Almost every destructive earthquake opens social and political fault lines as well as natural ones, and those of the Great East Japan Earthquake of March 2011 are notable only for their local depth and global circulation. Seismicity is a geographically selective concern, but because radiation is involved in the present instance, the stake-holders in debates surrounding the earthquake are much more numerous and spread over a far wider map. In fact seismicity, as opposed to radiation, has receded in most contemporary discussions of the earthquake's aftermath, just as “Sanriku,” the area most damaged by the tsunami, has been displaced by “Fukushima” as shorthand for the event as a whole.
While “Fukushima” stands for the latest (and epic) failure of the Japanese nuclear power industry to operate safely, a related crisis has been less discussed outside expert circles—the failure of the science of seismology to predict or even model the largest earthquake in Japanese history. Were it not for the meltdown of the power plant and the breaching of the anti-tsunami sea walls, media attention might be focused on the shortcomings of scientists' computer models in calculating the probabilities of earthquake hazards. These models failed to predict the location, timing, and magnitude of the event, despite Japan's having the world's longest-running and arguably strongest seismological research program, and one uniquely committed to prediction and societal protection. This was, moreover, no isolated failure. Japan's last “great earthquake,” which struck Kobe in 1995, was likewise not predicted by computer models or hazard maps. If popular faith in engineering has been shaken by these last two seismic disasters, then faith in science has no firmer ground to stand on.
While the Fukushima plant disaster may or may not involve engineering negligence, that word is too strong to apply to earthquake science, which does its best to model and predict a notoriously elusive natural phenomenon. Nonetheless, the president of the Seismological Society of Japan, Hirahara Kazurō, publicly accepted responsibility for his profession's failings in August 2011, telling the Asahi Shimbun, “There are many excuses we can make, but it amounted to a defeat for us. The only thing we can say is that it was beyond our expectations” (Segawa Reference Segawa Shigeko2011). While he and his colleagues expected large seismic events to regularly recur along the Tōhoku (northeast) coast, they greatly underestimated the magnitude of this one, and had given much greater attention to a fault line farther south, between Tokyo and Nagoya, as the likely location of the next great rupture (Cyranoski Reference Cyranoski2011; Kerr Reference Kerr2011; Monastersky Reference Monastersky2011). The “Tōkai Earthquake,” as the long-anticipated and much-modeled south-central event has been known since the 1970s, now seems a phantom in comparison to the stark, brutal, and costly reality of the Great East Japan Earthquake of 2011.
This essay will shift attention from the failure of Japanese nuclear plant engineering to consider the less-noticed crisis for Japanese seismology, a profession whose birth in the Meiji era I have traced previously (Clancey Reference Clancey2006a, Reference Clancey2006b). In doing so, I hope not to add to the existing discomfort of Japan's earth scientists, nor to imply criticism of their efforts or techniques. As a historian of Japanese earthquakes and the discourses and practices that grew up around them, my aim is to put this latest crisis in a longer historical context, and more specifically to try to make sense of how prediction emerged as so vital a component of Japanese seismology, yet one so elusive and, ultimately, so punishing to disciplinary and societal hopes and expectations. Japan has the longest and arguably most sophisticated tradition of instrumental seismology in the world, so if the profession feels “defeated,” to use the language of Dr. Hirahara, then it is not just a national or regional setback, but has repercussions for science globally, and earth science in particular.
The Predictive Imperative
Popular expectations to the contrary, not all seismology has been concerned with earthquake prediction or even the protection of populations. Seismologists in northern Europe, where destructive earthquakes were unknown, were instrumental in organizing an international association in 1901–03 around the exciting prospect that tracking seismic waves could help reveal the internal structure of the Earth. This and other “pure science” projects (and applied ones having to do with nuclear test ban treaties and oil exploration) drove global seismology for much of the twentieth century, and continue to drive much of it today. One could (and still can) be a seismologist and contribute to the protection of vulnerable populations only indirectly or in the abstract, through the publication of highly quantitative articles modeling the movement of waves and the mechanics of subterranean geological features. Even the most well known of Western seismologists, Charles E. Richter, could dismiss prediction altogether as the province of “fools and charlatans,” without jeopardizing public funding for his science (Sholz Reference Sholz1997).
Japanese seismology, however, which was the first scientific community in the world organized under that name, held prediction as an explicit and principal goal from its inception. A unique collaborative effort between expatriate European (mostly British) scientists and their Japanese graduate students in late nineteenth-century Tokyo, Japanese seismology gave birth to the seismograph, a new device that held out to its inventors the possibility of evolving into a predictive instrument. By the first quarter of the twentieth century, when their European teachers had returned home or retired, Japanese seismologists like Ōmori Fusakichi were making global reputations through intercontinental fieldwork and the publication (in Tokyo) of hundreds of empirical and theoretical works in perfect English, often lavishly but precisely illustrated with isoseismal maps, graphs, and seismograms. Backed by a local network of 700 meteorological observation stations that recorded over a thousand earthquakes a year, and with seismologists visiting seismic zones as far afield as Italy, India, and California, Japanese seismology was the most elaborate and productive center of earthquake research in the world at the turn of the twentieth century (Clancey Reference Clancey2006a; Kim Reference Kim Boumsoung2007).
Prediction was central to the Japanese research program, being of major concern to the Japanese government, which funded it, and to a vulnerable national population. The first high-level multidisciplinary research council in Japanese science, founded in 1892, was named the Board of Investigation for Preventing Earthquake Disasters (known more commonly in English as the Imperial Earthquake Investigation Committee or IEIC). It included physicists, geologists, and architects as well as seismologists. With access to so much data (an average of four earthquakes a day in the archipelago), and a widespread network to record and interpret it, its members had faith in the inevitability of prediction and the prestige that would accrue for the profession at home and abroad. This faith was temporarily shaken by the Great Kantō Earthquake of 1923, which destroyed the national capital and caused an incredulous government to abolish the IEIC and restructure seismology root and branch, putting a naval architect from Mitsubishi in charge of a new Earthquake Research Institute and turning decisively to math and physics (Kim Reference Kim Boumsoung2007). This did not, however, shake the determination of either the government or its scientists that prediction should remain a goal, though fortification through “engineering seismology” (now called earthquake engineering) commanded equal or greater attention as a mitigation strategy (Clancey Reference Clancey2006a).
The major theoretical fruit of early Japanese efforts to predict earthquakes was Ōmori Fusakichi's “gap theory,” which continues to form the basis for global earthquake hazard assessment today. Ōmori postulated that one can predict where an earthquake will likely occur next along a known fault by first mapping the locations of all previous earthquakes along the same line. The “gaps” left in the map—the areas along the fault that had been seismically quiet the longest—were likely to be the next sections to shift or release strain. Ōmori used this technique to successfully predict the locations of future earthquakes along a known fault in southern Italy, and went on to apply the same methodology in Japan and around the Pacific Rim (Clancey Reference Clancey2006a; Ikegami Reference Ryohei1981; Imamura Reference Akitsune1926; Valensise and Pantosti Reference Valensise and Pantosti2001). It was on the strength of this and other contributions that the American board of inquiry that convened following the San Francisco Earthquake of 1906 called Ōmori and other Japanese into close collaboration. Hailed by one San Francisco newspaper as “the world's greatest seismologist,” Ōmori declared the city safe for the next few decades, the gap theory having overturned earlier assumptions that epicenters were more or less fixed.
It was a dispute over the application of the gap theory to the undersea section of a fault near Tokyo in the early twentieth century that caused an infamous rift between Ōmori and a close colleague and rival, Imamura Akitsune. When Imamura predicted in print that a gap under Sagami Bay would likely be the source of the next great earthquake, Ōmori responded skeptically and accused the junior seismologist of grandstanding. It was a matter of evidence rather than a dispute over the theory, Ōmori contending that the history of the underwater fault was not adequately understood. Imamura was vindicated in November 1923 when the Great Kantō Earthquake leveled Tokyo and Yokohama, originating precisely in the gap he had located nearly twenty years before (Aki Reference Keiiti1980; Clancey Reference Clancey2006a; Ikegami Reference Ryohei1981). Ōmori was disgraced and Imamura widely celebrated. Imamura's reputation was most recently burnished by an NHK bio-drama produced for broadcast on “Disaster Preparedness Day” in September 2011. So badly did Ōmori's reputation suffer for his failure to predict the capital's destruction (he had gone so far as to declare Tokyo safe) that the gap theory itself has since been more often attributed to Imamura (or to Western scientists who later explained its mechanism), and the significant contributions of Ōmori and his colleagues before 1923 have been largely forgotten.
It is not surprising that the gap theory originated in Japanese rather than Western seismology because prediction was accorded such high priority within that community. Japanese scientists also had a uniquely long historical record of earthquake observation to draw upon when mapping faults, something lacking among colleagues in the United States and northern Europe, or European seismologists who studied earthquakes in colonies like India. “Historical seismology,” or research in written sources to determine the date, location, and intensity of previous earthquakes, thus became an essential aid and corollary to the gap theory, and characteristic of Japanese science. In other parts of the world, such as the American west coast, the discipline of paleo-seismology would later have to be invented in order to provide the same information that in Japan was largely available through written records.
While the gap theory helped identify the locations of future earthquakes, it gave little clue as to their exact timing or magnitude. Given that thirty to forty years is a short cycle for seismically active faults, and many run over centuries, seismologists (and stakeholders within the proposed disaster zone) could wait a lifetime for locational theories to be vindicated or disproven. The theory also required that one knew where faults were, and only the largest, most obvious, and most active ones had actually been charted in Ōmori's lifetime. Many remained elusive until they revealed their presence in a major earthquake, as happened under Kobe in 1995. The gap theory was never a panacea, in other words, and did not satisfy anyone's criteria as the endpoint of earthquake prediction.
For this reason, Ōmori and later generations of Japanese scientists also studied “precursor” phenomena, which might manifest just prior to an earthquake as a result of subterranean movement, or even constitute its trigger. In the former category was the behavior of animals (particularly pheasants, eels, and catfish) and variations in water tables, while in the latter were changes in the atmosphere, lunar variation, magnetism, and electrical ground currents (Clancey Reference Clancey2006a; Terada and Matuzawa Reference Torahiko and Matuzawa1926). At least through the mid-twentieth century, most seismologists in Japan and throughout the world believed that such precursors existed, and often included them in their research programs, though their heterogeneity and perceived location “on the border” of seismology (whose center was increasingly mathematical) and other disciplines meant they were rarely pursued with vigor. In lecturing at Stanford, UC Berkeley, Caltech, and MIT in 1930, Japanese seismologist Suyehiro Kyoji mentioned that “it is very probable that a minor cyclone which passed over the earthquake zone shortly before the occurrence had acted as a trigger that precipitated the destructive 1923 [Great Kantō] earthquake.” In nearly the same breath, however, he dismissed “meteorological phenomena” as “not essential data for earthquake prediction” because they were not reliable (i.e., “the intensity of the induced earthquake” may have no relation to the trigger, and not all storms cause earthquakes). So while Suyehiro (and presumably his audiences) accepted that “certain meteorological phenomena do unmistakenly bring about an earthquake,” he seemed content to leave the investigation of their mechanism to meteorologists (Suyehiro Reference Kyoji1934). At the end of the twentieth century, such open-ended curiosity would increasingly disappear from the discipline as it began to suggest that possibilities it had not actually tested were in fact not testable, or that they had been somehow disproven rather than simply dropped.
The Tōkai Gap
Imamura was preoccupied in his later career with the Nankai Trough, a major undersea fault between two tectonic plates running to the south of Japan from Kyushu to the center of Honshu. In 1944 a major earthquake and tsunami in the center of this fault convinced him that a second such event would occur at a “gap” at the southern end, opposite Shikoku, which indeed happened in 1946. Imamura was guided by the awareness of two similar great earthquakes in the same region in 1854, and records that indicated a cycle of such events every hundred years or more, extending back at least ten centuries (Kanamori Reference Hiroo, Simpson and Richards1980). With the completion of the cycle again in 1944–46, there remained one segment of the fault, named the “Tōkai gap,” which now stood out as the likely candidate for the next major rupture. This gap extended from opposite the city of Nagoya to the Izu Peninsula just south of Tokyo. A major earthquake and tsunami in this area would presumably cause even greater devastation and casualties than the previous pair of 1944–46. Serious attention to this and other seismic risks had to await Japan's postwar recovery, however.
In Japan and globally, seismology was reinvigorated in the 1960s with the acceptance of the plate tectonic theory and the related sense, given a new explanatory model of crustal movement, that prediction was closer to being realized. For Japanese researchers, the Tōkai gap became the natural (and national) laboratory in which prediction would be tested and proven. Beginning in the early 1970s, batteries of instruments and large numbers of seismologists concentrated their attention on this one region as offering the best hope that precursor phenomena would be detected, and a general warning thus issued ahead of the expected disaster. Seismologist Ishibashi Katsuhiko coined the term “Tōkai Earthquake” in 1976 to give the theoretical event more reality, and in 1978 the Japanese parliament passed the Large-Scale Earthquake Countermeasures Act, which set up an official monitoring and preparedness regime for the Tōkai region that was without precedent in Japan or the world. It established a committee of six, initially headed by seismologist Mogi Kiyoo, to be quickly convened if precursory events were detected in the Tōkai area, with the responsibility of advising the prime minister on whether to issue an evacuation order. Here was the ultimate model of emergency preparedness for a sudden-onset event, but one restricted to the single region that experts thought most likely to experience disaster (Kanamori Reference Hiroo, Lee, Kanamori, Jennings and Kisslinger2003).
Enthusiasm for prediction was also sustained by other events occurring outside Japan. The Chinese government had announced in 1975 that the city of Haicheng had been successfully evacuated before a destructive earthquake, and many lives saved, based on precursor phenomena (Chen et al. Reference Yong, Tsoi, Feibi, Zhenhuan, Qijia and Zhangli1988). Although this was met with skepticism by some overseas, and the failure to detect precursors of the Tangshen earthquake the following year set back even the Chinese program, China's commitment to prediction was clear. A similar program would be pursued in the USSR, and was increasingly demanded in newly independent former colonies and “third world countries” whose burgeoning urban areas made seismic vulnerability a concern that even first-world scientists could not ignore (Simpson and Richards Reference Simpson and Richards1980). The Japanese, with their long interest in prediction and their unique status as Asian leaders in this “Western” science, were in a good position to promote their style of seismology in a postcolonial world. And if seismology could be marshaled for the Cold War project of detecting nuclear explosions, it was reasoned, why not reaffirm the discipline's commitment to the safety of cities?
The Great Hanshin (Kobe) Earthquake and the Crisis of Prediction
The problem with the Tōkai Earthquake, of course, is that as of 2011 it is long overdue. On the other hand, there have since been two “great earthquakes” with large numbers of casualties in other places—Kobe, and now Tōhoku—which had been assigned lower probability estimates by the central authorities (1–8 percent and 20 percent respectively, as compared with over 80 percent for Tōkai). The fault beneath the city of Kobe had been designated in 1981 as “precautionary,” but because it had not moved within historic time, there were no “gaps” that might have been identified along its length (Sholz Reference Sholz1997). Thinking the city at low risk from an earthquake, the local and national authorities had done relatively little to prepare themselves or the citizenry. This was apparent in the tardy relief effort that followed, in which the local yakuza famously established relief stations in advance of the government.
One moral of the Ōmori-Imamura story is how dangerous it is for earthquake scientists to make predictions if the consequence is an unexpected disaster, or one in the wrong place. Ōmori's illustrious career was virtually wiped from the record and long forgotten because of a predictive mistake. Similarly, the failure of seismologists to issue clear warnings about the fault under Kobe before it roared to life in 1995 brought widespread public censure upon the profession. The effects, according to seismologist Ishibashi Katsuhiko, were “drastic.” Budgets for research into prediction were slashed—with Ministry of Education funding suffering a 60 percent cut—as more resources were directed toward earthquake engineering through Japan's Science and Technology Agency. “The term ‘prediction’ is taboo these days,” seismologist Megumi Mizoue told a reporter for an American news magazine in 1997. The reporter seemed to take satisfaction that the Japanese were “now starting to concur” with “seismologists elsewhere” that prediction was a waste of resources—resources that could be diverted to fortification (Kunii Reference Kunii1997). But the failure of expressways and other structures in Kobe masked the reality that Japan, arguably the birthplace of modern earthquake engineering, was even then the best-fortified urban society in any seismic zone. The destruction in Kobe was the result of the extreme proximity of the epicenter to the city, the fact that much of the land was fill, and that, for these and related reasons, the waves produced almost unprecedented levels of vertical movement, as opposed to the horizontal shaking that earthquake engineering is normally prepared to resist. It was, in other words, an unusual scenario that many Japanese experts thought did not warrant root-and-branch reform, as had happened after the Great Kantō Earthquake. But then again, the idea of a “characteristic” or even predictable earthquake was increasingly seen to be untenable even by some inside the profession.
The seeming failure of earthquake forecasting in Japan, described in a Time article as the “last bastion” of prediction research, helped fuel what one scientist called a “crusade” against prediction within seismology more generally (Kunii Reference Kunii1997; Main Reference Main1999). This was largely led by a Tokyo University–based American seismologist, Robert Geller. Geller used the Japanese case to argue that not only had the predictive project failed, but also that it may be theoretically and practically impossible because earthquake triggers were too complex if not entirely chaotic (Geller et al. Reference Geller, Jackson, Kagan and Mulargia1997). In a Nature-sponsored debate in 1999, Geller raised the ire of other seismologists, still involved in precursor research, by calling for their defunding if not defrocking (Main Reference Main1999). Whatever the merits of individual arguments, the debate reflected anxiety within the community of seismologists over its boundary with Richter's “frauds and charlatans” (the specter of the “pseudo-scientist”) and its status vis-à-vis fields like astronomy and physics, whose objects of study were perceived to have yielded “harder” and more reproducible results, and ones freer from social and political entanglement (what Geller referred to as the “sociological rather than scientific”). Treating prediction as a residual “social” limb was intended to usher in a new paradigm, and Geller at one point explicitly referenced Thomas Kuhn. While to proponents this was simply realism, for others it was a retreat from curiosity (Wyss Reference Wyss1997), and for yet others it was a strategic abandonment of social responsibility within a science that had long claimed a social function. To Russian scientist E. N. Khalilov, Geller's rejection of prediction was allowing “seismologist-pessimists” to “scientifically avow their failures” (Khalilov Reference Khalilov2008).
The Great East Japan Earthquake and Seismologist-Cassandras
Most destructive earthquakes and disasters of all kinds have their “Cassandras,” figures who through luck, insight, or both are remembered as warning voices. Imamura played that role in the Great Kantō Earthquake. If the spirit of Imamura is present in any contemporary Japanese seismologist, the likely candidate is Ishibashi Katsuhiko, the Tokyo University seismologist who had given the Tōkai Earthquake its name in 1976.
In the aftermath of Kobe, around 1997, Ishibashi coined a second term, “genpatsu-shinsai” (nuclear earthquake disaster) to describe the nightmare scenario that would eventually unfold at Fukushima. This new awareness of the vulnerabilities of nuclear plants was clearly a fruit of the Kobe disaster, especially the collapse of the concrete Hanshin Expressway and other civil engineering icons, which Ishibashi had warned about even before the earthquake (and been censured for). Thus at the moment that earthquake prediction was being called into question, Ishibashi channeled the predictive imperative into additional warnings about earthquake engineering (Clenfield Reference Clenfield2011). In one sense this was the latest chapter in a long and complicated rivalry between seismologists and architects/engineers, dating from the Meiji period, though contemporary seismologists, unlike their Meiji predecessors, had been partly complicit by modeling “characteristic” earthquakes for particular regions (Clancey Reference Clancey2006a).
The initial focus of concern for Ishibashi in the 1990s was not Fukushima, but the Hamaoka nuclear plant in Shizouka Prefecture, located squarely within the zone of the future Tōkai Earthquake. In 2006, however, Ishibashi resigned in protest from a committee considering the upgrading of building codes for nuclear power plants generally, and wrote an Asahi Shimbun article a year later that would seem particularly prescient after March 2011 (Ishibashi 2007). Most of Japan's nuclear plants had been built in a period of relative seismic quietude, he pointed out, but with the Great Hanshin (Kobe) quake, Japan “entered a period of brisk seismic activity” that was likely to continue for forty or more years. Kobe had also taught that “even perfect seismic research could not discover [all] active fault lines,” meaning anywhere in the country was a potential epicentral zone for an earthquake of magnitude 7 or greater. All nuclear plants would have to be upgraded to withstand events of this magnitude, he concluded, regardless of location. Those that could not be upgraded should be shut down (Ishibashi Reference Katsuhiko2007).
Ishibashi remained a believer in the gap theory and the Tōkai Earthquake, and he continued the tradition of nineteenth-century seismologists in considering “historical seismology” (knowing when previous earthquakes occurred and where) a crucial branch of his profession. Another article he wrote in 2004 reiterated this need in the context of the new post-Kobe sensitivity to hidden or neglected faults, which might best be discovered through documents, supplemented by the new sub-discipline of paleo-seismology to reveal physical signs of ancient earthquake and tsunami damage (Ishibashi Reference Katsuhiko2004).
In fact Japanese paleo-seismologists would emerge as a second group of Cassandras in the aftermath of the Great East Japan quake. In a 2001 article, Tōhoku University geologist Minoura Koji and an interdisciplinary team of colleagues had reported physical evidence that a tsunami of 869 CE had inundated the Sendai Plain to a depth of 4 kilometers, or twice as far as tsunamis of more modern times. This Jōgan tsunami (named after the era in which it occurred) had been recorded and was thus well-known to historical seismology (Imamura Reference Akitsune1934; Smits Reference Smits2011), but the report by Minoura et al. (and subsequent paleo-seismological reports published through 2008) confirmed a magnitude and effect that, in hindsight, made it the true predecessor of the March 2011 event (Minoura et al. Reference Minoura, Imamura, Sugawara, Kono and Iwashita2001; Sawai et al. Reference Yuki, Fujii, Fujiwara, Kamataki, Komatsubara, Okamura, Satake and Shinshikura2008). While it took the Great East Japan Earthquake to reveal what may be a 1,000- to 1,500-year cycle of mega-tsunamis in this region, Minoura et al. had demonstrated well beforehand that Japanese seismicity was capable of far greater destruction than any local fortification was designed to bear.
Japanese Seismology at the Crossroads
Despite Hirahara Kazurō's admission of Japanese seismology's “defeat,” quoted at the beginning of this essay, the scientific enterprise clearly will not be shut down, and more rather than fewer resources likely will be put into the effort in the aftermath of March 2011. The question is whether the reforms will be root and branch, as favored by voices like Robert Geller, or whether Japanese seismology and the institutions around it will retain historic coherencies, theoretically as well as institutionally (Geller Reference Geller2011; Gilhooly Reference Gilhooly2011). Seismologists in the early twenty-first century might be compared to economists, who elicit frustration and dislike outside their own community when they predict the future wrongly, as they often do, but whose expertise cannot be fully dispensed with. We may hope they will find a way for us to achieve stability and prosperity. But we cannot make them, and we are constantly suspicious as to whether they are actually working for our safety or some more theoretical goal.
The future of prediction is likewise uncertain, though the debate over its efficacy, and possibility, has never been so clearly drawn. The public (and some seismologists) will certainly want more research put into forecasting. Other seismologists will want to wash their hands of it. The Tōkai Earthquake has not happened, and may not happen any time soon, but who at this stage can abandon the possibility that it might? If Ishibashi is right that Japan is in the midst of a heightened seismic period, and that all regions of the country are at risk, then fortification, awareness, monitoring, and even historical investigation will have to be stepped up on all fronts. What long-term effect this will have on institutions, landscapes, and psyches is anyone's guess. But in the short-term, it is sure to quicken the Japanese sense of vulnerability. Seismology has never been a rock Japan could stand on, but it has been a bastion of hope and a source of some pride, despite historic setbacks. At this moment in time it needs a few victories.
