Seismological Research Letters
 © 2004 by the Seismological Society of America
Abstract
The moment magnitude determination for prehistorical earthquakes in the Wabash Valley seismic zone traditionally has relied on modified magnitudebound liquefaction curves. The calibration of the moment magnitude and maximum distance to surface evidence of liquefaction curve for the central United States depended, in part, on the scaling of the 31 October 1895 earthquake as an M 6.8 event. Returning the M of the 1895 earthquake to a previously accepted value range of M 5.9 to M 6.2, and permitting the distance between the epicenter and the resulting liquefaction sites to vary, lowers the M of the prehistorical earthquakes by at least 0.6 units.
INTRODUCTION
Obermeier et al. (1993), Munson et al. (1995, 1997), Pond and Martin (1997), and others have identified and dated several paleoliquefaction sites in southwestern Indiana and southern Illinois that were subsequently used to estimate the magnitudes and epicenters of paleoearthquakes in the area. Magnitude scaling of the prehistorical earthquakes relies, to a large extent, on empirical data relating moment magnitude to the maximum distance of surface liquefaction evidence (Figure 1). The solid line in Figure 1 is Ambraseys' (1988) limiting distance for liquefactioninduced ground failure versus moment magnitude for worldwide shallowfocus earthquakes. The shaded area in the figure is the Obermeier et al. (1993) estimated range of an upper boundary for central United States earthquakes.
It is well established that attenuation of seismic waves in central and eastern North America is low compared to attenuation in other areas of active tectonics (e.g., California) (Nuttli, 1973; Benz et al., 1997, among others). In addition, Obermeier et al. (1993) described the sediment in the Wabash River Valley as having high liquefaction susceptibility. Because of the low seismicwave attenuation in the central United States and the high liquefaction susceptibility of the Wabash Valley sediment, any liquefactioninduced ground failure distance versus momentmagnitude relationship for the Wabash River Valley area would be expected to plot at or near Ambraseys' (1988) “maximum epicenter range” line. The shaded area in Figure 1, however, which is the liquidinduced ground failure versus momentmagnitudebound curve used by Obermeier et al. (1993) and others for the Wabash River Valley area, is indicative of either higher attenuation of seismic waves, or lower susceptibility to liquefaction of the Wabash River Valley sediments.
The objective of this short note is to offer a correction to scaling of prehistorical earthquakes in the Wabash River Valley and suggest a revision of their moment magnitudes.
REVISED MOMENT MAGNITUDES
In Figure 1, the 31 October 1895 earthquake, which occurred in the southeastern Missouri/southern Illinois area (Figure 2), is plotted as having a moment magnitude of 6.8. The M 6.8 (Hamilton and Johnston, 1990) for the 1895 earthquake is higher than what a number of other investigators have assigned to the earthquake. Alternative estimates for the magnitude of the 1895 earthquake follow:

Nuttli (1974) as m_{bLg} 6.2 (M 6.2)

Street et al. (1986) as m_{bLg} 6.25 (M 6.2)

Stover and Coffman (1993) as M_{fa} 5.9 (M 5.9)

Bakun et al. (2003) as ranging between M 5.7 and M 6.3, with the preferred magnitude being M 6.0.
Stover and Coffman's (1993) M_{fa} is their nomenclature for a bodywave magnitude based on the felt area. The moment magnitudes shown in parentheses for Nuttli (1974), Street et al. (1986), and Stover and Coffman (1993) are all based on their intensityderived estimates of the bodywave magnitudes for the event, the bodywave to seismicmoment relationship of Johnston (1996), and the moment magnitude scale of Hanks and Kanamori (1979).
The 31 October 1895 earthquake induced liquefaction at several sites in the Charleston, Missouri area, and for this reason Nuttli (1974) picked the epicenter of the earthquake as being at Charleston. Alternatively, Bakun et al. (2003) concluded from the distribution intensity data assigned by Hopper and Algermissen (1980) to the 1895 event that its epicenter was in southern Illinois (Figure 2). Bakun et al.'s (2003) epicenter is 100 km north of Charleston, the area where liquefaction was observed. These two studies define the variation we suggest can be applied to central United States calibration of the maximum distance to observed surface evidence of liquefaction.
Ambraseys' (1988) relationship between the limiting epicentral distance to sites at which liquefaction has been observed, and the moment magnitude for shallow earthquakes, is shown as the middle curve in Figure 3. The upper curve in the figure is the maximum distance to liquefaction derived (from 1D soil response modeling) by Pond and Martin (1997) for scaling prehistorical earthquakes in the Wabash River Valley. The dashed rectangle represents the variation of suggested magnitudes and distances to liquefaction for the 1895 earthquake. The epicentral distances to liquefaction bounded by the rectangle are 16 and 100 km. The 16 km distance is from Obermeier et al. (1993) (i.e., using the Nuttli [1974] epicenter), whereas the 100 km distance is the distance between Bakun et al.'s (2003) epicenter for the 1895 event and the liquefaction in the Charleston, Missouri area. The M 5.9 to M 6.2 vertical range for the rectangle is within Bakun et al.'s (2003) suggested range of M and includes their preferred value of M 6.0. In addition, the estimated moment magnitudes from Nuttli's (1974), Street et al.'s (1986), and Stover and Coffman's (1993) bodywave magnitudes lie within the rectangular area. The location of the rectangle for the 1895 event with respect to Ambraseys' (1988) limiting distance curve is also reasonable for an area characterized by low attenuation of seismic waves, and whose sediments have a high susceptibility to liquefaction (e.g., Wabash River Valley).
Like the Wabash River Valley, eastern Canada is an area characterized by low attenuation of seismic waves (Benz et al., 1997). As one of several examples, the M 6.3 St. Lawrence earthquake of 1 March 1925 resulted in numerous ground cracks in parts of Quebec City, where the depth of the soil was considerable (Coffman and von Hake, 1973). As noted by Bakun et al. (2003), the distance between the epicenter and the area of liquefaction for the 1 March 1925 earthquake is approximately 148 km (Figure 3). The distance to liquefaction observed for the 1925 earthquake is significantly greater than Ambraseys' (1988) limiting distance curve. Because the attenuation of ground motions in southeastern Canada is approximately equivalent to that of the central United States (Benz et al., 1997), the liquefactioninduced ground failure distance for the 1925 earthquake is evidence that earthquakes in the Wabash River Valley area are capable of inducing liquefaction at distances greater than Ambraseys' (1988) limiting epicentral distance curve.
Along the bottom of Figure 3 are the maximum epicentral distances to liquefaction used by Pond and Martin (1997) for backcalculating the magnitudes of the four prehistorical earthquakes. The dashed lower curve shown in Figure 3 parallels Ambraseys' (1988) limiting distance curve and passes through the moment magnitudes versus maximum distances to liquefaction for the 1895 and 1925 earthquakes. Based on the intercepts of the maximum epicentral distances to Pond and Martin's (1997) curve, Ambraseys' (1988) limiting distance curve, and the dashed curve, the moment magnitudes for the four events can be variously estimated, as shown in Table 1.
Peak ground accelerations recorded for the southwestern Indiana M 4.5 earthquake of 18 June 2002 also indicated that the magnitudes of the prehistorical earthquakes could be overestimated. Pond and Martin (1997) concluded that for sites in the Wabash Valley seismic zone with soil overburden between 15 and 75 m deep and shearwave velocities between 180 and 360 m/s, the peak ground acceleration (PGA) can be described by the relationship (1) where PGA is the peak ground acceleration in units of gravity and R is the hypocentral distance in km. However, 12 of the 34 peak ground accelerations recorded for the 18 June 2002 M 4.5 earthquake exceeded those predicted by Equation 1 for an M 6 event (Street et al., 2004). These results indicate that earthquakes somewhat smaller than those suggested by Pond and Martin (1997) can generate the groundmotion amplitudes (but not necessarily the durations) needed for inducing liquefaction in the sediments of the Wabash River Valley.
DISCUSSION AND CONCLUSIONS
By using a moment magnitude for the 31 October 1895 earthquake closer to that suggested by most investigators, the maximum distance to surface evidence of liquefaction is shifted nearer Ambraseys' (1988) limiting distance curve. This shift is also physically more plausible for an area of relatively low seismicwave attenuation and where the sediments have a high susceptibility to liquefaction. Consequently, we suggest that the moment magnitudes for the four prehistorical earthquakes shown in Figure 3 be approximated by the intercepts of their maximum distances to liquefaction with Ambraseys' (1988) limiting distance curve.
Lowering the moment magnitudes of the four prehistorical earthquakes by 0.6 units will change the perceived seismic hazard for critical infrastructure in the southern Indiana and Illinois area. The lower magnitudes will also require adjusting the magnituderecurrence calculations and reassessing the maximum magnitude earthquake for the area as suggested by Wheeler and Cramer (2002).
Acknowledgments
We are grateful for the constructive comments of Drs. Pradeep Talwani, Roy Van Arsdale, and Martin Chapman. Their assistance resulted in significant improvements to the manuscript. We also wish to thank Meg Smath and Collie Rulo of the Kentucky Geological Survey for their editorial comments and assistance in figure preparation, respectively.
Footnotes
Kentucky Geological Survey
Illinois State Geological Survey
University of Kentucky