logo image
Helen YatesJune 15, 2021
Tohoku earthquake
Tohoku earthquake
The Earthquakes That Changed Everything
June 15, 2021

In the decade since the devastating 2011 Tohoku and Christchurch Earthquakes, risk modelers such as RMS have been proactive in honoring the data generated from these events. It is a process of continuous learning and refinement, explains Chesley Williams and Laura Barksby from RMS, and the journey is still ongoing Within the space of just 17 days in 2011, Christchurch in New Zealand was rocked by an M6.3 earthquake — almost directly beneath the central business district (CBD) — and Japan was reeling from the most powerful earthquake in its history. At RMS, event response and reconnaissance teams were mobilized with the implicit knowledge they were there to do justice to those affected and to gather as much data as possible in order to advance seismic hazard and risk understanding. The tsunami waves triggered by the M9.0 Tohoku Earthquake inundated approximately 532 square kilometers (205 square miles) of the country's northeastern coastline. At its highest point, the waves reached over 15 meters (49 feet) in some localized areas. They overtopped seawalls, destroyed 122,000 buildings with over a million buildings severely or partially damaged, and damaged 230,000 vehicles. The event also triggered level seven meltdowns at Fukushima Dai-ichi Nuclear Power Station. The disaster at Chernobyl in 1986 was the only previous level seven event. The catastrophe was watched, in horror, in real time on news channels around the world. In total, it caused 15,899 deaths with 2,527 missing. Estimated economic damage totaled US$235 billion. When initiating a risk model update, cat modelers generally start with published national seismic hazard maps, which are typically built on consensus-based research. This represents an important first step — to review the key assumptions in such studies to make sure that they are consistent with the latest data and methods for seismic hazard and risk assessment. “When we developed our RMS® Japan Earthquake Model in 2005, the starting point was the first version of the national seismic hazard maps released in that year,” says Chesley Williams, senior director at RMS. “We had detailed discussions with the Japanese researchers who developed the maps to understand the key assumptions, particularly with a focus on the sources impacting Tokyo.” After the 2011 event, it is now clear that the Japan Trench can produce M9 events. The 2005 national hazard maps were focused on the M7-M8 events that had occurred in the preceding 450+ years. Looking at the geologic record that there have likely been large, possibly M9 events in the past, for example the Jogan Sanriku Earthquake in 869. Honoring the Data So much about both these events in 2011 — the Christchurch Earthquake on February 22 and Tohoku Earthquake on March 11 — was unexpected. Although New Zealand is a highly seismic region, the 2010-11 Canterbury Earthquake Sequence occurred in an area that historically had relatively low seismicity. Prior to the Canterbury Earthquake Sequence there were fewer than 30 earthquakes of magnitude four or greater in Christchurch and the immediate surrounding area. In the last decade, there have been more than 370 earthquakes in this region. The Christchurch Earthquake caused higher-than-expected ground motions and unprecedented liquefaction. As a result, it was the costliest event in the sequence and the second-highest insured loss from earthquake in history, after the Tohoku Earthquake. Japan is also highly seismic, but the Tohoku event occurred on structures that had not shown their full potential during the historical record. The intensity of M9.0 Tohoku was particularly surprising in scale — the highest-ever economic losses from a natural disaster — and the tsunami impact was unprecedented for Japan. “Both Christchurch and Tohoku taught us an awful lot about earthquake risk, including the secondary impacts of earthquakes — tsunami, liquefaction, landslides, nuclear disaster, aftershocks, business interruption, contingent business interruption, and post-event loss amplification,” commented Williams. “They transformed how we think about and model seismic hazard and risk.” New Insights Into Large-Magnitude Events Media coverage of Tohoku clearly showed that the damage in the tsunami induction zones was catastrophic. Once the search and rescue work had been completed, RMS sent a reconnaissance team to Japan to examine tsunami damage and also damage from strong ground shaking, which was extensive. Key observations from this work included that older (pre-1981) concrete buildings often sustained significant damage at high ground motions, traditional wooden homes with heavy tile roofs were more heavily damaged than more modern home construction, and contents damage in high-tech industrial facilities was particularly problematic for production continuity. Tsunami damage from the Tohoku Earthquake In the period immediately following a disaster, the Japanese government posts running tallies for the damage statistics as they are collected. This data is invaluable for understanding the scale of damage but also provides important insights with regard to drivers of loss. RMS used these damage statistics during the early event response process to help inform economic and insured loss estimates. In subsequent months, more comprehensive damage statistics compiled by Japan’s Ministry of Land, Infrastructure, Transport and Tourism proved vital for refinement of building performance modeling by RMS during strong ground shaking as well as for the development of vulnerability functions for tsunami inundation. Japan has created and maintained what is arguably the best and most dense national seismic network in the world. This network recorded more than 1,000 observations of the ground motions produced by the Tohoku Earthquake. Because large M9+ events are so rare (only five in the last 150 years), this observation dataset is key for understanding the strong ground motions produced by these extremely strong earthquakes. “Prior to this event, modeling of ground motions for events in this magnitude range had to be extrapolated from observation of smaller magnitude events,” says Williams. “Having more data to constrain M9+ ground motions helps refine seismic hazard and risk for all regions that can experience events in this magnitude range. Additionally, the observation data captured the details of the interaction of sedimentary basins and shallow site conditions on ground motion amplitude and frequency content.” The information has allowed RMS to dramatically improve the assessment of site condition impacts (both shallow and deep) to allow for a better assessment of localized interactions of ground motions on structural performance. Following the 2011 events, the Japanese government commissioned a series of comprehensive research studies to better understand earthquake potential for the key subduction zones (i.e., Japan Trench, Kuril Trench, Sagami Trough and Nankai Trough) and key crustal faults. The goal was to extend understanding of the historical record by utilizing the geologic record and providing information on past events over the last several thousand years. Key geologic datasets that were examined included paleotsunami deposits in coastal regions, coastal terraces uplifted in past events and paleoseismic studies to examine past ruptures on faults. The RMS Japan Earthquake Model was informed by all these datasets, allowing for a better representation of the range of events that can occur as well as better constraining the recurrence of future events on these structures. Advances in Tsunami Modeling Prior to the Tohoku event, RMS tsunami solutions had been focused on key tsunami scenario footprints that were developed to allow for an understanding of exposure accumulations at risk. “With the 2011 event and the contribution of approximately 30 percent of the loss from tsunami, it was clear that RMS needed to start providing fully probabilistic tsunami solutions,” said Williams. “The unique characteristics of the Tohoku tsunami event and its generation were key for guiding the RMS tsunami hazard and risk development.” The extremely high fault slip and large ocean bottom deformations highlighted the importance of modeling a range of slip models. RMS has chosen to use analytical slip modeling, and the sampling of alternative slip models for a given earthquake rupture allows for a more comprehensive understanding of tsunami and seismic risk. Tsunami insights from Tohoku also informed tsunami modeling in New Zealand. Following Tohoku, GNS Science, the New Zealand geoscience research institute, updated the maximum magnitude potential for the Hikurangi Subduction Zone to the east of the North Island. This assumption is reflected in the RMS® New Zealand Earthquake HD Model, and when combined with other updates, the larger magnitude has consequential impacts for portfolios with exposure in the capital of Wellington.      Lessons in Liquefaction Residents in Christchurch had certainly felt the initial M7.1 Darfield Earthquake on September 4, 2010, some 40 kilometers (25 miles) west of the city, and power and water supplies were disrupted. The event caused moderate damage, the worst of which was to unreinforced masonry chimneys and walls. Damage was also observed in historic buildings. Following the Darfield event, assessments were made to repair the damaged buildings. However, despite the lower magnitude of the February 2011 earthquake, its proximity almost directly beneath the CBD meant that the ground motions were considerable. The Christchurch Earthquake generated widespread liquefaction and was part of an ongoing sequence of events, the largest of which, following February 2011, were M5.9, M5.5 and M5.3. A number of buildings that had been compromised during the September 2010 quake crumbled under the more intense ground motion of February 22, 2011. “It was the way the sequence moved eastward from Darfield to Christchurch so that it was virtually under the CBD that made it so devastating," said Laura Barksby, product manager at RMS. "It occurred in the wrong place at the wrong time.” The Christchurch event exacerbated preexisting damage, as well as damaging previously unscathed structures. Damage was so severe in some areas of Christchurch that a red zone was established, within which it was considered uneconomical to repair buildings, and structures were demolished regardless of their state. In total, the Canterbury Earthquake Sequence caused 185 fatalities and around NZ$40 billion in economic damage, of which an estimated NZ$33-38 billion was insured. The sudden change in seismicity was traumatic for residents and hampered efforts to assess the damage and begin the rebuild and restoration process. Access inside the CBD was restricted as many older structures, mostly unreinforced masonry buildings, were deemed unsafe. In the years immediately following the earthquake, demolitions outnumbered rebuilds by four to one. Aftermath of the Christchurch Earthquake in 2011 “There has been a huge societal impact. The CBD was cordoned off and many businesses had to close,” says Barksby. “From a community perspective, they went from years of no earthquakes to almost waiting for the next to happen. The fact that the events were part of a sequence added to that sense of nervousness.” The overall headline, she explains, was the damage caused by liquefaction. “When we think about earthquakes, our immediate thoughts are about the ground shaking, but with Christchurch the focus was the liquefaction. It was responsible for around 30 to 40 percent of the losses, which is considerable.” During an earthquake, the ground motions can cause an increase in water pressure in soil layers beneath the ground. This can cause a reduction in the strength of the soil particles, and they subsequently behave like a liquid, which can cause significant ground deformation. In Christchurch, for buildings with shallow foundations, there was significant damage. One aspect that had not been appreciated prior to Christchurch was the scale of the destruction liquefaction could cause — and the loss it could generate. RMS reconnaissance observed that some buildings experienced no shake damage but considerable liquefaction damage. “The damage was particularly bad along the River Avon in Christchurch," says Barksby. "Due to the lateral displacement, it looked as though some of the buildings had been pulled apart — the Christchurch Earthquake really demonstrated the different types of liquefaction displacement.” This represented an important distinguishing feature when modeling liquefaction risk. “What was seen in Christchurch was a notable difference in the damage severity depending on the liquefaction process that had occurred. There was a correlation between the type of liquefaction displacement and building damage,” said Barksby. “Lateral spreading versus vertical displacement can have very different outcomes when it comes to loss. This distinction is not something we were able to capture before Christchurch, but thanks to data we can now model it at a high resolution and directly relate it to damage at a location.” The liquefaction impact was highly variable, a feature best captured by aerial photographs taken in the immediate aftermath. While some streets were largely unscathed, others looked as though they had been inundated by flood waters from liquefaction expressed at the surface. Barksby added, “We also saw streets with the whole spectrum of liquefaction damage, ranging from none at all to severe damage just a few hundred meters down the road.” Geotechnical engineering experts from around the world seized the opportunity to better understand the hazard, using Christchurch as a liquefaction laboratory. Through its collaboration with the Canterbury Geotechnical Database (now known as the New Zealand Geotechnical Database), RMS was able to analyze borehole data along with claims insights in order to better understand how soil characteristics, water table depth and proximity to water courses influenced the exposure. It was also really important to establish the link on how liquefaction translated into damage, as ultimately this was the main area of concern. Given the significant advances in seismic understanding after Christchurch, New Zealand was chosen as the location for the first RMS high-definition (HD) earthquake model. Released in 2016 and updated in 2020, the model leveraged the surge in data available from the Earthquake Commission (EQC) and GNS Science, among others, together with collaboration partners in the insurance market to offer a more granular view of the hazard and vulnerability. The RMS New Zealand Earthquake HD Model was also the first to include an advanced liquefaction module in addition to landslide, fire following earthquake and probabilistic tsunami. ʺWe applied all the lessons from Christchurch to the rest of the country at a more localized level than had been possible before,” says Barksby. “New Zealand was selected for the first high-definition model because we had so much data arising from the sequence that we knew we could leverage HD methodologies and Cloud-computing technology, plus the country has a sophisticated insurance market.” Barksby describes it as a paradigm shift, with the same underlying principles and framework rolled out to improve the granularity and level of hazard and vulnerability detail captured by the other earthquake models, including those for Japan and North America. Striving for a Resilient Future A decade on from Tohoku and Christchurch, communities in Japan and New Zealand are still coming to terms with the tragedies and how the quakes have shaped their lives. While very large earthquakes remain relatively rare, it is important to understand the potential, including from the associated perils. The return period for earthquakes on major faults or subduction zones is hundreds to thousands of years​. Because they are so rare, each earthquake disaster has its own unique characteristics​. The events of 2011 were an opportunity to learn and to continue to push the boundaries of earthquake science and seismic engineering​. The earthquakes provided a unique opportunity to fundamentally improve the scientific and engineering communities’ understanding of earthquakes and their impacts​. RMS has used this opportunity to redefine its perspective on seismic risk in Japan and in New Zealand, and beyond​. Chesley Williams concludes: “At RMS, the goal is ​to implement the best available science​, to understand the limitations of the modeling, to apply appropriate uncertainty assumptions and to ensure that we make the best estimate of seismic risk based on the information we have today.”

Helen YatesMay 10, 2018
Where Tsunami Warnings Are Etched in Stone
May 10, 2018

EXPOSURE looks back at the 2011 Tohoku event and other significant events that have shaped scientific knowledge and understanding of earthquake risk incorporated into the RMS® Japan Earthquake and Tsunami HD Model Hundreds of ancient markers dot the coastline of Japan, some over 600 years old, as a reminder of the danger of tsunami. Today, a new project to construct a 12.5-meter-high seawall stretching nearly 400 kilometers along Japan’s northeast coast is another reminder. Japan is a highly seismically active country and was well prepared for earthquakes and tsunami ahead of the Tohoku Earthquake in 2011. It had strict building codes, protective tsunami barriers, early-warning systems and disaster-response plans. But it was the sheer magnitude, scale and devastation caused by the Tohoku Earthquake and Tsunami that made it stand out from the many thousands of earthquakes that had come before it in modern times. What had not been foreseen in government planning was that an earthquake of this magnitude could occur, nor that it could produce such a sizable tsunami. The Tohoku Earthquake was a magnitude 9.0 event — off the charts as far as the Japanese historical record for earthquakes was concerned. A violent change in the ocean bottom triggered an immense tsunami with waves of up to 40 meters that tore across the northeast coast of the main island of Honshu, traveling up to 10 kilometers inland in the Sendai area. The tsunami breached sea walls and claimed almost everything in its path, taking 16,000 lives (a further 2,000 remain missing, presumed dead) and causing economic losses of US$235 billion. However, while the historical record proved inadequate preparation for the Tohoku event, the geological record shows that events of that magnitude had occurred before records began, explains Mohsen Rahnama, chief risk modeling officer at RMS. “Since the Tohoku event, there’s been a shift … to moving further back in time using a more full consideration of the geological record” Mohsen Rahnama RMS “If you go back in the geological record to 869 in the Tohoku region, there is evidence for a potentially similarly scaled tsunami,” he explains. “Since the Tohoku event, there’s been a shift in the government assessments moving away from a focus on what happened historically to a more full consideration of the geological record.” The geological record, which includes tsunami deposits in coastal lakes and across the Sendai and Ishinomaki plains, shows there were large earthquakes and associated tsunami in A.D. 869, 1611 and 1896. The findings of this research point to the importance of having a fully probabilistic tsunami model at a very high resolution. Rahnama continues: “The Tohoku event really was the ‘perfect’ tsunami hitting the largest exposure concentration at risk to tsunami in Japan. The new RMS tsunami model for Japan includes tsunami events similar to and in a few cases larger than were observed in 2011. Because the exposure in the region is still being rebuilt, the model cannot produce tsunami events with this scale of loss in Tohoku at this time.” Incorporating Secondary Perils RMS has incorporated the lessons from the Tohoku Earthquake and other major earthquakes that have occurred within its model. There have been several large earthquakes around the world, and they all inform thinking about the largest events, particularly how the ground motions they produce are modeled. Crucially, it includes a fully probabilistic tsunami model that is integrated with the earthquake stochastic event set. On understanding strong ground shaking, information across events is utilized. For example, petrochemical facilities around the world are built with relatively consistent construction practices. This means that examination of the damage experienced by these types of facilities in Chile and Japan can inform the understanding of the performance of these facilities in other parts of the world with similar seismic hazard. The Maule Earthquake in Chile in 2010, the Canterbury sequence of earthquakes in New Zealand in 2010 and 2011, and the more recent Kumamoto Earthquakes in Japan in 2016, have added considerably to the data sets. Most notably they have informed scientific understanding of the nature of secondary earthquake perils, including tsunami, fire following earthquake, landslides and liquefaction. The 2016 Kumamoto Earthquake sequence triggered extensive landsliding. The sequence included five events in the range of magnitude 5.5 to 7.0 and caused severe damage in Kumamoto and Oita Prefectures from ground shaking, landsliding, liquefaction and fire following earthquake. “Liquefaction is in the model as a secondary peril. RMS has redesigned and recalibrated the liquefaction model for Japan. The new model directly calculates damage due to vertical deformation due to liquefaction processes,” says Chesley Williams, senior director, product management at RMS. “While the 1964 Niigata Earthquake with its tipped apartment buildings showed that liquefaction damages can be severe in Japan, on a countrywide basis the earthquake risk is driven by the shaking, tsunami and fire following, followed by liquefaction and landslide. For individual exposures, the key driver of the earthquake risk is very site specific, highlighting the importance of high-resolution modeling in Japan.” The RMS model accounts for the clustering of large events on the Nankai Trough. This is an important advancement as an examination of the historical record shows that events on the Nankai Trough have either occurred as full rupturing events (e.g., 1707 Hoei Earthquake) or as pairs of events (e.g., 1944 and 1946 and two events in 1854). This is different from aftershocks, explains Williams. “Clustered events are events on different sources that would have happened in the long-term earthquake record, and the occurrence of one event impacts the timing of the other events. This is a subtle but important distinction. We can model event clustering on the Nankai Trough due to the comprehensive event record informed by both historical events and the geologic record.” The Tohoku event resulted in insurance losses of US$30 billion to US$40 billion, the costliest earthquake event for the insurance industry in history. While the news media focused on the extreme tsunami, the largest proportion of the insurance claims emanated from damage wrought by the strong ground shaking. Interestingly, likely due to cultural constraints, only a relatively low amount of post-event loss amplification was observed. “In general for very large catastrophes, claims costs can exceed the normal cost of settlement due to a unique set of economic, social and operational factors,” says Williams. “Materials and labor become more expensive and claims leakage can be more of an issue, so there are a number of factors that kick in that are captured by the RMS post-event loss amplification modeling. The Japan model explicitly models post-event loss amplification but limits the impacts to be consistent with the observations in recent events in Japan.” Supply chain disruption and contingent business interruption were significant sources of loss following the Tohoku event. This was exacerbated by the level seven meltdown at the Fukushima nuclear power plant that resulted in evacuations, exclusion zones and rolling blackouts. “We sent reconnaissance teams to Japan after the event to understand the characteristics of damage and to undertake case studies for business interruption,” says Williams. “We visited large industrial facilities and talked to them about their downtime, their material requirement and their access to energy sources to better understand what had impacted their ability to get back up and running.” Recent events have re-emphasized that there are significant differences in business interruption by occupancy. “For example,  a semiconductor facility is likely going to have a longer downtime than a cement factory,” says Williams. “The recent events have highlighted the impacts on business interruption for certain occupancies by damage to supply sources. These contingent business interruptions are complex, so examination of the case studies investigated in Japan were instrumental for informing the model.” Rebuilding in the seven years since the Tohoku Tsunami struck has been an exercise in resilient infrastructure. With nearly half a million people left homeless, there has been intense rebuilding to restore services, industry and residential property. US$12 billion has been spent on seawalls alone, replacing the 4-meter breakwaters with 12.5-meter-high tsunami barriers. An endless convoy of trucks has been moving topsoil from the hills to the coastline in order to raise the land by over 10 meters in places. Most cities have decided to elevate by several meters, with a focus on rebuilding commercial premises in exposed areas. Some towns have forbidden the construction of homes in flat areas nearest the coasts and relocated residents to higher ground. Tokyo-Yokohama: The World’s Most Exposed Metropolis The Japanese metropolis of Tokyo-Yokohama has the world’s greatest GDP at risk from natural catastrophes. Home to 38 million residents, it has potential for significant economic losses from multiple perils, but particularly earthquakes. According to Swiss Re it is the riskiest metropolitan area in the world. A combination of strict building codes, land use plans and disaster preparedness have significantly reduced the city’s vulnerability in recent decades. Despite the devastation caused by the tsunami, very few casualties (around 100) related to partial or complete building collapse resulting from ground shaking during the magnitude 9.0 Tohoku Earthquake.  

close button
Overlay Image
Video Title

Thank You

You’ll be contacted by an RMS specialist shortly.