operational earthquake forecasting: Some thoughts on why and how, Seis. Res. Letts., 81, 4, 571-574.

• Kanamori, H., 1981. The nature of seismicity patterns before large earthquakes. Earthquake Prediction: An International Review, AGU Monograph. AGU, Washington, D.C., pp. 1 – 19.

• Kovacs, P., 2010. Reducing the risk of earthquake damage in Canada: Lessons from Haiti and Chile. ICLR Research Paper Series, 49.

• Rundle, J.B., Tiampo, K.F., Klein, W., Sá Martins, J., 2002. Self-organization in leaky threshold systems: The influence of near mean field dynamics& its implications for earthquakes, neurobiology and forecasting. PNAS, Suppl. 1, 99, 2463.

• Tiampo, K.F., Rundle, J.B., McGinnis, S., Gross, S., Klein, W., 2002. Mean-field threshold systems and phase dynamics: An application to earthquake fault systems. Eur. Phys. Lett. 60, 481-487.

• Wu, Y. M., Kanamori, H., Allen, M. and Hauksson, E, 2007. Determination of Earthquake Early Warning Parameters, _c and Pd , for Southern California. Geophys. J. Int., 170: 711-717.