Earthquake Risk in Japan After Tohoku: Updates to the AIR Model

A New Understanding of the Hazard

AIR's updated model reflects new findings and updated rupture probabilities that will complement HERP's National seismic Hazard Maps. The Tohoku earthquake has had significant implications for the seismic source zones of the Japan Trench, increasing the probability of rupture on some segments and lessening it on others.

The event has also raised the question of whether a similarly unexpected event could occur elsewhere on the Japan Trench. For example, the area south of Tohoku rupture zone, near the Boso Peninsula, did not rupture during Tohoku and has not produced any large earthquakes in recorded history. In fact, the 2007 HERP model did not consider large magnitude earthquakes on this segment. However, the Tohoku earthquake suggests that this is no guarantee that a large earthquake cannot occur here in the future. Understanding and accounting for earthquake risk on this segment is very important, especially considering its proximity to Tokyo and a number of other large metropolitan areas. AIR researchers have therefore modeled a separate subduction segment for the area around the Boso Peninsula, which they refer to as the Boso segment. Although the uncertainty is high, it is possible that this area may have accumulated enough energy to generate a major event with a magnitude of 8.0-8.5. An event of this type could severely impact Tokyo.

Stress Transfer

AIR scientists also conducted original research to evaluate the impact of the Tohoku earthquake on crustal faults and on seismic source zones outside those affected by the Tohoku event, including the Kanto Plain, the Nankai Trough, and the Sagami Trough (Figure 1). They analyzed changes in Coulomb failure stress² in light of observed changes in the rate of seismicity in different regions to identify and model areas most affected by the Tohoku earthquake. AIR's study indicates some increase in the state of stress on the interface of the Pacific plate under Kanto Plain of Japan, which contains some of the country's most populous cities, including Tokyo.

Improved Understanding of Fault Geometry

Another important enhancement to the AIR model is a more detailed and precise understanding of fault geometry, particularly for the Nankai subduction zone. Japan's government has recently developed earthquake scenarios for this complex subduction area that take into consideration the latest research regarding the geometry of the plates. Several studies indicate that, near central Japan, the Pacific Plate is subducting at a more gradual angle than previously thought, effectively bringing earthquake events closer to the surface and thus increasing their intensity. This is supported by geophysical studies that suggest an extra fragment of plate appears to be lodged above the Pacific Plate under the Kanto Plain region, separating it from the Philippine Sea and Okhotsk Plates.

The Industry's First Fully Probabilistic Numerical Tsunami Model

The Tohoku event has had an enormous impact on the awareness of tsunami risk in Japan. While Japan is no stranger to tsunamis, those on the scale of the Tohoku event are rare. The Tohoku tsunami easily overtopped Japan's manmade levees in many coastal areas, with tragic results, and has highlighted the need to model such tsunami events to understand their future likelihood and possible impact. AIR is thus introducing the industry's first fully probabilistic tsunami model. Offering more than just deterministic scenarios or inundations footprints, AIR's model explicitly captures tsunami occurrence, intensity, and damage.

To model the tsunami (and seismic) risk, AIR used GPS data to evaluate the state of coupling for the Nankai Trough and Japan Trench, which are the most likely locations of future earthquakes that could generate the large amount of slip associated with damaging tsunamis.

Once a simulated earthquake generates a tsunami, the entire life of the tsunami is modeled, from fault rupture, to the tsunami's propagation across the sea, and throughout its entire inundation period. This simulation is based on the laws of physics that govern the motion of a fluid. For each event in the catalog capable of generating a tsunami, the two major damage determinants—tsunami height and forward speed—are modeled based on the characteristics of the originating earthquake, the bathymetry of the ocean, and overland surface features. The approach accommodates the non-uniform nature of earthquake slip to provide a realistic simulation that allows variations in a tsunami's height and speed at different inundation points.

A tsunami's height and speed are greatly affected by the friction it encounters along the floor of the ocean and on land. This friction increases as the wave approaches shallow water near the coastline and then over land. In the AIR model, land cover data was used to specify the friction characteristics and their effects on the tsunami. High resolution (50m) terrain and coastline data were used to enhance onshore terrain heights and the depths of waterways, harbors, and ports. The AIR model also takes into account Japan's extensive defense system, including in all simulations the probabilistic failure of a levee due to the hydrodynamic force of the wave. Therefore, even if a levee is not overtopped by a tsunami, the model includes any additional damage caused by a levee breach.

In the updated model, explicit tsunami damage estimates include both property damage and injury estimates. Property damage estimates incorporate the effects of debris impact as well as the force of the water. The personal injury line of business takes into account factors that impact the number of claims, such as the time of day and the time it takes the tsunami to arrive.

Modeling Liquefaction

Although eclipsed by the tsunami and shake damage, the liquefaction damage produced by the Tohoku event was severe. When an area saturated with groundwater experiences heavy ground shaking, the water pressure in the soil can cause it to lose its stiffness and behave like a heavy liquid. When this happens, the soil loses its capability to support structures, which can then tilt or sink even if they remain structurally intact. Widespread areas of the Kanto Plain, especially around Tokyo Bay and along the Tone River, exhibited deformed streets, sidewalks, and train tracks, as well as damage to underground pipes.

AIR's liquefaction module incorporates many recent studies on the surface geology, geomorphology, and shear wave velocity for regions that are prone to liquefaction. AIR researchers obtained groundwater data along with the borehole database of Japan from several sources including Japan's Ministry of Land, Infrastructure, Transportation, and Tourism (MLIT). The liquefaction settlement estimation follows the F_L method, established by Japan Road Association in 2002, which is now widely used in many studies, including many of the earthquake scenarios provided by Japan's government. The liquefaction damage estimation leverages Japan government recommendation as well as findings from the Christchurch, New Zealand earthquake.

Closing Thoughts

AIR understands the critical need for reliable risk assessments that include scenarios of a Tohoku-like event near the Kanto Plain. With the update to the AIR Earthquake Model for Japan, AIR is providing companies with a model that addresses the recent and dramatic changes to Japan's seismic risk landscape, as well as the need to explicitly model associated perils like tsunami and liquefaction.

Included in the updated model are damage functions for new construction classes, including large industrial facilities. These damage functions have been validated against an unprecedented amount of detailed claims data from the Tohoku earthquake. The fire damage module is also being updated with a major enhancement to the methodology used to determine fire outbreaks. It features a newly-developed algorithm for determining fire spread, which leverages an updated high-resolution industry exposure database.

While it is still too early to specify exactly how AIR-modeled loss estimates will change as a result of incorporating the new science and engineering, in the end, companies will benefit from a more realistic and comprehensive view of the risk.