Identifying the Rupture Plane of the 2001 Nisqually, Washington, Earthquake by Honn Kao, Kelin Wang, Rong-Yuh Chen, Ikuko Wada, Jiangheng He, and Stephen D. Malone Abstract The 2001 Nisqually earthquake occurred within the subducting Juan de Fuca plate. Previous seismic and geodetic studies could not confidently identify its actual fault plane from the two nodal planes. In this study, we apply the recently de- veloped source-scanning algorithm to local seismic waveforms and show unambigu- ously that the steeply east-dipping plane is the rupture plane. The rupture began near the bottom of the subducting crust and propagated downward into the subducting uppermost mantle. If intraslab earthquakes are assumed to be due to dehydration em- brittlement, the source dimension is unlikely to grow any larger because the warm thermal state of the subducting Juan de Fuca plate limits dehydration to a shallow depth below the slab surface. Numerical modeling of the thermal structure indicates that dehydration embrittlement can only take place in the top 10 km of the subducting mantle, implying that the maximum size of an intraslab earthquake in northern Cas- cadia would be M w ∼7 or less. Introduction Earthquakes that occur within the young subducting Juan de Fuca plate in the depth range of 40–70 km are responsible for most of the damaging earthquakes in the Pa- cific Northwest over the past century. The most recent exam- ple is the 28 February 2001 Nisqually earthquake (M w 6.8, Fig. 1), which caused widespread, mainly nonstructural, damage to buildings and bridges in the greater Seattle area (Fig. 1; Malone, 2001). Other events of this type include the 1949 Olympia (M w 6.8) and the 1965 Seattle–Tacoma (M w 6.6) earthquakes (Baker and Langston, 1987; Ichinose et al., 2004; Ichinose et al., 2006). It is widely accepted that these intraslab earthquakes are associated with dehydra- tion of the subducting slab (Kirby et al., 1996; Peacock and Wang, 1999). Dehydration of hydrous minerals may cause pore fluid pressure to rise to near-lithostatic values to facil- itate brittle failure along preexisting faults, a process referred to as dehydration embrittlement. Without elevated fluid pres- sure, rock failure under the confining pressure of the slab environment is expected to be aseismic, either as cataclastic flow if the temperature is low or as plastic creep if the tem- perature is high (Stöckhert and Renner, 1998). Most of the bonded H 2 O in the oceanic lithosphere prior to subduction is in the highly fractured basaltic-gabbroic crust. It is thus expected that intraslab earthquakes occur mostly in the dehydrating subducting crust. If the subducting mantle is not significantly involved, the size of earthquake rupture may be limited by the thickness (∼7 km) of the sub- ducting crust. However, if H 2 O percolates to the peridotitic uppermost oceanic mantle to cause serpentinization prior to subduction, breakdown of serpentine minerals during slab subduction may also lead to dehydration embrittlement (Hacker et al., 2003; Yamasaki and Seno, 2003), allowing fault rupture to propagate into the subducting oceanic man- tle. In this case, the rupture dimension of an intraslab earth- quake may be controlled by the distribution of serpentine in the subducting mantle. The possibility of significant mantle involvement in the rupture processes of large intraslab earth- quakes in northern Cascadia is debated. The most direct ap- proach to resolving the issue of mantle involvement is to study rupture geometry of past large intraslab earthquakes, among which the 2001 Nisqually earthquake is the most well recorded. The hypocenter of the Nisqually earthquake is located at a depth of 54 km, just above the Moho discontinuity of the subducting plate (Fig. 1; Malone, 2001). Focal mechanism solutions show one nodal plane steeply dipping to the east and the other shallowly dipping to the west (Fig. 1). If the rupture were on the shallowly west-dipping nodal plane, the entire rupture could barely fit in the slab crust (Creager et al., 2003), whereas if it were on the steeply east-dipping plane, it must have extended into the slab mantle (Fig. 1). Inversion of strong-motion waveforms shows that, if the steeply east- dipping plane is taken as the rupture plane, the rupture must have propagated downward and that the strongest moment 1546 Bulletin of the Seismological Society of America, Vol. 98, No. 3, pp. 1546–1558, June 2008, doi: 10.1785/0120070160