U33C-01
What Controls Subduction Earthquake Size and Occurrence?
There is a long history of observational studies on the size and recurrence intervals of the large underthrusting earthquakes in subduction zones. In parallel with this documentation of the variability in both recurrence times and earthquake sizes -- both within and amongst subduction zones -- there have been numerous suggestions for what controls size and occurrence. In addition to the intrinsic scientific interest in these issues, there are direct applications to hazards mitigation. In this overview presentation, I review past progress, consider current paradigms, and look toward future studies that offer some resolution of long- standing questions. Given the definition of seismic moment, earthquake size is the product of overall static stress drop, down-dip fault width, and along-strike fault length. The long-standing consensus viewpoint is that for the largest earthquakes in a subduction zone: stress-drop is constant, fault width is the down-dip extent of the seismogenic portion of the plate boundary, but that along-strike fault length can vary from one large earthquake to the next. While there may be semi-permanent segments along a subduction zone, successive large earthquakes can rupture different combinations of segments. Many investigations emphasize the role of asperities within the segments, rather than segment edges. Thus, the question of earthquake size is translated into: "What controls the along-strike segmentation, and what determines which segments will rupture in a particular earthquake cycle?" There is no consensus response to these questions. Over the years, the suggestions for segmentation control include physical features in the subducted plate, physical features in the over-lying plate, and more obscure -- and possibly ever-changing -- properties of the plate interface such as the hydrologic conditions. It seems that the full global answer requires either some unforeseen breakthrough, or the long-term hard work of falsifying all candidate hypotheses except one. This falsification process requires both concentrated multidisciplinary efforts and patience. Large earthquake recurrence intervals in the same subduction zone segment display a significant, and therefore unfortunate, variability. Over the years, many of us have devised simple models to explain this variability. Of course, there are also more complicated explanations with many additional model parameters. While there has been important observational progress as both historical and paleo-seismological studies continue to add more data pairs of fault length and recurrence intervals, there has been a frustrating lack of progress in elimination of candidate models or processes that explain recurrence time variability. Some of the simple models for recurrence times offer a probabilistic or even deterministic prediction of future recurrence times – and have been used for hazards evaluation. It is important to know if these models are correct. Since we do not have the patience to wait for a strict statistical test, we must find other ways to test these ideas. For example, some of the simple deterministic models for along-strike segment interaction make predictions for variation in tectonic stress state that can be tested during the inter-seismic period. We have seen how some observational discoveries in the past decade (e.g., the episodic creep events down-dip of the seismogenic zone) give us additional insight into the physical processes in subduction zones; perhaps multi-disciplinary studies of subduction zones will discover a new way to reliably infer large-scale shear stresses on the plate interface?
U33C-02 INVITED
Structural and Physical Controls on Segmentation and Earthquake Rupture of the Sumatran Subduction Zone
A major UK and international project is investigating how the 3D structure and physical properties of the subduction zone (including the plate boundary) affect earthquake rupture within and between the 2004 and 2005 rupture zones. These earthquake ruptures terminated at clear segment boundaries of the subduction zone and, in part, repeated historic rupture patterns. The project includes marine surveys (seismic refraction and reflection, heatflow measurements, sidescan sonar and coring), a combined onshore and offshore earthquake recording experiment and refinement of earthquake slip distribution models. The project focuses on two segment boundaries: between the 2004 and 2005 rupture zones and at the southern termination of the 2005 rupture. The data will be used to (a) determine how prism and plate boundary structure, thermal properties and seismic velocity structure change across the rupture boundaries, (b) further constrain the seismogenic zone position based on well located seismicity and thermal properties, (c) link the slip distribution to forearc structure and (d) assess the paleoseismic history of these rupture segments. The first phase of the marine experiment (seismic refraction and reflection) was conducted May-July 2008 with passive seismic recording ongoing. Initial results reveal details of prism structure and thrust development including changing vergence patterns along the toe of the prism; clear imaging of the plate boundary to ~75 km landward of the deformation front; and structure of the oceanic plate including basement topography and a thick sedimentary section. The 3D topography of the subducting plate at the southern termination of the 2005 earthquake (Nias island) and 1935 rupture zone (Batu Islands) is particularly complex and variable due to subduction of oceanic plate basement structure. These structures may be responsible for the 2005 segment boundary and the anomalous rupture history beneath the Batu Islands. Reflection profiles across the incoming trench wedge offshore Simeulue island (2004-2005 boundary) reveal a prominent reflector at the base of the sedimentary section with changing polarity along strike. This reflector is interpreted as the proto-decollement. Further analysis will investigate its changing properties and potential relationship to earthquake rupture.
U33C-03
Earthquake Variations Along the Middle America Trench, Central America
Subduction zones, home to the majority of the world's earthquakes, have also been the hosts for some of the world's most devastating tsunami. Seismic characteristics of subduction zone earthquakes, including the time constant of rupture, range from typical fast (few-10s seconds) ruptures to slow, long duration (100s of seconds) tsunami events, to silent (days-years) earthquakes that are observed geodetically rather than seismically. Causes for variations in the seismic parameters are currently not well understood, yet the causative factors influencing rupture could be very important in seismic and tsunami hazard assessment. Several aspects of subduction zone conditions may be important for influencing earthquake rupture, with our focus here on the effects of subducting plate inputs on earthquake rupture characteristics. Our main focus area for this study is along the Central American subduction zone because of the well-characterized input into the zone as well as detailed seismic catalogs. This region has also experienced slow, silent, and tsunami earthquakes in the past. We suggest that complexity on the subducting Cocos plate dictates variations in rupture characteristics for earthquakes over a range of magnitudes. Preliminary results for earthquakes larger than Mw=5.7 suggest earthquakes with slow rupture durations from southern Mexico through Costa Rica, with the slowest events occurring in the region between the high slip zones of the 1992 Nicaragua tsunami earthquake, away from imaged seamounts in the high slip regions. Slow earthquakes are also observed in southern Mexico. In addition, analysis of rupture characteristics of small magnitude (Mw<4) along the Costa Rica margin also suggests variations in stress drop that correspond with variations in geodetically estimated plate locking and rupture zones of past large earthquakes.
U33C-04 INVITED
Issues and Advances in Understanding Landslide-Generated Tsunamis: Toward a Unified Model
The physics of tsunamis generated from submarine landslides is highly complex, involving a cross- disciplinary exchange in geophysics. In the 10 years following the devastating Papua New Guinea tsunami, there have been significant advances in understanding landslide-generated tsunamis. However, persistent issues still remain related to submarine landslide dynamics that may be addressed with collection of new marine geologic and geophysical observations. We review critical elements of landslide tsunamis in the hope of developing a unified model that encompasses all stages of the process from triggering to tsunami runup. Because the majority of non-volcanogenic landslides that generate tsunamis are triggered seismically, advances in understanding inertial displacements and changes in strength and rheologic properties in response to strong-ground motion need to be included in a unified model. For example, interaction between compliant marine sediments and multi-direction ground motion results in greater permanent plastic displacements than predicted by traditional rigid-block analysis. When considering the coupling of the overlying water layer in the generation of tsunamis, the post-failure dynamics of landslides is important since the overall rate of seafloor deformation for landslides is less than or comparable to the phase speed of tsunami waves. As such, the rheologic and mechanical behavior of the slide material needs to be well understood. For clayey and silty debris flows, a non-linear (Herschel-Bulkley) and bilinear rheology have recently been developed to explain observed runout distances and deposit thicknesses. An additional complexity to this rheology is the inclusion of hydrate-laden sediment that commonly occurs along continental slopes. Although it has been proposed in the past that gas hydrate dissociation may provide potential failure planes for slide movement, it is unclear how zones of rigid hydrate-bearing sediment surrounded by a more viscoplastic matrix affects the overall rheologic behavior during slide dynamics. For more rigid materials, such as carbonate and volcanic rocks, models are being developed that encompass the initial fracturing and eventual disintegration associated with debris avalanches. Lastly, the physics dictating the hydrodynamics of landslide-generated tsunamis is equally complex. The effects of non-linearity and dispersion are not necessarily negligible for landslides (in contrast to most earthquake-generated tsunamis), indicating that numerical implementation of the non-linear Boussinesq equations is often needed. Moreover, we show that for near-field landslide tsunamis propagating across the continental shelf, bottom friction (bottom boundary layer turbulence) and wave breaking can be important energy sinks. Detailed geophysical surveys can dissect landslide complexes to determine the geometry of individual events and help estimate rheological properties of the flowing mass, whereas cores in landslide provinces can determine the mechanical properties and pore-pressure distribution for pre- and post-failure sediment. This information is critical toward developing well-documented case histories for validating physics-based landslide tsunami models.
U33C-05
Hazards in volcanic arcs
Volcanic eruptions in arcs are complex natural phenomena, involving the movement of magma to the Earth's surface and interactions with the surrounding crust during ascent and with the surface environment during eruption, resulting in secondary hazards. Magma changes its properties profoundly during ascent and eruption and many of the underlying processes of heat and mass transfer and physical property changes that govern volcanic flows and magmatic interactions with the environment are highly non-linear. Major direct hazards include tephra fall, pyroclastic flows from explosions and dome collapse, volcanic blasts, lahars, debris avalanches and tsunamis. There are also health hazards related to emissions of gases and very fine volcanic ash. These hazards and progress in their assessment are illustrated mainly from the ongoing eruption of the Soufriere Hills volcano. Montserrat. There are both epistemic and aleatory uncertainties in the assessment of volcanic hazards, which can be large, making precise prediction a formidable objective. Indeed in certain respects volcanic systems and hazardous phenomena may be intrinsically unpredictable. As with other natural phenomena, predictions and hazards inevitably have to be expressed in probabilistic terms that take account of these uncertainties. Despite these limitations significant progress is being made in the ability to anticipate volcanic activity in volcanic arcs and, in favourable circumstances, make robust hazards assessments and predictions. Improvements in monitoring ground deformation, gas emissions and seismicity are being combined with more advanced models of volcanic flows and their interactions with the environment. In addition more structured and systematic methods for assessing hazards and risk are emerging that allow impartial advice to be given to authorities during volcanic crises. There remain significant issues of how scientific advice and associated uncertainties are communicated to provide effective mitigation during volcanic crises.
U33C-06
Strong Influence of Near Tropopause Winds on Distal Volcanic Ash Transport
Comparison of ash cloud satellite images, the resulting fallout deposits, and advection simulations that utilize high resolution wind-field data can provide important insights into the atmospheric levels where dominant ash transport occurs during explosive eruptions. The assumption that high intensity eruptions result in ash transport at significantly higher levels than moderate-sized events may not necessarily be true for the distal portion of widespread ash fall events. The Lagrangian ash dispersal model PUFF was used to simulate two historic eruptions, the 1980 eruption of Mt. St. Helens (< 1 km3 dense rock equivalent of magma, DRE) and the 1991 eruption of Mt. Pinatubo (~ 5 km3 DRE). Results from PUFF simulations were directly compared to visible, infrared, and ultraviolet satellite images taken during the eruptions. Despite significant differences in peak eruption column heights the best match between the observed and simulated ash clouds is produced when the majority of ash is concentrated near the tropopause in both cases. The position of these simulated ash clouds also matches the position of the observed fall deposits. These results indicate that the long range distribution of distal ash clouds and ash deposits may be controlled primarily by ash transport near the tropopause, a zone characterized by significant shear and high velocity winds. If these results can be shown to be more generally applicable, they may lead to rapid forecasting capabilities for geohazards generated by the distal dispersion of ash from large volcanic eruptions, including but not limited to hazards to aviation.
U33C-07
Holocene Paleoenvironments, Relative Sea-Level Changes and Marine Incursions On The Mexican Pacific Coast
Although the relative sea-level and environment variability have received considerable attention elsewhere, relatively little has been directed toward sea-level and significant environmental changes on the Mexican Pacific coast in the Holocene (11,500 cal yr BP to present) noted in this study. Examination of previous published work and sediment-stratigraphic records of lagoon sediments reveal strong evidence for changes in salinity and environment of lagoonal marshes, which provide scientific bases for past relative sea-level change and marine incursions of the on the Mexican Pacific coast. Sediment records show that sea-level was at or close to the present level by at least c. 4630 yr BP, when the brackish lagoons and beach barriers where established. A marine inundation (evidence of a probable tsunami is presented), that occurred by c. 3400-3500 yr BP. A return to marginal lagoon conditions, indicating a drop in relative sea-level, occurred in the most recent time (c. 2300 yr BP). Climate records in the mid- to late-Holocene, although scarce and clustered in Yucatan, central and northern Mexico, cannot explain the observed environmental changes. In general, most of the Holocene relative sea-level changes observed obscure any climatic signal in this period on the tectonically-active Mexican Pacific coast.
U33C-08
Geological record of severe storm impacts along the Texas Coast
Hurricanes act as one of the primary controls on barrier island migration through wave and wind energy, and their frequency has been suggested to indicate changes in climate (El Niño) cycles. Texas has an extensive coastline containing barriers in various stages of evolution. Through a detailed sedimentological examination and radiocarbon age constraints of offshore storm sands, beach ridge breaching events, storm surge channels, and washovers, we offer a geologic record of severe storm impacts along the Texas Coast. From offshore core data, we ascertain that sand storage along the upper and lower shoreface (the profile of which is controlled by catastrophic storm impacts) is minimal over geologic timescales (i.e. 100-1000 years). Hence, an offshore record of storm impact is lacking. Using high resolution LIDAR data, we map breaching events of prominent beach ridges. Storm surge channels on the bayside of barriers (which are cut by water flowing towards the Gulf of Mexico when storm surge recedes) are also being dated, although they likely record lower magnitude storms. This study reveals that hurricane washover formation is only a minor contributor to sand transport within the system, as accumulation rates in back-barriers range from .095 - .4m/C. By examining the sedimentological components of hurricane impacts, we establish a hurricane impact chronology and conclude that the frequency of major storms along the Texas Coast is actually quite minimal.