Eric L. Geist

Local Tsunamis and Earthquake Source Parameters

Publisher Summary This chapter establishes the relationship among earthquake source parameters and the generation, propagation, and run-up of local tsunamis. In general terms, displacement of the seafloor during the earthquake rupture is modeled using the elastic dislocation theory for which the displacement field is dependent on the slip distribution, fault geometry, and the elastic response and properties of the medium. Specifically, nonlinear long-wave theory governs the propagation and run-up of tsunamis. A parametric study is devised to examine the relative importance of individual earthquake source parameters on local tsunamis, because the physics that describes tsunamis from generation through run-up is complex. Analysis of the source parameters of various tsunamigenic earthquakes have indicated that the details of the earthquake source, namely, nonuniform distribution of slip along the fault plane, have a significant effect on the local tsunami run-up. Numerical methods have been developed to address the realistic bathymetric and shoreline conditions. The accuracy of determining the run-up on shore is directly dependent on the source parameters of the earthquake, which provide the initial conditions used for the hydrodynamic models.

Upper crustal structure and Neogene tectonic development of the California continental borderland

Multichannel seismic-reflection data, sonobuoy seismic-refraction data, and regional geology are used to define the upper crustal structure of the southern California continental borderland and to delineate the characteristics of the main lithotectonic belts of the region. The Catalina Schist belt is separated on its west side from the gently deformed Nicolas forearc belt by faults that have steep west dips and pronounced normal separations. On its east side the schist belt is bounded by a large detachment fault that dips gently to the east beneath the west edge of the Peninsular Ranges belt at the coastline near Oceanside. The Catalina Schist was uplifted from middle crustal depths and exposed during a major event of extensional tectonism that started in early Miocene time in conjunction with about 10° of clockwise rotation of the western Transverse Ranges belt. Part of the uplift of the Catalina Schist could have occurred on the detachment fault, but it is thought to have mostly occurred on the steep faults that bound the west edge of the schist belt. A large amount of uplift is required, and it probably involved strong footwall flexural deformation in the wake of the translating and rotating western Transverse Ranges and Nicolas forearc belts. Extension, accompanied by probable large amounts of right slip, continued in the borderland region during and after middle Miocene time. The later stage of extension was accompanied by rapid clockwise rotation of the western Transverse Ranges of at least 90°. Most of the borderland, including the belt of schist that was uplifted in early Miocene time, was further deformed into numerous basins and ridges during this stage of oblique extension. The primary driving force for the deformation is thought to have been derived from the rapid northwest motion of the Pacific plate after it had become coupled to the Farallon plate system, which had previously been subducted beneath the borderland.

Crustal structure of accreted terranes in southern Alaska, Chugach Mountains and Copper River Basin, from seismic refraction results

Seismic refraction data were collected along a 320-km-long “Transect” line in southern Alaska, crossing the Prince William, Chugach, Peninsular, and Wrangellia terranes, and along several shorter lines within individual terranes. Velocity structure in the upper crust (less than 9-km depth) differs among the four terranes. In contrast, layers in the middle crust (9- to 25-km depth) in some cases extend across projected terrane boundaries. The following observations can be made: (1) An intermediate-velocity layer (6.4 km/s) at 9-km depth extends across the deep projection of the suture between the Chugach and Peninsular terranes, suggesting that the northern Chugach and southern Peninsular terranes are detached and rest on a deeper terrane of unknown origin. (2) The top of a gently north dipping sequence of low- and high-velocity layers (5.7–7.8 km/s), more than 10 km thick, extends from near the surface in the southern Chugach terrane to more than 20-km depth beneath the southern Peninsular terrane. This sequence, truncated by the suture between the Prince William and Chugach terranes, is interpreted to be an underplated “terrane” made up of fragments of the Kula plate and its sedimentary overburden that were accreted during subduction in the late Mesozoic and/or early Tertiary, during or between times of accretion of the Prince William and Chugach terranes. (3) A thick crustal “root”, with a laminated sequence at its top, extends from a depth of 19 km to as much as 57 km beneath the northern Peninsular and Wrangellia terranes. This root extends across the deep projection of the suture between the Peninsular and Wrangellia terranes, although resolution of this apparent crosscutting relationship is relatively poor. This root may represent tectonically or, possibly, magmatically emplaced rocks. The lower crust beneath the Prince William, Chugach, and southern Peninsular terranes includes a north dipping, 3- to 8-km-thick section of subducting oceanic crust.

Implications of the 26 December 2004 Sumatra–Andaman Earthquake on Tsunami Forecast and Assessment Models for Great Subduction-Zone Earthquakes

Results from different tsunami forecasting and hazard assessment mod- els are compared with observed tsunami wave heights from the 26 December 2004 Indian Ocean tsunami. Forecast models are based on initial earthquake information and are used to estimate tsunami wave heights during propagation. An empirical forecast relationship based only on seismic moment provides a close estimate to the observed mean regional and maximum local tsunami runup heights for the 2004 Indian Ocean tsunami but underestimates mean regional tsunami heights at azimuths in line with the tsunami beaming pattern (e.g., Sri Lanka, Thailand). Standard forecast models developed from subfault discretization of earthquake rupture, in which deep- ocean sea level observations are used to constrain slip, are also tested. Forecast models of this type use tsunami time-series measurements at points in the deep ocean. As a proxy for the 2004 Indian Ocean tsunami, a transect of deep-ocean tsunami amplitudes recorded by satellite altimetry is used to constrain slip along four subfaults of the M 9 Sumatra-Andaman earthquake. This proxy model performs well in comparison to observed tsunami wave heights, travel times, and inundation patterns at Banda Aceh. Hypothetical tsunami hazard assessments models based on end- member estimates for average slip and rupture length (M w 9.0-9.3) are compared with tsunami observations. Using average slip (low end member) and rupture length (high end member) (M w 9.14) consistent with many seismic, geodetic, and tsunami inversions adequately estimates tsunami runup in most regions, except the extreme runup in the western Aceh province. The high slip that occurred in the southern part of the rupture zone linked to runup in this location is a larger fluctuation than expected from standard stochastic slip models. In addition, excess moment release (9%) deduced from geodetic studies in comparison to seismic moment estimates may gen- erate additional tsunami energy, if the exponential time constant of slip is less than approximately 1 hr. Overall, there is significant variation in assessed runup heights caused by quantifiable uncertainty in both first-order source parameters (e.g., rupture length, slip-length scaling) and spatiotemporal complexity of earthquake rupture.

Effect of depth‐dependent shear modulus on tsunami generation along subduction zones

Estimates of the initial size of tsunamis generated by subduction zone earthquakes are significantly affected by the choice of shear modulus at shallow depths. Analysis of over 360 circum-Pacific subduction zone earthquakes indicates that for a given seismic moment, source duration increases significantly with decreasing depth (Bilek and Lay, 1998; 1999). Under the assumption that stress drop is constant, the increase of source duration is explained by a 5-fold reduction of shear modulus from depths of 20 km to 5 km. This much lower value of shear modulus at shallow depths in comparison to standard earth models has the effect of increasing the amount of slip estimated from seismic moment determinations, thereby increasing tsunami amplitude. The effect of using depth dependent shear modulus values is tested by modeling the tsunami from the 1992 Nicaraguan tsunami earthquake using a previously determined moment distribution (Ihmle, 1996a). We find that the tide gauge record of this tsunami is well matched by synthetics created using the depth dependent shear modulus and moment distribution. Because excitation of seismic waves also depends on elastic heterogeneity, it is important, particularly for the inversion of short period waves, that a consistent seismic/tsunami shear modulus model be used for calculating slip distributions.

Mw 8.6 Sumatran earthquake of 11 April 2012: Rare seaward expression of oblique subduction

The magnitude 8.6 and 8.2 earthquakes off northwestern Sumatra on 11 April 2012 generated small tsunami waves that were recorded by stations around the Indian Ocean. Combining differential travel-time modeling of tsunami waves with results from back projection of seismic data reveals a complex source with a significant trench-parallel component. The oblique plate convergence indicates that ∼20–50 m of trench-parallel displacement could have accumulated since the last megathrust earthquake, only part of which has been taken up by the Great Sumatran fault. This suggests that the remaining trench-parallel motion was released during the magnitude 8.6 earthquake on 11 April 2012 within the subducting plate. The magnitude 8.6 earthquake is interpreted to be a result of oblique subduction as well as a reduction in normal stress due to the occurrence of the Sumatra-Andaman earthquake in 2004.

Differences in tsunami generation between the December 26, 2004 and March 28, 2005 Sumatra earthquakes

Source parameters affecting tsunami generation and propagation for the Mw > 9.0 December 26, 2004 and the Mw = 8.6 March 28, 2005 earthquakes are examined to explain the dramatic difference in tsunami observations. We evaluate both scalar measures (seismic moment, maximum slip, potential energy) and finite-source representations (distributed slip and far-field beaming from finite source dimensions) of tsunami generation potential. There exists significant variability in local tsunami runup with respect to the most readily available measure, seismic moment. The local tsunami intensity for the December 2004 earthquake is similar to other tsunamigenic earthquakes of comparable magnitude. In contrast, the March 2005 local tsunami was deficient relative to its earthquake magnitude. Tsunami potential energy calculations more accurately reflect the difference in tsunami severity, although these calculations are dependent on knowledge of the slip distribution and therefore difficult to implement in a real-time system. A significant factor affecting tsunami generation unaccounted for in these scalar measures is the location of regions of seafloor displacement relative to the overlying water depth. The deficiency of the March 2005 tsunami seems to be related to concentration of slip in the down-dip part of the rupture zone and the fact that a substantial portion of the vertical displacement field occurred in shallow water or on land. The comparison of the December 2004 and March 2005 Sumatra earthquakes presented in this study is analogous to previous studies comparing the 1952 and 2003 Tokachi-Oki earthquakes and tsunamis, in terms of the effect slip distribution has on local tsunamis. Results from these studies indicate the difficulty in rapidly assessing local tsunami runup from magnitude and epicentral location information alone.

Is There a Basis for Preferring Characteristic Earthquakes over a Gutenberg–Richter Distribution in Probabilistic Earthquake Forecasting?

The idea that faults rupture in repeated, characteristic earthquakes is cen- tral to most probabilistic earthquake forecasts. The concept is elegant in its simplicity, and if the same event has repeated itself multiple times in the past, we might anticipate the next. In practice however, assembling a fault-segmented characteristic earthquake rupture model can grow into a complex task laden with unquantified uncertainty. We weigh the evidence that supports characteristic earthquakes against a potentially simpler model made from extrapolation of a Gutenberg-Richter magnitude-frequency law to individual fault zones. We find that the Gutenberg-Richter model satisfies key data constraints used for earthquake forecasting equally well as a characteristic model. Therefore, judicious use of instrumental and historical earthquake catalogs enables large-earthquake-rate calculations with quantifiable uncertainty that should get at least equal weighting in probabilistic forecasting.

Source parameters controlling the generation and propagation of potential local tsunamis along the cascadia margin

The largest uncertainty in assessing hazards from local tsunamis along the Cascadia margin is estimating the possible earthquake source parameters. We investigate which source parameters exert the largest influence on tsunami generation and determine how each parameter affects the amplitude of the local tsunami. The following source parameters were analyzed: (1) type of faulting characteristic of the Cascadia subduction zone, (2) amount of slip during rupture, (3) slip orientation, (4) duration of rupture, (5) physical properties of the accretionary wedge, and (6) influence of secondary faulting. The effect of each of these source parameters on the quasi-static displacement of the ocean floor is determined by using elastic three-dimensional, finite-element models. The propagation of the resulting tsunami is modeled both near the coastline using the two-dimensional (x-t) Peregrine equations that includes the effects of dispersion and near the source using the three-dimensional (x-y-t) linear long-wave equations. The source parameters that have the largest influence on local tsunami excitation are the shallowness of rupture and the amount of slip. In addition, the orientation of slip has a large effect on the directivity of the tsunami, especially for shallow dipping faults, which consequently has a direct influence on the length of coastline inundated by the tsunami. Duration of rupture, physical properties of the accretionary wedge, and secondary faulting all affect the excitation of tsunamis but to a lesser extent than the shallowness of rupture and the amount and orientation of slip. Assessment of the severity of the local tsunami hazard should take into account that relatively large tsunamis can be generated from anomalous ‘tsunami earthquakes’ that rupture within the accretionary wedge in comparison to interplate thrust earthquakes of similar magnitude.

Were Global M ≥8:3 Earthquake Time Intervals Random between 1900 and 2011?

The pattern of great earthquakes during the past ∼100 yr raises questions whether large earthquake occurrence is linked across global distances, or whether temporal clustering can be attributed to random chance. Great-earthquake frequency during the past decade in particular has engendered media speculation of heightened global hazard. We therefore examine interevent distributions of Earths largest earth- quakes at one-year resolution, and calculate how compatible they are with a random- in-time Poisson process. We show, using synthetic catalogs, that the probability of any specific global interevent distribution happening is low, and that short-term clusters are the least repeatable features of a Poisson process. We examine the real catalog and find, just as expected from synthetic catalogs, that the least probable M ≥8:3 earth- quake intervals during the past 111 yr were the shortest ( t< 1 yr) if a Poisson process is active (mean rate of 3.2%). When we study an M ≥8:3 catalog with locally trig- gered events removed, we find a higher mean rate of 9.5% for 0-1 yr intervals, com- parable to the value (11.1%) obtained for simulated catalogs drawn from random- in-time exponential distributions. We emphasize short interevent times here because they are the most obvious and have led to speculation about physical links among global earthquakes. We also find that comparison of the whole 111-yr observed M ≥8:3 interevent distribution (including long quiescent periods) to a Poisson process is not significantly different than the same comparison made with synthetic catalogs. We therefore find no evidence that global great-earthquake occurrence is not a ran- dom-in-time Poisson process.