Underestimating nuclear accident risks: Why are rare events so common?

More than a month has passed since the one-two punch of an earthquake and tsunami added a third dimension to the tragedy in Japan: a major nuclear crisis at the Fukushima Daiichi nuclear power station. The situation remains serious, and radioactivity continues to be released. Yet media reports of the disaster have become sporadic, reduced to headlines on the news ticker at the bottom of television screens, as the world’s attention turns to other events. Over the next year, the impact of the Fukushima disaster on the public’s perception of nuclear power will evolve, with advocates portraying the event as an opportunity to make an indispensable source of energy safer, and critics characterizing it as a final indictment of the dangers of nuclear energy. As this debate develops, the public would be well served by answers to a few simple but critical questions.

In the United States, the Nuclear Regulatory Commission (NRC) and the American Nuclear Society (ANS) will conduct safety reviews of US nuclear facilities. Both the NRC and the ANS have already expressed confidence in the safety of nuclear power plants in the United States, even before completing their reviews. The NRC will carry out two assessments: A near-term review, due by mid-July, will focus on operating reactors and spent fuel pools in the United States, and a longer-term review will deal with broader technical and policy issues. The ANS has created a Special Commission on Fukushima Daiichi that will examine the technical aspects of the event to “help policymakers and the public better understand [the disaster’s] consequences and its lessons for the US nuclear industry.” The quality of these reviews, and the depth to which they probe our understanding of safety, will be important to the final judgment on the benefits and risks of nuclear power.

Critical questions. As with all learning experiences, the usefulness of these reviews depends on the questions that are raised and addressed. The first step in such reviews should be to pose critical questions that will guide the review toward fundamental issues. The questions can be simple, but the answers must be honest and fully accessible to an educated and concerned public. One question in particular demands attention: Why was the actual event in Japan, an earthquake and tsunami, so different from the “credible” event that was expected?

From our perspective as geoscientists, this is the most important question because the definition of the credible event provides the basis against which a nuclear power plant is designed. In the case of the Fukushima Daiichi power station, the magnitude of the earthquake (9.0 on the Richter scale, or M9) and subsequent tsunami (with a reported wave height of 14 meters) exceeded the credible event on which the nuclear power plant’s design was based. The site has six nuclear reactors; three of them were operating at the time of the quake and successfully shut down in response to the ground shaking. Nevertheless, the power station and its spent fuel storage pools were overwhelmed by an event that had not been planned for — a “larger-than-expected” tsunami wave, leading to a sequence of catastrophic failures.

Some experts have since described the tsunami as a “rare” or “exceptional” event that was entirely out of the range of reasonable or credible expectation. But shallow, offshore earthquakes can cause tsunamis, and the height of the tsunami at Daiichi was certainly not unexpected for a 9.0 magnitude earthquake. In addition, there have been three 9.0 magnitude earthquakes during the past decade: Indonesia in 2004, Chile in 2010, and now Japan in 2011. The fact that such earthquakes occur infrequently over historical periods does not explain why the Fukushima nuclear power plant was not designed to withstand this type of geologic event.

Reconsider the definition of “credible.” From a geologic perspective, the earthquake and its great magnitude should not have been a surprise. Ten years ago, Japanese earth scientists, led by Koji Minoura at Tohoku University in Sendai, described a major earthquake and tsunami that happened in July 869 and was recorded in an historical document. This event, which is also clearly recorded in the coastal sediment of the Sendai plain, extended inland about four kilometers from the coast. Based on even older tsunami deposits that go back some 3,000 years, Minoura and his colleagues suggested a 1,000-year recurrence interval for large-scale earthquakes and tsunamis in Japan and presciently published their results in the Journal of Natural Disaster Science.

Their results and conclusions did not go unnoticed. Based on the Minoura et al. paper, Yukinobu Okamura, the director of Japan’s Active Fault and Earthquake Research Center, raised the possibility that a large tsunami could damage the Fukushima Daiichi plant. The plant operator, Tokyo Electric Power Company, dismissed these warnings.

The essential question is: Why were these very clear results and reasonable concerns not included in an updated safety assessment? This question is not meant to point fingers or establish blame, but rather to understand why critical information was excluded from the safety analysis. The oldest nuclear reactor at Daiichi was built 40 years ago; the intervening years have seen remarkable progress and insight into Japan’s tectonic setting, but this new information seems not to have raised much alarm or concern.

Put the risk in a broader perspective. Part of the explanation may lie in the inherent difficulty of reducing geologic data and interpretation to a form that is amenable to the probabilistic risk assessment (PRA) methodology. Probabilistic risk assessments focus on determining the risk that a credible event may pose to a specific nuclear power plant. From this narrow perspective, the risk of a very large tsunami hitting a specific nuclear power plant along the coast of Japan may be very low; however, in a broader geologic context, the risk may be very different.

Nuclear power is vital to Japan’s electricity production and future energy plans. Consider that 54 nuclear power reactors are operating in Japan, with a dozen more planned or under construction. Nuclear reactors generate about 25 percent of Japan’s electricity, and the country’s energy policy calls for doubling the number of reactors — to generate 50 percent of the nation’s electricity. Most of Japan’s reactors are or will be located on the coast. Imagine Japan in 2030, with some 100 reactors — approximately 50 of them along the eastern coast. These reactors, which may be re-licensed or replaced over time, are likely to be essential parts of Japan’s energy landscape for the next several hundred years.

This picture of nuclear power in Japan takes on special significance when it is put into the geologic context of the Japanese islands: Japan sits on the western edge of the Pacific Ring of Fire, a distinct boundary between oceanic and continental plates where some 90 percent of the world’s earthquakes occur. Along the eastern coast of Japan, two tectonic plates are colliding. The rate of convergence (8.3 centimeters per year) is relentless, driving the Pacific plate to plunge westward beneath the Eurasian plate to form a subduction zone along the Japan Trench. When the interface between these two plates suddenly ruptures during an earthquake, a huge volume of water can be displaced—forming a tsunami. Importantly, long periods without major earthquakes may be the precursors to large-scale events in which the accumulated strain from hundreds of years of convergence is suddenly released in a single earthquake.

When the tectonic setting and the distribution of nuclear power plants along the east coast of Japan are taken into account, the probability that a major earthquake and tsunami will strike a nuclear power plant somewhere along the coast increases significantly, particularly when viewed over periods of hundreds of years. Regions of low seismicity may deserve special attention and concern. Distinguished Japanese geoscientist Hiroo Kanamori and colleagues noted in the Geophysical Journal International in 2010:

This [seismic] behaviour raises another important question regarding tsunami potential of subduction zones where no large historical event has been documented (e.g., a segment of subduction zone south of Sanriku in Japan). Because most seismic and tsunami hazard mitigation measures heavily rely on the past experience, such “quiet” subduction zones tend to receive less attention, but slow accumulation of strain in such subduction zones can lead to extremely serious, though infrequent, tsunami hazard, and special attention needs to be paid to such possibilities.

Clearly, assumptions about the probability and intensity of credible events drive the results of a risk analysis. Unfortunately, probabilistic risk assessments have failed to incorporate all of the geologic data available — and to interpret this information in a broad context. As a result, the risk assessment for the nuclear power station at Fukushima Daiichi underestimated the probability and intensity of the actual geologic event. The definition of credible events used for reactor design must consider Japan’s tectonic setting and its past history of major earthquakes and tsunamis.