The Seismic History of Vermont

The earthquake history of the northeastern U.S. and adjacent areas in Canada is known from a variety of sources (Chiburis, 1981). For earthquakes which took place prior to modern instrumental recording, the sources of information on the earthquake activity come from documents such as diaries, local histories, newspaper articles, and other historical archive material. Starting in the 1930's press reports of earthquakes were supplemented by instrumental recordings of the earthquake activity. A further improvement in earthquake monitoring took place in the mid 1970's when a regional seismic network was installed in New England and vicinity by several university research groups in seismology, primarily those at the Weston Observatory of Boston College, at the Massachusetts Institute of Technology, and at the Lamont-Doherty Geological Observatory of Columbia University. This network, funded primarily by the U.S. Nuclear Regulatory Commission and by the U.S. Geological Survey, allowed for the first time the regular recording of earthquakes too small to be felt. This information on the small earthquake activity is important in learning more about the causes, the probabilities, and possible effects of earthquakes in the region.

Several investigators have taken the earthquake information from these various different sources to compile earthquake catalogs for the region. An earthquake catalog is a listing of all the earthquakes from a region, typically including such information as the date, time, location and size for each event as well as other information deemed important by the compiler. In this study we used the northeastern earthquake catalog presently available at Weston Observatory. This catalog is comprised of the earthquake activity through 1977 compiled by Chiburis (1981) and supplemented since 1977 with the earthquake data from the Northeastern United States Seismic Network Bulletins published by Weston Observatory. In this catalog the locations or epicenters of the events, in other words the points on the surface of the earth below which the earthquake radiated its energy, are listed by latitude and longitude.

The sizes of the earthquakes in the Weston Observatory catalog are indicated in one or both of two different ways. For the older earthquakes the sizes of the events are indicated by the maximum intensity of the event. In earthquake seismology the term intensity refers to a number (normally listed as a Roman numeral) assigned to a given description of ground shaking. The most commonly used intensity scale in the United States is the Modified Mercalli intensity scale. This scale, described more fully in Appendix B, runs from intensity I (not felt) to intensity XII (total destruction). Minor damage to structures is assigned intensity VI, moderate damage intensity VII, major damage intensity VIII and severe damage intensities IX to XII. The maximum intensity of an earthquake is the highest intensity reported from the earthquake, usually near or at the epicenter of the event.

The other way that the size of an earthquake is listed in the catalog is by the instrumental magnitude of the earthquake. The magnitude of an earthquake, often called the Richter magnitude after the seismologist Dr. Charles Richter who proposed the magnitude scale in 1935, is a measure of the size of the earthquake based on the measurements made from seismographic instruments. The magnitude scale is designed to be a measure of the size of the earthquake at its source.

As part of this study a number of the earthquakes in the Weston Observatory earthquake catalog for Vermont were reexamined and updated with new information. As documented by Nottis (1983) some of the events reported from Vermont were not earthquakes at all, and these were stricken from the catalog. The magnitudes for a number of the events were recalculated by Ebel (1987), and the locations of several of the largest shocks from Vermont were re-computed by Dewey and Gordon (1984) or by J. Ebel in unpublished work.

A total of 63 earthquakes or possible earthquakes centered within Vermont through 1993 are contained in the final earthquake catalog from this study and are shown in Figure 2-2. As many as 11 of the 63 events, all from the early 1980's, may in fact be quarry blasts which were misidentified as earthquakes (see Appendix C for more information on these events).

The largest events in Vermont in this catalog are the July 6, 1943 earthquake near Swanton, VT and the April 10, 1962 event near Middlebury, VT. Both of these earthquakes had magnitude 4.1. Next to these there was an earthquake of magnitude 4.0 at Brandon, VT on March 31, 1953. The April 24, 1957 event at St. Johnsbury, VT is reported by Chiburis (1981) as having a maximum intensity of V, putting it among those earthquakes with the highest intensity centered in Vermont. However, Ebel (1987) reported its magnitude as only 2.4. Table 2-1 lists the dates, times, magnitudes and maximum intensities of the largest magnitude earthquakes centered in Vermont through 1993.

Mag. is the Richter magnitude of the earthquake from (Ebel, 1987). MMI is the maximum Modified Mercalli intensity of the earthquake, as listed in the publication U.S. Earthquakes for the appropriate year.

It is clear from Figure 2-2 that the earthquakes in Vermont scatter broadly across the state, although the largest events have occurred in the northwestern part of the state. Compared to the rest of region, there is relatively less seismicity centered within Vermont itself. However, there are quite active zones of earthquakes in the Adirondack Mountains of northern New York state, in southern Quebec, in central New Hampshire and in eastern Massachusetts. As these more active earthquake areas abut Vermont, they represent possible sources of significant earthquakes which can affect Vermont, as is discussed in the following sections.

All of the instrumental earthquakes in Vermont have been centered within about 15 km (about 9 miles) of the earth's surface, typical of the shallow earthquakes throughout the entire northeastern U.S. (Ebel and Kafka, 1991). Earthquakes centered near the surface of the earth are more capable of generating damaging ground shaking than deeper earthquakes (below 50 km depth), so the shallow focal depths of the earthquakes of the region mean that even moderate magnitude earthquakes (above 5.0) can be damaging.

There are no active faults confirmed in Vermont, nor is there even a clear association between the earthquakes and the geologic faults in the state (Ebel and Kafka, 1991). An earthquake itself is actually a combination of two related events. Pressure on the rock in the earth can cause the rock to crack over a large planar area and then to slide along this crack. The crack is the fault, and sliding of the rock on either side of the crack generates vibrations which are the seismic waves that shake the surface of the earth (Figure 2-3). Of course, once the rock is fractured, the crack remains permanently in the rock and can be observed even hundreds of millions of years later. An active fault is defined as one that is presently capable of sliding and thus releasing seismic waves. Many faults which geologists map can be inactive faults, ones which slipped in the geologically distant past but which are not capable of slipping today. Some faults occur entirely at depth and so never reach the earth's surface where they can be observed by geologists. Such buried or blind faults are an unsuspected seismic hazard because often they are not recognized until a large earthquake occurs on them.

In order to identify active faults, geoscientists look for several corroborating pieces of evidence. They first look for faults mapped at the surface or inferred in the subsurface from geophysical investigations. They second look for earthquakes along the fault, and in particular for earthquakes which align along the trend of the fault. They finally look for surface evidence that the fault has had movement in geologically recent time (within the last 10,000 years). Surface geologic evidence of recent fault movement is the most convincing argument that a fault is active, but it is also very difficult evidence to find.

While there are many faults which have been mapped in Vermont, no geologic evidence has been found for recent fault movement anywhere in the state (Ebel and Kafka, 1991). Most of the faults occur in the southwestern part of the state or in the eastern part of the state along the Connecticut River. Recent geologic work has found a number of faults in the Green Mountains. There is insufficient earthquake activity along any of these fault systems to argue that they are active faults. Furthermore, a number of the earthquakes in Vermont have occurred in places where there are no faults shown on geologic maps. These earthquakes could represent minor rock cracking which is not related to more significant earthquake activity, or they could represent earthquakes on buried faults which are not observed at the surface. The January, 1994 earthquake at Northridge, California is an example of a buried fault with no direct surface expression (Hall, 1994). Thus, as is true throughout the rest of the northeastern U.S., there are no confirmed active faults within the State of Vermont, and the identification of active faults in the region must await the accumulation of more earthquake and geologic data. One consequence of this conclusion is that the geology of Vermont provides no direct clues as to where strong earthquakes may be possible in the state.

Analysis of the seismic waves generated from the earthquakes in the region and of the pressure directions measured in boreholes strongly supports the idea that the pressures which cause New England earthquakes come from the movement of the North American plate over the earth (Ebel and Kafka, 1991). The surface of the earth is composed of a dozen major tectonic plates, each about 100 km (60 miles) thick. Heat escaping from the earth's interior slowly moves the plates over the surface of the earth. Places where the edges of two plates meet are zones of large pressures on the local rocks. In these areas mountains or valleys usually form, and earthquakes are frequent. Most active volcanoes are also found at the edges of plates. This process of plate motions and deformations is called plate tectonics. The Appalachian Mountains were formed during earlier geologic ages when the east coast of North America was at the edge of a tectonic plate.

Today, one boundary of the North American plate is at the center of the Atlantic Ocean, where North America is spreading away from Europe and Africa (Figure 2-4). The other North American plate boundary lies along the western coast of North America. There the North American plate is pushing against the Pacific Ocean plate and other smaller plates in the Pacific Ocean. Astronomical measurements show that this east-to-west movement of North America is quite constant, so the pressure in the interior of the plate is always slowly building up at a steady rate. This means that the pressures which drive the earthquakes in the northeast, and therefore the earthquake activity itself, will continue indefinitely in the future. A few decades ago it was thought that the earthquakes in the region were caused by a slow upward rebound of North America after the melting the continental glaciers about 10,000 years ago. The glacial rebound theory about the causes of the northeastern earthquakes is not supported by the latest seismological and geological evidence. If post-glacial rebound is the cause of the earthquakes in the region, then the direction of the maximum pressure in the rocks of New England would be quite different from that which is actually measured (Ebel and Kafka, 1991).

While it is easy to summarize the past seismic history of Vermont, estimating what earthquakes and earthquake effects may happen in the future is much more difficult. Several factors conspire to make this so. First, it is not possible, even in quite seismically active areas like California or Japan, to predict earthquakes. There have been no consistent forewarning signals before large earthquakes that seismologists can use to predict the coming of a large shock. Second, there is inherent uncertainty in postulating future strong earthquakes that are larger than those that have happened in the past or that are at localities that have not had significant earthquakes in the past. Third, since no active faults have been identified to date in the northeast, it is impossible to point to any particular geologic features as being the most likely to generate a large or damaging earthquake. Fourth, there is some uncertainty in estimating the probabilities of future strong earthquakes in the region. The documented earthquake history is only a few hundred years long, and it becomes progressively more incomplete as one goes backward in time. Finally, the strength of the ground shaking at different distances from an earthquake epicenter can only be approximately estimated. Past data are a guide to this estimation, but there can be variations due to the particular magnitude, depth and location of the earthquake as well as the sites where the earthquake shaking is felt.

Two different approaches are taken addressing the question of what future earthquake effects are possible in Vermont. One approach is a deterministic seismic hazard analysis, where several different large earthquakes are postulated at epicenters around Vermont and then the ground shaking effects in Vermont are estimated. The other approach is a standard probabilistic seismic hazard analysis, where the level of ground shaking that has a low probability of being exceeded in each of several different time periods (50 years, 100 years and 250 years) are calculated throughout Vermont. The latter is a standard approach that is often used as a basis in determining seismic design requirements in building codes.

The deterministic seismic hazard scenarios provide snapshots of what would happen in Vermont should one or more of the postulated earthquakes take place. However, that approach does not address an important question which often arises: how is a particular site in Vermont affected by all the different possible earthquake activity in the region? To address this question, the method of probabilistic seismic hazard estimation was developed. In this method the probabilities of different levels of ground shaking at a site due to earthquakes all throughout the region are calculated, resulting in the accumulation of a final set of estimates which are the probabilities of different levels of ground shaking at the site. From this result the chances of a site experiencing any level of ground shaking can be realistically estimated. For instance, as a result of a probabilistic hazard analysis the strength of ground shaking which has, say, 1 chance in 1,000 of occurring per year can be estimated. It is not possible to relate the ground motions in this type of analysis to any one particular earthquake since the method is based on the accumulation of probabilities from all of the earthquake activity in an area. However, the method is quite meaningful at a site since it takes into account all possible earthquakes around the site.

The output of a probabilistic seismic hazard analysis is an estimate of the likelihood of various ground motion levels being exceeded, typically as annual probabilities. This information is then used to determine the ground motion associated with a specified chance of being exceeded over various exposure periods. For example, the U.S. Geological Survey has published two national maps which show the levels of ground shaking which have only a 10% chance of being exceeded over given time periods (90% chance of non-exceedance as they often state it), one map for a 50-year time period and one map for a 250 year time period. The maps give the levels of ground shaking across the country that have a low probability of being exceeded during the stated time period. This idea is similar to that used in the analysis of flood potential where flood maps might show the once-in-100-year flood level, the once-in-250-year flood level, or the once-in-500-year flood level. Since many buildings and other structures have an estimated lifetime at construction of about 50 years, probabilistic hazard maps for a 50-year exposure period are often made. Probabilistic hazard estimates for 100-year and 250-year exposure periods are also used to cover longer projected time periods. Such maps have been used for determining the seismic design levels for structures such as landfills and dams.

In this study probabilistic seismic hazard maps for Vermont and vicinity have been computed using the latest information on earthquake locations and magnitudes and on strong ground motion attenuation models. Figures 4-7, 4-8 and 4-9 illustrate for Vermont and surrounding areas the peak ground accelerations for 50 years, 100 years and 250 years that have only a 10% probability of being exceeded, as computed in this study. The ground motions values in these figures were computed for a site on hard bedrock . Figures 4-7, 4-8 and 4-9 show that for all three time durations the northwestern corner of Vermont is likely to face the strongest ground shaking. The peak ground acceleration values are lowest in the central part of the state and rise to somewhat greater values along the eastern boundary in the state. This is consistent with the past seismic history where the strongest ground shaking in the past has been in the northwestern part of Vermont, and earthquakes have been felt most frequently in the northwestern part of the state.

It has long been known by those who study damage effects from earthquakes that the strength of earthquake shaking can differ quite significantly over distances as short as a few city blocks. Furthermore, observations made after destructive earthquakes have shown a correlation between damage and local geology, with the destruction being in general larger on unconsolidated sediments (also called "soft soils" by geotechnical engineers) or fill than on consolidated sediments (also called "hard soils" by geotechnical engineers) or on bedrock (ledge) (for example, Seed et al., 1972; Seed et al., 1987; and Loma Prieta Reconnaissance Report, 1990). In this context geotechnical engineers use the term soils interchangeably with terms like sediments and fill, referring to any clays, sands, silts or gravels above the bedrock or ledge. Research has revealed that the surface soil conditions at a site have a major effect on the strength of ground shaking experienced at that site. In particular, a thick layer of unconsolidated soils can significantly modify the ground shaking compared to that which is experienced at nearby bedrock sites. Such thick soils can occur naturally in places like river bottoms, or they can be man-made in areas where landfill was used to extend a city into a swamp, river, lake or ocean. This ground shaking amplification, or increase in the strength of ground shaking due to the existence of a thick layer of soft soils, can be quite pronounced.

Another set of phenomena that can take place in strong earthquake shaking are soil failure effects, such as soil liquefaction and lateral spreading of soils. These occur when water-saturated sandy layers a few feet below the surface of the earth are strongly shaken. In soil liquefaction pressure builds up in the water saturated layer to the point where sand and soil erupt up to the ground surface. This eruption can form what looks like a sand volcano or sand boil, typically a few feet to a few tens of feet in diameter. The ground can shift around the edge of such a sand volcano, distorting the foundations of buildings in the area due to settlement in the soils. Lateral spreading of soils occurs over large areas which are acres in size. In lateral spreading the water-saturated layer loses most of its strength to support the soils above, and the overlying soils slump toward lower-lying areas. What makes lateral spreading such a problem is that the slopes can be quite small (only a few degrees) and that under normal conditions (i.e., without strong earthquake shaking) no lateral spreading normally can take place. Once again, buildings and other constructed facilities founded on soils that undergo lateral spreading will have their foundations distorted.

Earthquake ground motions in Vermont can be locally modified by soil conditions. In particular, poorly consolidated or unconsolidated soils can significantly amplify ground shaking relative to the bedrock below the soils, up to a factor of 3 at some frequencies of ground shaking. In an analysis of Chittenden County in Vermont, the distribution of surficial soils suggests that a few areas in the county could experience significant amplification of earthquake ground shaking. These areas are generally in river valleys or along Lake Champlain, including some parts of the city of Burlington. Other areas in Chittenden County could experience minor ground shaking amplification.

Two major factors control the amount of groundshaking amplification that soils can undergo. The first is the stiffness of the soils from the surface to the bedrock. The second major factor is the thickness of the soil layers. The thicker the unconsolidated soil, the more likely it is that there will be strong groundshaking amplification, with soils that are over 100 feet thick being the most prone to amplification. However, while the 1970 Surficial Geologic Map of Vermont shows the types of the surficial layers from throughout the state, it gives no information about the thicknesses of these layers. A detailed study of the thickness of these sedimentary layers under Burlington would more clearly define the amount of local amplification that could occur during earthquake ground shaking.

For sites on level ground soils the most important earthquake effects are: (1) a modification of the amplitude, frequency content and duration of the ground shaking caused by the soil amplification (i.e. soil factor), and (2) the failure, settlement or liquefaction of the soil near the ground surface.

In our analysis of typical soils for Vermont, we chose to quantify and qualify the effect of the local soil conditions for our study using 4 different representative homogeneous soil columns with thicknesses varying from 25 to 200 ft. The shear wave velocities for the soil layers were determined using the empirical correlations between index and field soil properties and seismic velocities presented by Sykora (1987), and the input earthquake ground motion used in the computer program was that developed as part of the seismic hazard analysis for the new Boston Central Artery highway construction project, digitized every 0.015 seconds. The peak ground acceleration of this input earthquake ground motion was normalized to 16% g, consistent with the 50-year peak horizontal ground motions for Chittenden County shown on Figure 4-7.

As part of this study calculations were performed to estimate the amount of ground shaking amplification, relative to that in the bedrock, which could take place in typical soils in Vermont. Soil layers ranging in thickness from 25 feet to 200 feet were analyzed. The thinnest soil layer only amplified the ground shaking at very short periods (less than .1 seconds), while the thicker soils significantly amplified the ground shaking in the period range between 0.1 seconds and 1.0 second. This amplification could increase the damage to those structures situated on soils with properties similar to those used in this analysis. Significant amplification can occur on some thick soils at those frequencies of seismic waves to which many important structures in Vermont potentially are most sensitive.

The most likely damaging earthquake scenarios come from strong earthquakes (above magnitude 6.5 or so) centered in the Adirondack Mountains of New York or in southern Quebec or from moderately strong earthquakes (perhaps as large as magnitude 5.7) in northwestern Vermont. Modified Mercalli intensity VI to VII shaking could occur in the bedrock or on hard soils at Burlington in any of these scenarios. The probabilistic seismic hazard values computed in Section 4-2 (Figs. 4-7, 4-8, 4-9) show that in the bedrock in Chittenden County, the strongest peak horizontal ground acceleration likely to be experienced in a 50-year period is about 16% g. For time periods of 100 years and 250 years this peak acceleration value rises to about 25% g and 35% g, respectively. The peak horizontal ground acceleration threshold for intensity VI ground shaking, roughly that at which damage to buildings begins, is about 8% g at soft-soil sites and 14% g at hard-rock sites (Krinitzsky and Chang, 1988). Thus, there is a likelihood that buildings in the Burlington area will experience some level of potentially damaging ground shaking if those buildings last 50 years or more.

Vermont should begin to take steps to minimize the consequences of a damaging earthquake to the state. This should be a well-planned, steady effort. Any effort to mitigate the effects of earthquakes should have three primary aims: (1) to save lives and minimize injury, (2) to minimize the damage to structures, and (3) to enable the rapid recovery to normal life after the earthquake. The following gives some specific details about actions that could be taken in Vermont to achieve these three aims of earthquake hazard mitigation.

Vermont does have some seismic provisions in the building codes used in the state. The State of Vermont adopted the 1987 BOCA National Building Code (NBC) with the 1988 supplement and state amendments as the state building code. Only a few municipalities in Vermont have adopted the state code or its equivalent. Building plans are not reviewed for seismic design in any community except Burlington and at the state level. The seismic provisions in the 1993 BOCA NBC provide a level of seismic safety comparable to that of the 1988 NEHRP Recommended Provisions for the Development of Seismic Regulations for New Buildings (published by the Federal Emergency Management Agency) and better than that in the 1987 BOCA NBC. The implications of this information are clear; Vermont should adopt the latest BOCA code, including its seismic provisions, if it is to have an acceptable measure of seismic safety for its buildings. It is recommended that Vermont adopt immediately the latest BOCA NBC, currently the 1993 edition, including the seismic provisions for its new buildings.

Seismic design is also being required by current federal regulations for landfills and for all new federal construction, and suggested design provisions have been published for such structures as highway bridges. Design recommendations are also available for so-called lifelines such as roads, pipelines and utility systems. We recommend that the latest seismic provisions be followed to ensure the safety of these critical structures.

A number of public policy steps are suggested to address the hazard from earthquakes and to minimize injury and damage from earthquakes. In the area of public education recommended steps are: