Subduction Zone Processes

Subduction Zone Processes

Aftershock studies of the 2004 M 9.2 Sumatra-Andaman Islands and 2005 M 8.7 Nias earthquakes

The seismic and tsunami hazard posed by great subduction zone earthquakes has long been recognized, but in light of the Mw 9 Sumatra-Andaman Islands earthquake, the resulting tsunami, infrastructure damage and huge loss of life, it is clear that the hazard posed by such events is generally underestimated. The December 26, 2004 Mw 9.0 earthquake is one of the four largest in the instrumental record, comparable in many respects to the 1960 Mw 9.5 Chilean and 1964 Mw 9.2 Alaska subduction megathrust events. The main shock and aftershock sequence essentially illuminated the entire Andaman micro-plate along the Andaman subduction zone and back-arc spreading center, and the earthquake likely triggered the March 28, 2005 Mw 8.7 earthquake along the adjacent Sunda Trench.

The 2004 earthquake initiated along the Andaman subduction zone, north of the last great Sumatra earthquake along the Sunda Trench in 1861. During the Nias earthquake, a portion of the 1861 rupture subsequently failed. The boundary between the 2004 and 2005 ruptures broadly coincides with local trench rotation and the southern edge of the Andaman microplate, which suggests structural control on fault segmentation. Aftershock relocations of the 2004 and 2005 earthquakes using the EHB method show little overlap, and the sharp boundary between the series locates near the 2002 Mw 7.3 Northern Sumatra earthquake. We posit that these features represent the southern extent of the stable Andaman microplate, ~50-100 km northwest of what was previously reported. I have additionally used broadband analyses to show that the 2002 earthquake was a bilateral rupture.  The resulting rupture pattern for the 2002 event was used to model Coulomb stress changes near the 2004 hypocenter to assess stress interactions along adjacent fault segments [see DeShon et al., 2005].

Combined, the 2004 and 2005 sequences have generated over 5000 teleseismically-recorded earthquakes along 1700 km of the Andaman and Sunda subduction systems. The seismogenic portions of the Andaman and Sunda Trenches lie within an oceanic environment, which ultimately limits the amount of local seismic data that will become available and emphasizes the need for state-of-the-art analysis of the available teleseismic data. Developing and applying new teleseismic techniques to provide earthquake locations and velocity structure resolution at a scale comparable to that of regional network studies will help advance our understanding of the complex ruptures of the Sumatra great earthquakes.  To date, aftershocks through September 2005 have been relocated using EHB methods (pictured below), and while errors remain on the order of 10s of km, interesting patterns in both the seismicity and focal mechanism solutions have been revealed [Engdahl et al., 2007 in press].

Double seismic zone identification using teleseismic data. Understanding the seismicity and structure of subducting slabs has important geodynamic implications for volcanic and mantle processes. One interesting, but poorly understood, feature of some subduction zones is the development of two distinct layers of seismicity within the oceanic lithosphere at intermediate depths (~50-300 km).  Such observations are aptly termed Double Wadati-Benioff Zones or Double Seismic Zones (DSZ) and have been attributed to petrologic reactions that trigger earthquakes via dehydration embrittlement. Identification of DSZ has typically been limited to areas where dense local networks cover the subduction zone.  We are developing a new technique that strictly uses globally recorded teleseismic waveforms to more precisely determine the location of subduction zone earthquakes. By cross-correlating the depth phase relative timing (pP-P, sP-P, and sS-S) we improve differential travel time precision and hence the relative depths of hypocenters.  The technique facilitates identification of DSZ, and we are in the initial stages of a global review of subduction zones.  We are currently focusing on Alaska (H. Zhang), Java (M. Brudzinski), northern Chile (H. DeShon), and the Marianas Trench (H. DeShon), where either teleseismic data hints at DBZs or local data has shown the existance of a DBZ. We have found a significant variability between subduction zones of the characteristic upper and lower plane stress pattern, confirming that not all DSZ exhibit downdip compression overlying downdip compression.  The ability to identify DBZ away from local networks will be a key step in determining the overall prevalence and regularity of this feature on a global basis, which is necessary to understand the conditions, both seismologic and petrologic, in the evolution of subducting slabs in general.  Principle Investigators: Cliff Thurber and Bob Engdahl.

Seismogenic zone structure along the Middle America subduction zone, Costa Rica (CRSEIZE)

Most large (Mw > 7.0) and great (Mw>8.0) underthrusting earthquakes nucleate along a shallow region of unstable frictional stability on or near the subducting plate interface termed the seismogenic zone. My dissertation, entitled Seismogenic zone structure along the Middle America subduction zone, Costa Rica, explored spatial and temporal variability in microseismicity within the seismogenic zone of the Middle America Trench (MAT) offshore Costa Rica and investigated processes controlling the loci of shallow (<100 km depth) subduction zone seismicity.  The research was a primary component of the Costa Rica Seismogenic Zone Experiment, a series of interdisciplinary geodetic, seismic, and fluid flux experiments undertaken from September 1999-June 2001 to better understand subduction zone behavior near the Osa (southern) and Nicoya (northern) Peninsulas of Costa Rica.  We utilized a range of absolute and relative earthquake location and local earthquake tomography techniques to relocate microseismicity and detect seismic velocity variations (VP, VS, & VP/VS) along the subduction thrust in this region. The CRSEIZE seismological studies constrain the updip and downdip limits and along-strike variability of microseismicity along Costa Rica.  This data is essential for comparison with geodetic measurements, thermal modeling, and hydrologic modeling to better understand the process controlling seismicity along subduction megathrusts. Combined, the studies provide the highest resolution images of seismicity and seismic velocity along the Middle America seismogenic zone to date and lend further insight into the thermal, mechanical, hydrological, and compositional interactions potentially responsible for controlling shallow subduction zone seismicity.

Comparison of the Osa and Nicoya experiment-specific 1D velocity models to previously reported country-wide 1D VP models, generally centered in central Costa Rica, shows similar velocity gradients between 5 and 30 km depth with mantle velocities (VP>7.8 km/sec) constrained between 35-50 km depth in the CRSEIZE experiment models compared to the large depth range (35-65 km) reported by other 1D velocity model studies DeShon et al., 2003; DeShon, 2004]. Differences in deep velocity structure between local and regional-scale 1D velocity studies are attributed to the relative location of the reference seismic station with respect to the subducting Cocos Plate and continental Moho.  A single station receiver function computed for Global Seismic Network station JTS in northwestern Costa Rica confirms the location of the continental Moho between 33-40 km depth and indicates 15% serpentinization of the mantle wedge near the Nicoya Peninsula [DeShon and Schwartz, 2004].  Hypocenter relocations across the Osa and Nicoya regions indicate a deepening of the updip limit from south to north, paralleling the steepening of the subducting Cocos Plate along strike of the Middle America Trench.

Absolute and relative relocations of ~300 aftershocks of the 1999 Quepos, Costa Rica, underthrusting earthquake are used to analyzes seismogenic zone structure offshore central Costa Rica during a period of increased seismicity rate.  DeShon et al. [2003]  shows that subduction of highly disrupted seafloor north of the Osa Peninsula has established a set of conditions that presently limit the seismogenic zone to be between 10-35 km below sea level, 30-95 km from the trench axis.  Aftershock locations define a plane dipping at 19° that marks the interface between the Cocos Plate and the Panama Block.  Relative event relocation produces a seismicity pattern similar to that obtained using absolute locations, increasing confidence in the geometry of the seismogenic zone. The aftershock locations spatially correlate with the downdip extension of the oceanic Quepos Plateau and reflect the structure of the mainshock rupture asperity. This strengthens an earlier argument that the 1999 Quepos earthquake ruptured specific bathymetric highs on the downgoing plate [Bilek et al., 2003] The areas of mainshock moment release are marked by the star & box on figure to the right.

We recently calculated high resolution earthquake locations and P-wave and P-wave/S-wave 3D velocity models for the locked Nicoya Peninsula segment of the Middle America subduction zone using an iterative, damped least squares local tomography method [DeShon et al., 2006].  In the southern Nicoya Peninsula, microseismicity along the plate interface extends from 12-26 km depth, 73-100 km from the trench axis while in the northern Nicoya Peninsula where hydrothermally cooled oceanic crust subducts, interplate seismicity extends from 17-28 km depth, 75-87 km from the trench axis. These results are in good agreement with preliminary CRSEIZE results presented in Newman et al. [2002].  Seismic velocity results suggest 10-30% serpentinization of the mantle wedge along the Nicoya Peninsula with the continental Moho occurring at 30-34 km depth [DeShon and Schwartz, 2004].  In this region, background microseismicity and geodetic measurements of locking along the subduction megathrust do not coincide [Norabuena et al., 2005], suggesting there are at least two mehcanical transitions occurring along the plate interface [Schwartz and DeShon, 2007; DeShon et al., 2006].  Recent focal mechanism determinations confirm that  events identified  as occurring on the subduction megathrust have thrust mechanisms [Hansen et al., 2006].The CRSEIZE experiment was a collaborative project that integrated land and ocean bottom seismic data, geodetic, and offshore fluid flow meter data as part of the NSF MARGINS program; principle investigators and collaborators included Susan Schwartz, Kevin Brown, Tim Dixon, LeRoy Dorman, Ernst Flueh, Paul Lungren, Marino Protti, Susan Bilek, Andrew Newman, Mike Tryon, Victor Gonzalez, Edmundo Norabuena, and Dan Sampson.