EGU 2011: Geodetic and inundation models of the Tohoku earthquake and tsunami

The recent meeting of the European Geosciences Union in Vienna included a session on the March 2011 Tohoku earthquake. Don Dingwell, the EGU president, introduced the session as the geoscientific community’s way of paying respect and expressing condolences to the victims of March’s seismic events. With researchers around the world doing their “best science to characterize and understand [the events],” he said, “we will move forward in mitigation, in the best possible way.”

James Daniell of Earthquake-Report pointed out that before the Tohoku earthquake, there had been a downward trend in economic losses from earthquakes since 1900. Over the last month, GPS surveys and geodetic models have provided some insight into the quake and subsequent tsunami that caused an unprecedented loss of life.

Another of the session speakers, Hiroshi Sato of the Earthquake Research Institute , noted, “The Tohoku earthquake [occurred in] a classic example of a trench arc–back arc system.” Trench arcs are formed when a heavy tectonic plate moves beneath a more buoyant one. The movement causes magma to rise and form an arc-shaped chain of volcanoes. As the crust extends, back arc basins form on the opposite side of the arc. Back arc spreading is the process that separates the Japanese islands from mainland Asia.

As the Pacific plate moves westward and subducts under the Okhotsk plate, Japan’s coastal area undergoes subsidence while the area further inland, near the volcanic crest, experiences uplift. The arc is relatively stable, but has been lifting at a rate of 0.3–0.6 mm/year since the Pliocene epoch, according to Sato. The side of the arc nearest the Pacific plate sinks due to compressional deformation. Although the plate is creeping westward at 80–90 mm/year, an earthquake occurs when there’s a bigger, faster slip than normal. The sudden and large deformation of the sea floor that ensues leads to tsunamis.

A bigger, faster slip than normal happened in the subduction zone at 38° N on 11 March 2011 and caused a displacement of 5 m to the east and 1 m to the south, according to Teruyuki Kato, also of the University of Tokyo. Half an hour later, the largest aftershock at 36° N caused a shift of 0.1 m east and 0.1 m south. And the maximum offset reached 20 m during the main shock and 1.5 m in the aftershock.

Sato explained that “a large amount of strain release is needed along the interface to create this scenario.” Understanding the process of strain buildup and release in the subduction zone is critical for evaluating the risk of a subduction zone “megathrust” quake—that is, magnitude 9.0 or greater—and Andreas Hoechner of the GFZ German Research Centre for Geosciences has been developing a model to do just that.

By examining GPS data for the 22 earthquakes that occurred in this region over the past century, Hoechner determined whether the plates are sliding or have been locked in place over the years. “Locking is very high on this [the Tohoku epicienter] location,” he said. “There’s lots of deformation and lots of slip accumulates.” Slip is the amount of relative motion of the plates at the contact interface of the plates during the earthquake. When plates are locked, strain builds up as they push against each other and deform. This results in a slip deficit, which is potentially released during an earthquake although the situation may be stable for centuries. The Tohoku earthquake ruptured offshore from Sendai in the region that showed over 75% locking.

To calculate the slip distribution, Hoechner’s model uses the coseismic displacement—that is, the difference between the GPS information (30-minute-interval kinematic solution from onshore measurements) before and after the main shock. To test the procedure, he makes a baseline model using minimal slip, then adds noise to three separate slip areas and to a smooth slip distribution. “The offshore part is not very well resolved,” he said. Simulating the offshore part by breaking it into 216 subfaults on a curved surface, he inverts the GPS data for a slip and shows a maximum slip of 36 m.

Deformation at the sea floor also uplifts a water column to the sea surface and is calculated by the slip model. Whereas Hoechner calculated a maximum uplift of 9 m, which can then be used as an initial condition for a tsunami model in which gravity drives water dissipation, Kato stressed that more ocean bottom measurements are needed to provide data about both locking and floor deformation.

A tsunami early warning system is one very important goal of understanding the subduction region locking conditions. Deepak Vatvani of Deltares research institute specified that the most important thing is “how fast you can tell people the inundation possibility once you know the propagation.”

Once the initial conditions of the tsunami are known, Vatvani doesn’t need very high resolution data to model how far the waves will propagate. But predicting the inundation requires high-resolution data and intensive computation. By the time standard models calculate the inundation of an oncoming tsunami, it would be too late to be useful in a warning system, even though the calculations agree with measured wavelengths and heights.

To speed things up, Vatvani uses a propagation grid that is generated from a simulated wave and translates it to a finer grid using an empirical formula derived from hundreds of different topographies. This gives him a faster way of estimating tsunami heights on the shore. “We would compare this with satellite data, but the satellite only shows water left behind and not the rate,” he said, adding that we “can’t predict inundation based on earthquake magnitude alone.” The Sumatra 2004 earthquake was comparable in magnitude to Tohoku 2011, but Tohoku ruptured over a shorter geographic region so the waves were much higher. Japan knew a tsunami was coming, but not how large it would be.

Does all of this mean that more earthquakes will happen in the near future due to aftershocks? “We’re drawn to think that there’s periodicity [in earthquake behavior], but studies have shown that earthquakes are capricious,” said Emile Okal of Northwestern University’s department of Earth and planetary sciences . Earthquakes are expected around active subduction zones, but stress transfer at a plate boundary becomes complicated. And other sources of stress may or may not be involved in earthquake triggers.

No matter how comprehensive our models and warning systems become, Okal reminded the EGU audience that “an earthquake takes time to prepare, like good food. It can start small and grow bigger. No matter how good these plans might seem, we shouldn’t try to outsmart Mother Nature.”

Rachel Berkowitz