July/August 2008

The Importance of Small Earthquakes

Regional and local seismic networks have become ubiquitous throughout the world to study local seismotectonics, to provide immediate hypocentral and magnitude information following felt or damaging earthquakes, to create ShakeMaps of the ground shaking due to earthquakes, and to form the sensing backbone of incipient early warning systems of earthquake shaking. They also provide data for earthquake forecasting research, nuclear test monitoring, and analyses of nonearthquake events such as explosions, intense storms, mine collapses, etc. Some seismic networks are aimed at detecting and recording seismic events down to the lowest possible magnitude, while others are strong-motion networks that aim to properly record the strongest possible ground shaking that can take place. Significant resources are invested annually in regional and local seismic monitoring, and such networks have become a permanent fixture in many areas of the world.

Is it important to record and process seismograms of all earthquakes down to magnitude 3? Magnitude 2? Magnitude 1? Magnitude 0 or below?

My purpose in writing this opinion piece is not to question the intrinsic need for regional and local networks. On the other hand, I do feel it is important that there should be a full and informed public discussion about the importance of the small earthquakes that we record. Is it important to record and process seismograms of all earthquakes down to magnitude 3? Magnitude 2? Magnitude 1? Magnitude 0 or below? Informed discussions on these issues sometimes take place when a decision must be made about a particular site, such as whether to put a high-level radioactive waste repository at Yucca Mountain in Nevada. But for the most part, regional and local seismic networks, once they have been initiated, tend to grow based on a “more is better” philosophy as funding is available. I believe that the existence and growth of regional seismic networks should be driven by performance-based scientific and societal needs, such as to better determine the seismotectonics, seismic hazard, and strong-motion wave propagation characteristics of the region being monitored and to produce and deliver better earthquake information that is vital for public safety issues.

The demand to gather as much seismological data as possible when large, damaging earthquakes occur is beyond discussion. Scientists and nonscientists alike agree that this is a priority. Furthermore, all regional and local seismic networks try to gather sufficient data to locate and compute magnitudes for all felt earthquakes within the area they monitor. The public can get quite excited when an earthquake is felt, even if it is much too small to cause any damage, and people want to know as much about these smaller shocks as they do about the larger, destructive events. Thus, operators of regional and local seismic networks invariably work to satisfy this public need-to-know with information on the locations and magnitudes of all locally felt earthquakes. Furthermore, many regional and local seismic networks are able to gather at least some information about local earthquakes that are too small to be felt.

Putting aside these public-information considerations, are there scientific reasons why seismologists should devote time and money to monitoring and studying small earthquakes in seismically active areas? I believe that there are. From a scientific perspective, accurate locations and magnitudes, along with focal mechanisms, stress drops, and other technical source parameters for smaller earthquakes, are vital for those who study seismotectonics and seismic hazard on a local or regional basis. It has been recognized for many decades that in seismically active regions, many of the small earthquakes occur on active faults, and determinations of the hypocenters of small earthquakes help delineate where the active faults lie. Small earthquakes also help identify volcanically active zones, other active geologic structures like growing folds, and areas of induced seismicity. Thus, to learn more about the seismotectonics of a region, the first tool that seismologists always want to employ is a regional seismic network with the capability of detecting and locating earthquakes to the smallest possible magnitude. The smallest magnitude that such networks need to record for seismotectonic studies has never been established by any scientific analysis of which I am aware, but for most regions of the world, regional seismic networks seem to aim for detecting all earthquakes down to about magnitude 2.0 within their monitored region. In many cases, local seismic networks are sufficiently dense to achieve completeness thresholds well below magnitude 2.0.

There likely is useful information other than for seismotectonic analyses in the very small earthquake activity of a region. In a paper some colleagues and I published several years ago (Ebel et al.2000), we argued that the small earthquake activity in central and eastern North America and other intraplate regions may be leftover aftershock activity from strong earthquakes that took place hundreds or perhaps even a few thousand years ago. If this is indeed the case, then the small earthquake in central and eastern North America can help us better determine the past seismic history and give us a better handle on the seismic hazard in this heavily populated region. The smaller the earthquakes that we can detect and locate in the eastern United States, the faster we can learn about the seismotectonics of this region. Another interesting set of studies has been published by Alan Kafka (2002) and in GSA Special Paper 425, 2007). Kafka has found that a very high percentage of large earthquakes occur at localities where smaller earthquakes (down to M 2.0) had been detected in previous years or decades by regional or local seismic networks. It would be interesting to extend this work to even smaller earthquake magnitudes to see if the association of large and small earthquakes remains. These and other studies show that small earthquakes do tell us quite a lot about the potential in time and space of future large earthquakes in regions far from plate boundaries.

One important benefit of regional and local seismic network monitoring is that when strong earthquakes occur, the networks are in place to immediately record the aftershocks, which help delineate the fault plane or planes that were active during the mainshock rupture. Smaller aftershocks are more abundant than larger aftershocks, and so the smaller the magnitude threshold to which the aftershock monitoring is sensitive, the better the delineation of the fault surface upon which the main earthquake occurred. While portable instrumentation certainly helps with detailed aftershock monitoring, it cannot be employed until hours or days after a mainshock, often after the most intense part of an aftershock sequence normally has taken place.

…improving the detection capabilities of a regional or local seismic network by one or two orders of magnitude entails an increased cost that is far from trivial. It is this tradeoff between scientific benefit and costs that I think needs a more extensive public discussion.

Another reason that regional and local seismic network operators collect data on small earthquakes is to estimate the recurrence rates of larger earthquakes. The empirical Gutenberg- Richter magnitude distribution of earthquakes seems to apply almost universally across the globe and down to the smallest natural earthquakes that have been recorded, making this relation an important tool for estimating how often strong earthquakes might be expected to occur, even in places where damaging earthquakes have never been observed in historical time. Seismic hazard analyses rely heavily on this key assumption, and hence they rely heavily on a complete record of the small earthquakes in the region of the seismic hazard analysis.

While I believe that the above arguments are quite compelling for many seismically active geologic structures, they clearly cannot be applied universally. For example, neither of the two segments of the San Andreas fault that had major earthquake ruptures in 1857 and 1906 is delineated by small earthquake activity. Perhaps this lack of small-magnitude seismicity reflects the current state of the rupture cycle on this major plate-boundary fault, or it is an indication of low strength of the fault in the areas of these two ruptures. Also, many small earthquakes take place in California that do not relate in any obvious way to known, active faults. Perhaps these small earthquakes are indications of faults that will experience large earthquakes sometime in the future, much as happened in the 1971 San Fernando earthquake and the 1994 Northridge earthquake.

I think the need to collect data on small earthquakes to study seismotectonics, especially in populated areas where the seismic hazard needs to be determined for engineering purposes, is extremely strong. But my real question is, “How small is small enough?” Improving the detection and location capabilities of a regional or local seismic network by one magnitude unit (say from M 3.0 to M 2.0) should increase by about a factor of 10 the number of earthquakes that are recorded per year. Improving network capabilities by two orders of magnitude should garner about 100 times more earthquakes annually. However, improving the detection capabilities of a regional or local seismic network by one or two orders of magnitude entails an increased cost that is far from trivial. It is this tradeoff between scientific benefit and costs that I think needs a more extensive public discussion.

My opinion is quite simple. I think the scientific evidence is compelling that much can be learned about seismotectonics and seismic hazard by collecting a complete dataset of high-quality locations, magnitudes, and waveforms for earthquakes as small as M 1.0, and likely even smaller. In a plate boundary region such as California, the smallest earthquakes that are located routinely by the regional seismic networks (~ M 0.5) generally occur on the same geologic structures where the larger earthquakes are detected, and so even M 0.5 earthquakes are useful for delineating seismically active structures. In those places where a complete dataset has been collected, the Gutenberg-Richter distribution seems to hold true down to magnitudes as small as magnitude 0.0 or below, and so the rates of very small earthquakes probably tell us something about the average rates of much stronger earthquakes on those geologic structures that are known or thought to be seismically active.

I believe that the same scientific arguments that are used to justify monitoring the small earthquake activity in plate boundary regions are also applicable to stable cratonic regions like central and eastern North America. Many parts of the eastern United States and southeastern Canada have high population densities and many built structures with little or no seismic design, and so the seismic risk is quite high even if the seismic hazard is not as high as along the west coast of North America. Much has been learned about the active seismotectonic structures in the New Madrid seismic zone, at Charleston, South Carolina, and at the Charlevoix seismic zone in Quebec, from monitoring over a long period of time the small local earthquake activity in these seismic source zones. More needs to be learned about the active seismotectonics along the important urban corridor from Boston to Washington, D.C., in the heavily industrial eastern Great Lakes basin, and in the Wabash Valley area of the Midwest. The eastern Tennessee seismic zone is thought by some (including me) to perhaps represent the after- shock zone of a major prehistoric earthquake, but the existence and extent of this enigmatic seismic zone is known only from long-term monitoring of the small earthquakes of the southern Appalachians. A better quantification of the seismic hazard in these and other areas of eastern North America requires a better understanding of the seismotectonics and of the potential for future strong earthquakes in this stable cratonic region. Studying the future small earthquakes in this area is one important way to achieve this better understanding.

An investment in the development of new data processing techniques for regional seismic networks can help with monitoring small earthquakes even in those areas where the seismic station density is not high.

In setting the priorities for funding seismic networks in a large and diverse region like the United States, I think it is vital that scientific issues be clearly separated from political realities. For example, lack of funding (a political reality) does not justify the idea that it is unimportant to monitor the small earthquake activity of eastern North America (a scientific argument). Rather, tight funding simply means that there are not enough resources to support high-density regional seismic networks in all parts of the country. Is operating high-density regional seismic networks the only way to study small earthquakes? Not necessarily. In the western United States, the USArray experiment has already shown that an increase in the density of seismic stations in poorly monitored parts of the country coupled with powerful new data analysis techniques can reveal previously unknown earthquake activity that may be illuminating unstudied seismotectonic processes. An investment in the development of new data processing techniques for regional seismic networks can help with monitoring small earthquakes even in those areas where the seismic station density is not high. Such techniques could include: improved identification of depth phases on seismograms; more accurate local and regional waveform modeling, particularly at higher frequencies than is possible today; better methods to identify and pick seismic phases, especially at low signal-to-noise ratios; better event association methods for sparse regional seismic networks; and combining targeted local deployments of seismic stations with regional network monitoring. Such new techniques could benefit seismic monitoring in all parts of the United States. 


Ebel, J. E., K.-P. Bonjer and M. C. Oncescu (2000). Paleoseismicity: Seismicity evidence for past large earthquakes. Seismological Research Letters 71, 283–294.

Kafka, A. L. (2002). Statistical analysis of the hypothesis that seismicity delineates areas where future large earthquakes are likely to occur in the central and eastern United States. Seismological Research Letters 73, 990–1,001.

Kafka, A. L. (2007). Does seismicity delineate zones where future large earthquakes are likely to occur in intraplate environments? In Continental Intraplate Earthquakes: Science, Hazard and Policy Issues, S. Stein and S. Mazzotti, eds. Boulder, Colorado: Geological Society of America, 35–48, doi: 10.1130/2007.2425(03).

John E. Ebel
Weston Observatory of Boston College
Email: ebel [at] bc [dot] edu


To send a letter to the editor regarding this opinion or to write your own opinion, you may contact the SRL editor by sending e-mail to <lastiz [at] ucsd [dot] edu>.




Posted: 20 June 2008