U43A-01 INVITED
Understanding where a great earthquake can happen
Nearly every subduction zone in the world is long enough to produce a Mw = 9.0 or larger earthquake. For decades, Earth scientists have sought explanations as to why some do and some don't. As a consequence, prior to December 26, 2004, the North Sumatra - Andaman - Nicobar subduction zone was thought to be among the least likely to slip in a great earthquake. I will discuss some of the logic behind this notion and how new observations relate to it. Hindering these efforts is our very incomplete history of great earthquakes. Hence, correlations based on this earthquake history lead to incorrect conclusions. The Dec. 2004 earthquake and tsunami taught us that we are much more vulnerable to such natural disasters than we previously thought. Perhaps the lesson we should be gleaning is that every subduction zone may be a future site of a great quake and, where conditions are right, a tsunami.
U43A-02
Possible Morphologic Indicators for the Location of Large Slow Earthquakes in Subduction Zones
Global observations of convergent margin morphology may identify regions more likely to produce tsunami earthquakes. Earthquake observations and laboratory modeling show that subduction of seafloor relief influences the accretionary prism (AP), interplate coupling and the occurrence of large interplate earthquakes. The amplitude, wavelength, and direction of convergence of seafloor roughness appear to relate to the scale of the disruption of the subduction process. Sandbox models of AP deformation caused by subduction of relatively small seafloor relief causes a displacement of the active decollement into the AP, rather than along its base. If the relief has a "wake" (e.g. seamount), passive entrance of the toe of the AP, in the wake of the descending seafloor relief, down into the subduction channel occurs, leaving an indented toe. More importantly, when seafloor relief is sufficiently deep in the subduction zone, the decollement descends to the base of the AP. However, this new base has been little faulted if at all, as it was carried passively down the subduction channel. Therefore, conditions for normal earthquake rupture may not exist, and a tsunami earthquake may be more likely. The morphology of an indented AP can be easily identified globally in ETOPO-2 bathymetry. As most convergent margins define small circles and to first order, the bathymetry of the inner wall of the trench is nearly smooth, trench inner wall bathymetry should define conic sections. Residual bathymetry would be the difference between a regionally predicted conic section and observed bathymetry. Anomalously deep residual bathymetry successfully identifies the location of the Tonga, Nicaragua, 1963 Kuril Islands, and eastern Java slow earthquakes. Both slow earthquakes in Peru (1960, 1996) occur along non-accreting margin segments presently subducting rough seafloor with sediment filled troughs. The 1968 and 1994 "hybrid" earthquakes off Honshu are anomalous in that their ruptures began updip as slow earthquakes and later became regular earthquakes. Their initial ruptures lie in the source region of the 1896 slow earthquake.
U43A-03 INVITED
What have we learned about the Guerrero Gap from GPS?
The Guerrero seismic gap is a 200 km segment of the Cocos-North American plate boundary located near Acapulco, Mexico. There has been no megathrust earthquake in the northwest portion of the gap since 1911. If the entire gap were to rupture, it would result in a > M8 earthquake. Such an event would devastate not only cities in Guerrero, but in Mexico D.F. as well. The development of GPS in the past decade has made it possible to directly measure strain accumulation in the region with unprecedented precision. GPS measurements begun by UNAM in the early 1990's has led to the development of continuous GPS instrumentation throughout Guerrero and nearby states. Although our original intent in collecting GPS data was to measure steady-state deformation rates, we have also observed very large aseismic slip events in 1995-1996, 1998, 2002, and 2006 which have motions opposite to the steady-state trends. The presence of these slow slip events - where slip rates can be twice the size of the steady-state deformation rates - means that seismic recurrence intervals for the region must be re-evaluated. In this presentation we will examine stress and strain accumulation in Guerrero and how that impacts the risk for megathrust earthquakes in the region.
U43A-04
Stress Drop Estimates Based on Coda Wave Measures for Earthquakes Along the Nicoya Peninsula, Costa Rica
The majority of the world's largest earthquakes are confined to the subduction zone regions. Earthquake stress drop is an important parameter to characterize as it can illuminate changes in fault zone conditions. Here we determine stress drops for thrust zone earthquakes along the Nicoya Peninsula, Costa Rica to explore spatial variations in the Middle America subduction zone. This area is interesting because seismicity patterns vary along strike of the subduction zone as we see differences in the dip of the interface given by earthquake locations between the northern and southern ends of the Nicoya Peninsula, Costa Rica. We analyze seismograms from 694 well-located earthquakes that were recorded on land and ocean bottom seismometers from the 1999 CRSEIZE experiment placed on and near the Nicoya Peninsula. Earthquake source spectra are obtained from coda amplitude measurements taken in multiple frequency bands by correcting for average path effects, correcting for source excitation via an empirical Green's function technique, and shifting to absolute seismic moment based on independently known seismic moments of one or more events. The method of using coda energy in calculating earthquake magnitude tends to be more stable than using the direct arrival data because we minimize errors associated with the wave energy distortion arising from the crustal heterogeneity. We then use these values to compute stress drop for each earthquake along the fault zone. These earthquake stress drops will be compared with along-strike variations in b-value as well as structural heterogeneity to explore connections between subduction zone conditions and earthquake processes.
U43A-05
Examining Stress Changes Due to Subducting Topography and Variable Rheology in the Middle America Trench at Nicoya Gulf, Costa Rica
Offshore of the Nicoya Gulf at the Middle America Trench, the Cocos Plate is subducting beneath the Caribbean plate at about 84 mm per year. A line of seamounts are entering the trench in this region, causing dramatic deformation of the seafloor landward of the thrust. It has been suggested that these seamounts are being subducted, causing coastal uplift and seismicity. The March 25, 1990 Mw 7.0 Nicoya Gulf earthquake is thought to have occurred as one of these seamounts ruptured. How do these seamounts affect the rupture process? Are they behaving as patches of increased or decreased friction along the seismic interface? How does the subducting topography change the stress field after an earthquake? Can triggered events be explained by static stress changes, or does the rheology down dip from the seismogenic zone influence subsequent events in the region? Using a three dimensional model with patches of variable friction to simulate the seamounts as asperities, we compare the location of aftershocks to the stress changes associated with increased and decreased friction. We compare this to a model of Coulomb static stress change, which displays lobes of static stress increase and decrease due to slip on the fault plane, and the distribution of aftershocks within these lobes. To examine the stress changes associated with a set of delayed inland triggered events, we also vary the rheology of our model, using a linear elastic half space for the seismogenic zone, and viscous creep along the lower, aseismic portion of the fault below 40 kilometers. These models allow us to examine the spatial and temporal relationship of seismicity associated with stress changes due to variable friction and rheology. Our results indicate that stresses increase away from the fault with time if viscous creep is included in the model. These stress increases roughly correspond to inland areas of noted increase in seismicity, suggesting that creep along the down dip, aseismic portion of the interface, transmits stresses into the upper, elastic crust.
U43A-06 INVITED
Einstein's meanders
What does Einstein have to do with subduction? Good question. Peaceful Lake Budi, lying at the heart of an Indian reservation in the Deep South of Chile, had subsided by two meters in the 1960 mega-thrust earthquake. This unique South American salt lake was hiding an awful secret: it was actually an oxbow, not a lake. But Einstein had realized in 1926 that meanders are natural freaks. Rivers will not flow uphill, yet - he claimed - they don't flow down the path of steepest descent either. This anomaly was put at the doorstep of a weak Coriolis Force. Thus Einstein problematized the dilemma of the earth sciences. How can a non-force produce margin-parallel compression in a convergent margin where extension is expected? In fact, where does the energy for meander formation come from? Good question . . . Even Wikipedia knows that Coriolis is not a “force” but an “effect”. So is the obliquity of plate convergence in subduction. Where did Einstein err, and where was he a pioneer? Coastal ablation plus alternating subsidence and emergence in giant earthquakes may yield an answer. Einstein, A. (1926). Die Ursache der Maeanderbildung der Flusslaeufe und das sogenannte Baersche Gesetz, Naturwissenschaften, 14, fascicle II.