Source Investigation and Comparison of the 1939, 1946, 1949 and 1965 Earthquakes, Cascadia Subduction Zone, Western Washington

Home , 1949 Olympia earthquake

Ó Birkha¨user Verlag, Basel, 2007 Pure appl. geophys. 164 (2007) 1905–1919 0033–4553/07/101905–15 Pure and Applied Geophysics DOI 10.1007/s00024-007-0255-y

1,3 1 1 2 KATY R. WIEST, DIANE I. DOSER, AARON A. VELASCO, and JAMES ZOLLWEG

Abstract—Over the past 65 years intraslab earthquakes have caused the most signiﬁcant damage in the western Washington region. This study examines regional and teleseismic seismograms for four historic, suspected intraslab events of M > 5.5 occurring within the Cascadia Subduction zone in 1939 (South Seattle), 1946 (Puget Sound), 1949 (Olympia) and 1965 (Sea-Tac) to better determine the source locations, mechanisms and rupture histories of these events. Our study is aided by digital seismograms of post-1990 intraslab events with well-determined focal depths and focal mechanisms that were recorded in the same locations as the historic events. Thus the recent events were used as empirical Greens functions to study the historic events. Our results suggest that the 1946 earthquake is not an intraslab event, that the 1939 event closely resembles the 1965 event, and that the 1949 event is similar to the 2001 Nisqually earthquake, although the 1949 event appears to have ruptured toward the south, causing signiﬁcantly more damage than the Nisqually event. These results suggest that earthquakes periodically rupture along the same or similarly oriented faults within the subducting slab.

Key words: Historic earthquakes, Cascadia, Washington, Greens function analysis.

Introduction

The Cascadia Subduction Zone (CSZ) (Fig. 1) is a highly complex, geologically diverse, and potentially hazardous region due to the collision of the Juan de Fuca, Gorda, and Explorer plates with North America. The CSZ extends from Cape Mendocino, California north to northern Vancouver Island. The subducting lithosphere of this region is young (<50 Ma), thin, and buoyant (KIRBY and WANG, 2002), creating a tectonically complex environment with the buildup of a thick accretionary prism in the forearc region and active compression in the backarc region.

1 Department of Geological Sciences, University of Texas at El Paso, El Paso, TX 79968-0555. E-mail: [email protected] 2 Department of Geosciences, Boise State University, Mail Stop 1535, 1910 University Drive, Boise, ID 83725-1535, U.S.A. 3 Chevron North America Exploration and Production, 1500 Louisiana, Houston, TX 77002, U.S.A. 1906 K. R. Wiest et al. Pure appl. geophys.,

Figure 1 Map of the Cascadia subduction zone showing major plate boundaries. Squares denote locations of Seattle and Olympia, Washington. PS is Puget Sound, GS is southern Georgia Strait. The box outlined in gray is the study area shown in Figure 2.

Seismicity within the CSZ occurs within the crust of North America, the subducting slab, and along the plate interface. Events along the interface have the potential to be very large, but are believed to occur only every 500 years (ATWATER et al., 2005), with the last known event occurring in 1700 (ATWATER and HEMPHILL- HALEY, 1997). Events within the crust have the capability to cause extensive damage due to their shallow depths beneath sedimentary basins, but appear to occur even less frequently than interface events (e.g., LUDWIN et al., 2005). Historically the greatest damage in the CSZ has been caused by intraslab events that appear to have recurrence intervals of 30 to 50 years. There appear to be two regions within the subducting Juan de Fuca plate where moderate to large (M > 5:0) intraslab events concentrate at depths of 40 to 60 km. One area is located beneath southern Puget Sound, and the other is located beneath the southern Georgia Strait near edges of a bend in the subducted slab (CASSIDY and ELLIS, 1993). There has been a notable lack of moderate to large intraslab Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1907

Figure 2 Location of historic earthquakes and recent events used in this study.

earthquakes beneath much of Oregon (e.g., MA et al., 1996); leading WONG (2005) to suggest this region may not be capable of producing large intraslab earthquakes. The occurrence of the 2001 Nisqually earthquake which was well recorded digitally at regional and teleseismic distances, gives us a unique opportunity to compare its seismograms to those of older, suspected intraslab earthquakes. We examine regional and teleseismic seismograms for four historic, suspected intraslab events of M > 5:5 occurring within the Cascadia Subduction zone in 1939 (South Seattle), 1946 (Puget Sound), 1949 (Olympia) and 1965 (Sea-Tac) to better determine the source locations, mechanisms and rupture histories of these events. A better understanding of the depth range and rupture extent of these older events will aid in determining the physical processes that control slab rupture and improve models for hazards expected in future similar type events.

Previous Studies

The earliest reported M > 5:5 intraslab event is the 1939 (M 53=4; GUTEN- BERG and RICHTER, 1954) South Puget Sound earthquake (Fig. 2, Table 1). Its seismograms have not been previously studied. The Puget Sound earthquake of 1946 (M 53=4; GUTENBERG and RICHTER, 1954) has been reported in some catalogs as an intraslab event, however STOVER and COFFMAN (1993) have proposed it was actually a crustal event (18-km depth) (Fig. 2, Table 1). The 1949 Olympia earthquake appears to have been the largest intraslab event (mb ¼ 6:9; ABE 1981) that has occurred in the region in the past 100 years (Table 1). BAKER and LANGSTON (1987) used long-period teleseismic body-wave seismograms and strong motion records to examine this earthquake. Their results suggest the event occurred at a depth of 54 km with a left-lateral strike-slip focal mechanism 1908 K. R. Wiest et al. Pure appl. geophys.,

Table 1 Earthquake information

Date Time Latitude Longitude Magnitude* Depth*+

3 11/13/1939 0745 47.5 122:554 (S, GR) [63] 3 02/15/1946 0317 47.4 122:754 (S, GR) 18 SC [26] 04/13/1949 1955 47.17 122:62 6.9 (b, ABE), 6.8 (w, BL, I06) 54 BL [55] 04/29/1965 1528 47.38 122:31 6.5 (S,U), 6.6 (w, I04) 63 LB [63] 07/03/1999 0143 47.08 123:46 5.7 (w, I03) 40 I03 02/28/2001 1854 47.14 122:72 6.7 (w, I04) 56 UW [55]

* GR = GUTENBERG and RICHTER (1965), ABE = ABE (1981), BL = BAKER and LANGSTON (1987), I03 = ICHINOSE et al. (2003), I04 = ICHINOSE et al. (2004), I06 = ICHINOSE et al. (2006), LB = LANGSTON and BLUM (1977), SC = STOVER and COFFMAN (1993), UW = University of Washington, S = surface wave magnitude, b = body-wave magnitude, w = moment-magnitude, U = unknown magnitude, +brackets indicate focal depth estimated in this study.

and eastward rupture propagation. They determined a rupture duration of 40 km and a moment-magnitude of 6.8. Further studies of seismograms and ﬁrst motion data by ICHINOSE et al. (2006) suggest the 1949 event occurred on a steeply dipping normal fault with rupture to the south in at least two subevents. LANGSTON and BLUM (1977) modelled the source parameters of the 1965 Sea- Tac earthquake through inversion of teleseismic body waveforms to obtain a focal depth of 63 km, normal fault mechanism, and Mw of 6.7 (Fig. 2, Table 1). ICHINOSE et al. (2004) inverted teleseismic data for this earthquake using a segmented, multi-fault model. The model that best matched the waveforms was composed of two small asperities with dimensions of 12 and 16 km2 and maximum slips of 2 and 2.8 m, respectively. This model gave an Mw of 6.6. ICHINOSE et al. (2004) produced a shake-map for the Sea-Tac earthquake that gave a simulated pattern of shaking that was 2 times greater than that of the Nisqually earthquake, primarily due to the proximity of the Sea-Tac event to the Seattle and Tacoma sedimentary basins.

Data and Analysis Techniques

Over 150 historic paper copies and original seismograms recorded at regional and teleseismic distances for the four Puget Sound earthquakes were collected during this study. We also scanned 50+ seismograms from several U.S. archives. The scanned seismograms could be readily digitized using SeisDig (BROMIRSKI and CHUANG, 2003), an interactive digitizing tool. Analog paper records were digitized manually, due to a lack of a large bed scanner at the University of Texas at El Paso. Digital Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1909 data for two recent intraslab events, the 1999 (Mw ¼ 5:8) Satsop and 2001 (Mw ¼ 6:7) Nisqually earthquakes (Fig. 2, Table 1), were acquired from the IRIS (Incorporated Research Institutions in Seismology) electronic archives. These events were selected because they occurred in the vicinity of the four historic events of interest and were of large enough magnitude to be well recorded at regional and teleseismic distances. We collected digital seismograms for recent events only at stations that had recorded at least one historic event or were located close (<35 km) to a station that had recorded a historic event. This selection allowed us to assume similar path effects when carrying out the Greens function analysis. All seismograms were converted to Seismic Analysis Code (SAC2000) format (GOLDSTEIN, 2004) for the analysis process. Seismograms in this study were analyzed using the relative source time function technique (RSTF) similar to that described by VELASCO et al. (1994), but using a different deconvolution process. This technique requires the seismograms of two earthquakes occurring at similar locations and depths, with a similar focal mechanism recorded at the same station. The smaller event, the empirical Greens function (EGF), is assumed to be a point source in both time and space, but also needs to be large enough to ensure a satisfactory signal-to-noise level at the recording distances. The technique isolates the source properties of the larger event by examining the ratio of the larger (mainshock) to smaller event, usually through deconvolution in the frequency domain. The directivity and complexity of the larger event is revealed through any observed azimuthal variations in the duration and amplitude of the RSTF. Because the historic and recent events were recorded on distinctly different seismographs, the first step in our process was to remove the instrument response from both seismograms. Next the seismograms were windowed around the phases of interest, the mean was removed and a taper was applied. Then we used the iterative, time-domain deconvolution technique of LIGORRIA and AMMON (1999) to determine the RSTF. The iterative technique first strips the largest amplitude Gaussian pulse from the seismogram and then models successively higher frequency details of the seismogram. The Gaussian function A(t) can be described by

2 AðtÞ¼eðt=aÞ ; where t is time and a is the Gaussian width factor (in seconds). The resulting RSTF is a summation of the stripped Gaussian pulses. Gaussian width factors of 0.02, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5 and 4.0 seconds were used in our analysis. Each historical event was considered separately with iterative deconvolutions performed for all possible EGF-mainshock pairs. In cases where the reported magnitudes of the two events were similar, two deconvolutions were performed; ﬁrst using one event as the EGF, and the second using the other event as the EGF. Then 1910 K. R. Wiest et al. Pure appl. geophys.,

Figure 3 Illustration of how convolution of Relative Source Time Function (RSTF) for the 1949 Olympia earthquake with Empirical Greens Function (EGF) (seismogram of 2001 Nisqually earthquake) for College, Alaska (solid line) replicates original 1949 seismogram (dashed line). The diﬀerence between these waveforms is shown by grey dotted line. See text for details.

the stability of the resulting RSTFs was evaluated to determine which event had the larger moment. The goodness of fit of the deconvolution is based on a least-squares (RMS) fit to signal power. An illustration of how well the convolution process replicates the source characteristics is shown in Figure 3. In this case we have deconvolved the north- south component seismogram recorded at College, Alaska (COLA) (see Table 2 for station information) for the 2001 Nisqually earthquake from the north-south component seismogram for the 1949 Olympia earthquake recorded at College, Alaska (COL) (dashed line, Fig. 3) to produce an RSTF for College. Then the RSTF was convolved with the 2001 seismogram to produce the result shown by the solid line in Figure 3. The difference between the original and the seismogram reconstructed from the RSTF is indicated by the grey dotted line (Fig. 3). Note the excellent fit of the original to the reconstructed seismogram, especially for the first 20 seconds of the waveform. The increase in misfit over time is due to the difficulty in resolving the shape of the tail of the RSTF. In addition to computing RSTFs, we made a rough estimate of the focal depth for each historic event using the IASPEI91 earth model (KENNETT and ENGDAHL, 1991) to predict P-pP times. For all events the pP and P arrivals were read from vertical or radial (naturally rotated) components of the broadest band seismograph available at that location. Table 3 provides pP-P times and focal depths estimated from individual stations for the oldest two earthquakes. Average focal depths for the other historic and recent events are comparable to those obtained in previous waveform modelling studies (Table 1). Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1911

Table 2 Station information

Station Code(s) Name Latitude Longitude Ave. Distance Ave. Azimuth Years

COL College, AK 64.9 147:7933 32 330 1949–1995 COLA College, AK 64.8738 147:8511 1996- FLO Florissant, MO 38.8017 90:37 25 99 1928–1971 LRA Little Rock, AR 34.7783 92:3517 26 109 1931–1967 UALR Univ. Arkansas 34.7753 92:3436 1999- MAT Matsushiro, Japan 36.54517 138.2089 70 302 1947- MAJO Matsushiro, Japan 36.5425 138.2073 1977- OTT Ottawa, Canada 45.3942 75:7167 32 76 1906- GAC Glen Almond 45.7032 75:4783 1982- PAS Pasadena, CA 34.1483 118:1717 14 165 1927- SLM St. Louis, MO 38.6361 90:2361 25 99 1909- WES Weston, MA 42.3847 71:3221 36 79 1930- HRV Harvard, MA 42.506 71:558 1988-

Results

November 13, 1939; South Puget Sound Event We were able to digitize data from ﬁve stations for this earthquake. Most seismograms were recorded by short period (Ts <5 sec) seismometers. Figure 4 shows the similarities between seismograms of the 1939 and 1965 earthquakes as recorded at stations FLO and OTT (see Table 2 for station information), suggesting the events had comparable focal mechanisms and depths. Depth phases observed for the 1939 event suggest a focal depth of 63 km. Stable deconvolutions for the 1939 event

Table 3 Focal Depth Estimates for 1939 and 1946 Earthquakes

Date Station pP-P time (seconds) Focal depth* (km)

11/13/1939 COL 16 62 FLO 15 62 CGM+ 17 74 LRA 18 68 PAS 8 50 SLM 15 62 2/15/1946 FLO 8 27 PAS 6 25 SLM 8 27

* + Based on IASPEI91 (KENNETT and ENGDAHL, 1991) velocity model. Cape Girardeau, Missouri distance =26°. 1912 K. R. Wiest et al. Pure appl. geophys.,

Figure 4 Comparison of seismograms for the 1939 South Seattle and 1965 Sea-Tac earthquakes recorded at: (a) Florissant (FLO) (east-west component) and (b) Ottawa (OTT) (vertical –component). were only obtained using the north-south component of the 1965 event recorded at OTT and the east-west component of the 1946 event recorded at FLO as EGFs (Fig. 5). The RSTFs suggest a total duration of 20 seconds for rupture in 1939, but with both stations located at nearly the same azimuth (Table 2), little can be said about rupture directivity in 1939.

February 15, 1946; Puget Sound Most deconvolutions using the 1946 event as the mainshock were unstable and gave high misﬁts. The best deconvolution only had a ﬁt of 27% (Fig. 6). This is Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1913

Figure 5 Azimuthal plot showing best RSTFs for the 1939 South Seattle earthquake obtained using the iterative deconvolution method of LIGORRIA and AMMON (1999). OTT was obtained using the north-south component of the 1965 Sea-Tac event as the EGF. FLO was obtained using the east-west component of the 1946 Puget Sound earthquake as an EGF. Percent fit (%fit) of the deconvolution (based on RMS fit of signal power) is also indicated.

likely because the 1946 event appears to be the shallowest and smallest magnitude event studied, and the EGF is inadequate for the deconvolution procedure. Depth phases from PAS (Fig. 7) and two other seismograph stations (Table 3) suggest an average focal depth of 26 km, which is shallower than all other events studied (Table 1).

Figure 6 Plot of best RSTF for 1946 Puget Sound earthquake using east- west component of the 1965 Sea-Tac earthquake as the EGF. 1914 K. R. Wiest et al. Pure appl. geophys.,

Figure 7 Pasadena (PAS) vertical-component seismogram of the 1946 Puget Sound earthquake. P and pP arrivals are indicated. These arrival times give an estimated hypocentral depth of 25 km.

April 13, 1949; Olympia The Olympia earthquake is the largest magnitude intraslab event to have occurred in past century. Its location is similar to that of the 2001 Nisqually earthquake. Comparisons between low-pass ﬁltered seismograms of the 1949 and 2001 events recorded at COL and PAS (Fig. 8) indicate the events were very similar. RSTFs for the 1949 Olympia earthquake, all obtained by using the 2001 Nisqually earthquake as the EGF, are shown in Figure 9. Note that the RSTF duration at PAS is 12 sec, while those at COL and SLM have durations of 22 and 20 secs. This suggests rupture toward the south, consistent with the waveform modeling of ICHINOSE et al. (2006) and the observations of BAKUN et al. (2002) who located the intensity epicenter 55 km south of the instrumental epicenter. There is some suggestion of source complexity at PAS and SLM, with the SLM RSTF exhibiting the best evidence for a double event, as suggested by ICHINOSE et al. (2006).

April 29, 1965; Sea-Tac As noted previously, the 1965 Sea-Tac event waveforms appear similar to the 1939 South Seattle event (Fig. 4). RSTFs for the Sea-Tac event are shown in Figure 10. The 1999 Satsop event was used as the EGF for MAT and OTT, while the 1946 earthquake was used as the EGF for FLO. Note that all RSTFs show two distinct pulses of energy release with rupture duration of 20 sec and little evidence for variations in pulse width that would suggest directivity. This is consistent with the two small asperities imaged by ICHINOSE et al. (2004). Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1915

Figure 8 Comparison of seismograms of the 1949 Olympia earthquake to the 2001 Nisqually earthquake as recorded at: (a) College (COL/COLA) (north-south component) and (b) Pasadena (vertical-component). All seismograms have been low-pass ﬁltered at 0.2 Hz.

February 28, 2001; Nisqually In order to assess the quality of our results and validate the use of historical seismograms in the EGF technique, we also analyzed seismograms of the 2001 Nisqually earthquake. We ﬁrst used the 1949 Olympia event as the EGF at stations COL/COLA and PAS, although we realized that the 1949 event is likely larger in moment than the 2001 event. Not surprisingly, we obtained relatively unstable RSTFs for these event pairs (Fig. 11). When we used the 1965 Sea-Tac event as the EGF our RSTFs for stations MAT/MAJO and HRV/WES were simpler and more stable (Fig. 11). 1916 K. R. Wiest et al. Pure appl. geophys.,

Figure 9 Plot of best RSTFs for the 1949 Olympia earthquake. Focal mechanism shown is from ICHINOSE et al. (2006). All three RSTFs used the 2001 Nisqually earthquake as the EGF.

ICHINOSE et al. (2004) suggest that the 2001 event was composed of at least two subevents, with the second subevent being 1.5 to 2 times larger in amplitude. Both nodal planes of the event produced slip models that ﬁt the data similarly (ICHINOSE et al., 2004), however, if the steeply dipping nodal plane is considered the fault plane, slip appears to have propagated downward and bilaterally from the hypocenter (ICHINOSE et al., 2004), with slightly more slip toward the south. Double-diﬀerence relocation studies of the Nisqually mainshock and its aftershocks (CREAGER et al., 2003), however, suggest northward rupture along the strike of the slab, with the near- horizontal nodal plane favored as the fault plane. Our RSTFs at COLA, HRV and PAS suggest at least two sources. Comparison of RSTF widths suggests rupture toward MAT and COL (northwest), which would be opposite the direction suggested by INCHINOSE et al. (2004), but consistent with the work of CREAGER et al. (2003).

Figure 10 Plot of best RSTFs for the 1965 Sea-Tac earthquake. Focal mechanism shown is from ICHINOSE et al. (2004). The 1999 Satsop event was used as the EGF at Matsushiro (MAT) and Ottawa. The 1946 Puget Sound earthquake was used as the EGF at Florissant. Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1917

Figure 11 RSTFs for the 2001 Nisqually earthquake. Focal mechanism shown is from ICHINOSE et al. (2004). The 1965 Sea-Tac earthquake was used as the EGF at Matsushiro and Harvard (HRV). The 1949 Olympia earthquake was used as the EGF at College and Pasadena.

Discussion and Conclusions

We have used a limited set of historical seismograms along with recordings of two recent (post-1990) earthquakes to study the source processes of four suspected intraslab earthquakes within the western Washington region with the EGF/RSTF technique. Our studies suggest the intraslab events can be arranged by moment in the following order: 1946 (lowest), 1999, 1939, 1965, 2001 and 1949 (highest). Source complexities observed for the 1965, 2001 and 1949 events are in good agreement with other studies involving modelling of teleseismic, regional, and local seismograms. Many researchers (e.g., KIRBY and WANG, 2002; PRESTON et al., 2003) suggest that metamorphic dehydration reactions may trigger intraslab earthquakes. Faults created at mid-ocean ridges and fracture zones may be reactivated in downgoing slabs by dehydration embrittlement. Volume changes associated with densifying the crust could also play an important role in internal slab deformation. The complexities of these historic earthquakes, even with limited station coverage, highlight the difficulty of interpreting fault properties. However, our results suggest that the 1946 earthquake is not an intraslab event, and that the 1939 event closely resembles the 1965 event based on waveform similarities (Fig. 4) and our deconvolution results. Meanwhile, the 1949 event is similar to the 2001 Nisqually earthquake, although the 1949 event appears to have ruptured toward the south, causing significantly more damage than the Nisqually event. Since these historic events appear to occur in similar regions and at similar depths as more recent events, it suggests that reactivation of faults could be occurring. Our results indicate that important information can be gleaned from the limited historical data if the historical seismograms can be compared with recent, well recorded events occurring in similar locations. The observed differences in source parameters between historic and recent Cascadia intraslab earthquakes will aid in 1918 K. R. Wiest et al. Pure appl. geophys., quantifying the variation in rupture scenarios and resulting differences in strong ground motion that may be expected from future intraslab events within the region.

Acknowledgements

C. Flores assisted us with locating records from the Pasadena archives and P. Murray and T. Theiner helped with the digitization of seismograms. We thank G. Inchinose for providing us with preprints of papers and reports and an anonymous reviewer for helpful comments. This research was supported by the U.S. Geological Survey under grant 02HQGR0106. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the oﬃcial policies, either expressed or implied, of the U.S. Government.

REFERENCES

ABE, K. (1981), Magnitudes of large shallow earthquakes from 1904 to 1980, Phys. Earth Planet. Inter. 1, 72–92. ATWATER, B.F., MUSUMI-ROKKAKU, S., SATAKE, K., TSUJI, Y., UEDA, K., and YAMAGUCHI, D.K. (2005), The Orphan Tsunami of 1700, U.S. Geol. Surv. Prof. Paper 1707, 133 pp. Baker, G.E. and Langston, C.A. (1987), Source parameters of the 1949 magnitude 7.1 South Puget Sound, Washington, earthquake as determined from long-period body waves and strong ground motion, Bull. Seismol. Soc. Amer. 77, 1530–1557. BROMIRSKI, P.D. and CHUANG, S. (2003), SeisDig Software to Digitize Scanned Analog Seismogram Images: User’s Manual, Scripps Inst. Of Oceanog. Tech. Rept., Scripps Inst. Of Oceanog., Integrative Oceanog. Div., UC San Diego, 28 pp. CASSIDY, J.F. and ELLIS, R.M. (1993), S-wave velocity structure of the northern Cascadia subduction zone, J. Geophys. Res. 98, 4407–4421. CREAGER, K.C., XU, Q., and PRESTON, L.A. (2003), The 2001 Mw6:8 Nisqually intraslab earthquake and its aftershocks, GSA Abstr. with Progs. 35(6), 309. GOLDSTEIN, P. (2004), Introduction to SAC 2000, www.iris.edu/manuals/sac/SAC_Manuals /Introduc- tion.html GUTENBERG, B. and RICHTER, C.F., Seismicity of the Earth and Associated Phenomena, (Hafner, New York, 1965) 245 pp. ICHINOSE, G.A., THIO, H.K., and SOMERVILLE, P.G. (2006), Moment tensor and rupture model for the 1949 Olympia, Washington, earthquake and scaling relations for Cascadia and global intraslab earthquakes, Bull. Seismol. Soc. Amer. 96, 1029–1037. ICHINOSE, G.A., THIO, H.K., and SOMERVILLE, P.G. (2004), Rupture process and near-source shaking of the 1965 Seattle-Tacoma and 2001 Nisqually, intraslab earthquakes, Geophys. Res. Lett. 31, L10604, doi:10.1029/2004GL019668. ICHINOSE, G.A., THIO, H.K., and SOMERVILLE, P.G. (2003), Source characteristics of modern and historical in-slab earthquakes applicable to strong ground motion prediction, U.S. Geol. Surv. NEHRP Final Tech. Rept., Award 02HQGR0018. KENNETT, B.L.N. and ENGDAHL, E.R. (1991), Traveltimes for global earthquake locations and phase identiﬁcation, Geophys. J. Roy. Astr. Soc. 105, 429–465. KIRBY, S.H. and WANG, K. (2002), Introduction to a global systems approach to Cascadia slab processes and associated earthquake hazards, The Cascadia Subduction Zone and Related Subduction Systems (Kirby, S.H., Wang, K., and Dunlop, S.G.) eds. U.S. Geol. Surv. Open File Rept. 02-0328, 5–9. Vol. 164, 2007 Historic Earthquakes of the Cascadia Subduction Zone 1919

LANGSTON, C.A. and BLUM, D.E. (1977), The April 29, 1965 Puget Sound earthquake and the crustal and upper mantle structure of western Washington, Bull. Seismol. Soc. Amer. 67, 693–711. LIGORRIA, J.P. and AMMON, C.J. (1999), Iterative deconvolution and receiver function estimation, Bull. Seismol. Soc. Amer. 89, 1395–1400. LUDWIN, R.S., THRUSH, C.P, JAMES, K., BUERGE, D., JONINENTZ-TRISLER, C., RASMUSSEN, J., TROOST, K., and DE LOS ANGELES, A. (2005), Serpent spirit-power stories along the Seattle fault, Seismol. Res. Lett. 76, 426–431. MA, L., CROSSON, R., and LUDWIN, R. (1996), Western Washington focal mechanisms and their relationship to regional and tectonic stress, U.S. Geol. Surv. Prof. Paper 1560, 257–284. PRESTON, L.A., CREAGER, K.C., CROSSON, R.S., BROCHER, T.M., and TREHU, A.M. (2003), Intraslab earthquakes: dehydration of the Cascadia slab, Science 302, 1197–1200. STOVER, C.W. and COFFMAN, J.L. (1993), Seismicity of the United States 1568-1989 (revised), U.S. Geol. Surv. Prof. Pap. 1527, 418 pp. VELASCO, A.A., AMMON, C.J., and LAY, T. (1994), Empirical Green function deconvolution of broadband surface waves: Rupture directivity of the 1992 Landers, California (M = 7.3) earthquake, Bull. Seismol. Soc. Amer. 84, 735–750. WONG, I. G. (2005), Low potential for large intraslab earthquakes in the central Cascadia subduction zone, Bull. Seismol. Soc. Amer. 95, 1880–1902.

(Received August 31, 2006, accepted February 5, 2007) Published Online First: July 4, 2007

To access this journal online: www.birkhauser.ch/pageoph