Shake-Up Time for Japanese Seismology Robert J

Home , Seismology

COMMENT MEDICINE Social networking CLIMATE The voyages and FICTION Ecology, PSYCHOLOGY Why are some takes patient-group pressure research patronage of an booze, crime and people cruel and others to new levels p.410 unusual businessman p.413 pig-hunting p.414 compassionate? p.416

Emergency drills such as this mislead the public into believing that the Tokai district is due a magnitude-8 quake soon. Shake-up time for Japanese seismology Robert J. Geller calls on Japan to stop using flawed methods for long-term forecasts and to scrap its system for trying to predict the ‘Tokai earthquake’.

or the past 20 years or so, some The ‘foreseen’ earthquakes were seismologists in Japan have warned presumably the hypothetical future earth- SUMMARY of the seismic and tsunami hazards quakes used by the Japanese government to ● The Japanese government should Fto the safety of nuclear power plants, most produce national seismic hazard maps for admit to the public that earthquakes KYODO/NEWSCOM notably Katsuhiko Ishibashi, now professor Japan1. The modellers assume that ‘charac- cannot be reliably predicted. emeritus at Kobe University. Their warnings teristic earthquakes’ exist for various zones, ● Use of the misleading term ‘Tokai went unheeded. Yet in the immediate after- choose the fault parameters for each zone as earthquake’ should cease. math of the magnitude-9.1 earthquake that the input to their model, and then produce ● The 1978 Large-Scale Earthquake struck Tohoku on 11 March, pundits could probabilistic hazard maps. Countermeasures Act should be be found on many Japanese TV stations Although such maps may seem authori- repealed. saying that it was “unforeseeable”. tative, a model is just a model until the

methods used to produce it have been well-documented 1896 Sanriku tsunami had long-term forecasts verified. The regions assessed as most dan- a maximum height of 38 metres and caused of large earthquakes. gerous are the zones of three hypothetical more than 22,000 deaths. The 869 Jogan The idea was very ‘scenario earthquakes’ (Tokai, Tonankai tsunami is documented to have had a height simple. It was hypoth- and Nankai; see map). However, since roughly comparable to, or perhaps slightly esized that zones 1979, earthquakes that caused 10 or more less than, that of the 11 March tsunami. where no large earth- FUKUSHIMA fatalities in Japan actually occurred in If global seismicity and the historical quakes had occurred CISIS places assigned a relatively low probability. record in Tohoku had been used as the for a while, dubbed WWW.NATURE. This discrepancy — the latest in a string of basis for estimating seismic hazards, the ‘seismic gaps’, were COM/JAPANQUAKE negative results for the characteristic earth- 11 March Tohoku earthquake could eas- ripe for imminent quake model and its cousin, the seismic-gap ily have been ‘foreseen’ in a general way, large events. However, the seismic-gap model2–4 — strongly suggests that the haz- although not of course its particular time, hypothesis failed the test of reality2. Over ard map and the methods used to produce it epicentre or magnitude. Countermeas- tens of thousands of years or longer, the net are flawed and should be discarded. ures for dealing with such events could slip released by earthquakes and aseismic Globally, in the past 100 years, there have and should have been incorporated in the slip must match net inter-plate motion. But been five subduction-zone earthquakes of initial design of the Fukushima nuclear we now know that this catching-up pro- magnitude 9 or greater (Kamchatka 1952, power plants. cess does not occur regularly or cyclically, Chile 1960, Alaska 1964, Sumatra 2004, as is further underscored by the 11 March Tohoku 2011), which suggests that the upper THE ‘TOKAI EARTHQUAKE’ earthquake. limit on the possible size of a subduction- In the 1960s, plate tectonics became generally In the mid-1970s, when enthusiasm for zone earthquake may not much depend accepted as the fundamental paradigm the seismic-gap model was still widespread on the details of the subduction modality5. of solid-Earth geoscience. Researchers in in the global geoscience community, sev- Large tsunamis have frequently struck the several countries made efforts to combine eral researchers in Japan proposed that Pacific coast of the Tohoku district. The plate tectonics with seismicity data to make the plate boundary off the Tokai district was a seismic gap where a magnitude-8 earthquake could be expected6. The neigh- bouring Tonankai and Nankai districts REALITY CHECK were also labelled as being seismic gaps7. The Japanese government publishes a national seismic hazard map like this No large earthquake has occurred in any of every year. But since 1979, earthquakes these districts since 1975, but they are still that have caused 10 or more fatalities in Eurasian classified as the most hazardous regions in Japan have occurred in places it plate designates low risk. the country by the Japanese government (see map).

1993 Over the past 30 years or so, government 1994 spokesmen and university scientists associ- 7.8 (230) 8.2 () ated with the government’s Headquarters for Earthquake Research Promotion (or its various predecessors) have used the term 1983 Fault plane 7.7 (04) ‘Tokai earthquake’ so often that the public and news media have come to view it as 2008 7.2 (23) a ‘real earthquake’ rather than merely an 2011 Tohoku arbitrary scenario (1.78 million hits in a 2007 earthquake 6.8 (5) Magnitude-9. Japanese-language Google search). This (>27,000 dead misleads the public into believing that the 1984 or missing) clock is ticking down inexorably on a mag- 6.8 (29) 2004 6.8 (68) nitude-8 earthquake that is certain to strike 1995 the Tokai district in the near future. Use of 7.3 (6,437) the term ‘Tokai earthquake’ (and its com- Okhotsk plate panions ‘Tonankai earthquake’ and ‘Nankai earthquake’) should therefore cease. Paci c plate UNPREDICTABLE EARTHQUAKES Throughout most of seismological history, the prediction of earthquakes hours or days in advance has, for good reason, Hypothesized 8 Nankai Tonankai Tokai fault planes been regarded with great scepticism (see ADAPTED FROM JAPAN HEADQUARTERS FOR EARTHQUAKE RESEARCH PROMOTION 2010 DATA; OTHER INFORMATION ADDED BY AUTHOR ADDED BY INFORMATION OTHER 2010 DATA; PROMOTION RESEARCH EARTHQUAKE FOR HEADQUARTERS FROM JAPAN ADAPTED Philippine go.nature.com/ahc6nx). However, in the Sea plate late 1960s and early 1970s, several studies, initially by researchers in the Soviet Union, and followed by similarly positive studies from major US institutions, led to a burst 0 0.1 3 6 26 100% of optimism. The editors of Nature wrote Government-designated probability of ground motion of seismic in 1973 that the “situation is in some ways intensity of level ‘6-lower’ or higher (on a 7-maximum intensity 100 km scale) in the 30-year period starting in January 2010 similar to that in 1939 when nuclear fission suddenly became a reality”9. Positive results were also published at roughly the same

408 | NATURE | VOL 472 | 28 APRIL 2011 © 2011 Macmillan Publishers Limited. All rights reserved COMMENT time in Science and some leading speciality journals. JMA The positive reports were based on claims to have observed ‘precursors’ of earthquakes. For example, some studies of the type discussed in Nature’s 1973 arti- cle claimed to have observed decreases of 10–20% in crustal seismic velocities before earthquakes, with the return of the velocities to their normal values being the sign that an earthquake was imminent. But the 1976 earthquake in Tangshan, China, which caused a reported 240,000 fatalities, was not predicted, and by the late 1970s it had become clear to most researchers that the supposed precursors were artefacts. The prediction boom then largely died out, but like many similar examples (such as poly- water and cold fusion), die-hard holdouts in several countries continue to make precursor claims. BASELESS PREDICTION LAW By the mid-1970s, public discussion of the supposedly imminent Tokai earthquake reached quasi-panic levels. This The Japan Meteorological Agency control room conducts monitoring to predict the ‘Tokai earthquake’. was exploited by the Japan Meteorological Agency (JMA) and university scientists, Bulge’, in the United States in the 1970s was scientific knowledge. Finally, as long as the who persuaded the Japanese parliament to later shown to be an artefact8. LECA stays on the books, the government enact the Large-Scale Earthquake Coun- Basing even a large-scale programme of can claim that it is obligated by law to try to termeasures Act (LECA) in 1978. This law observational research on the 1944 data predict the Tokai earthquake. in effect requires the JMA to operate a 24/7 would be uncalled for. It beggars belief, It is time to tell the public frankly that monitoring system to detect precursors then, that the Japanese government operates earthquakes cannot be predicted, to scrap indicating that the ‘Tokai earthquake’ (see a legally binding earthquake-prediction the Tokai prediction system and to repeal map) will occur within up to three days. If system on this basis. The JMA’s official the LECA. All of Japan is at risk from earth- and when signals thought to be precursors home page says (author’s translation): “At quakes, and the present state of seismolog- are ever observed, a panel of five geophysi- present the only place a system for predict- ical science does not allow us to reliably cists will review the data, the JMA director ing earthquakes exists is for a magnitude-8 differentiate the risk level in particular will inform the prime minister, and the earthquake with an epicenter offshore geographic areas. We should instead tell cabinet will then declare a state of emer- Suruga Bay, i.e. the ‘Tokai earthquake’. the public and the government to ‘pre- gency, which will stop almost all activity Science and technology have not progressed pare for the unexpected’10 and do our best in a wide area around the Tokai district. sufficiently to allow other earthquakes to be to communicate both what we know and This law, which has no precedent in any predicted.” But there are many more obser- what we do not. And future basic research other country, pre- vatories now than in 1978. If it really were in seismology must be soundly based on sumes of course that “We should possible to predict the ‘Tokai earthquake’ physics, impartially reviewed, and be led by reliable precursors instead tell the then, surely it would be possible to predict Japan’s top scientists rather than by faceless ■ exist. In particular, on public and the all magnitude-8 earthquakes now. bureaucrats. the basis of one report of a geodetic precur- government TIME FOR OPENNESS Robert J. Geller is in the Department of sor of an earthquake to ‘prepare How is it that the Tokai prediction system Earth and Planetary Science, Graduate in Japan in 1944 (see for the has been in place for more than 30 years, School of Science, University of Tokyo, Tokyo Fig. 2 in ref. 6), geo- unexpected’.” with barely a whimper from most main- 113-0033, Japan. detic slip is the main stream Japanese seismologists? The reasons e–mail: [email protected] target of the JMA observations. The 1944 for this silence are complex. First, many 1. Headquarters for Earthquake Research data, taken far from the epicentral region, researchers have been co-opted in various Promotion, National Seismic Hazard Maps for were interpreted as possibly suggesting ways (such as with funding and commit- Japan (2005); available at http://go.nature.com/ uplift of a few centimetres due to slow slip tee memberships). Second, government yw5e92. 2. Kagan, Y. Y. & Jackson, D. D. J. Geophys. Res. 96, on a deep part of the fault shortly before decisions are nominally reviewed, but 21419–21431 (1991). the main shock. Unfortunately, the data review panels are chosen by bureaucrats of 3. Kagan, Y. Y. Bull. Seism. Soc. Am. 86, 274–285 were measured using antiquated surveying the agency being reviewed. Third, cogent (1996). 4. Newman, A., Schneider, J., Stein, S. & Mendez, A. techniques, and are subject to considerable criticisms do get reported by print media, Seismol. Res. Lett. 72, 647–663 (2001). uncertainty. Nothing of this type has ever but are usually ignored by broadcasters, 5. McCaffrey, R. Geology 36, 163–266 (2008). been observed using Global Positioning so critics don’t get much traction. Fourth, 6. Ando, M. Tectonophysics 25, 69–85 (1975). 7. Ando, M. Tectonophysics 27, 119–140 (1975). System devices or other modern meas- through the ‘press club’ system, the gov- 8. Geller, R. J. Geophys. J. Int. 131, 425–450 (1997). urement techniques. A famous report of a ernment pipes its views directly into the 9. Nature 245, 174 (1973). supposed geodetic precursor, the ‘Palmdale media, often through reporters lacking in 10. Kanamori, H. Seismol. Res. Lett. 66(1), 7–8 (1995).