Foreshocks, main shock, and aftershocks distribution: Spatial distribution of foreshocks, main shock, and aftershocks (a); Temporal distribution of foreshocks, main shock, and aftershocks (b); M-T Diagram of the M7.3 foreshock and other foreshocks stronger than M4.4 prior to the Mw9.0 main earthquake (c) 

Context in source publication

Context 1

... strong earthquake prediction according to a great quantity of intensive small earthquakes is one of the common methods employed in earthquake prediction, and the prediction of the Haicheng Earthquake in China is the most success- ful example (Chen 2009; Xu et al. 1982). But it is difficult to tell whether an earthquake that has occurred is a foreshock of another quake or is itself the main event. On the other hand, due to a lack of foreshocks (Marzocchi and Zechar 2011), no forecast was issued for the Tangshan and Wenchuan Earthquakes, both of which resulted in considerable casualties. On 11 March 2011, a Mw9.0 earthquake happened in the northeast of Japan. Before this Mw9.0 earthquake, a M7.3 earthquake occurred on March 9 in the same place. The M7.3 earthquake did not attract much attention due to its occurrence under the sea over 130 km away from the shore. In this location, the M7.3 quake caused neither severe destruction nor a devastating tsunami, although the quake belonged in the strong earthquake category in terms of magnitude. By examining this earthquake and the subsequent earthquake sequence, researchers later concluded that these earthquakes were actually a foreshock sequence of a greater earthquake rather than a typical aftershock sequence of the M7.3 earthquake (He, Zhou, and Ma 2011; Ozawa et al. 2011). Why was the M7.3 earthquake not recognized earlier as a foreshock? Seismologists attributed this to the lack of a history of great earthquakes in northeastern Japan. Thus the prediction of such an earthquake went beyond the cognitive range of their seismic activities. Similarly, since no such strong earthquake occurred in the Longmenshan area before the Wenchuan Earthquake in China in 2008 (Wen et al. 2009), seismologists took it for granted that there would be no strong earthquake in the future in the area, and thus paid little attention to its possible occurrence. According to historical earthquake catalogues, earthquakes have occurred periodically. In China, there have been five active seismic periods since 1895, marked by the occurrence of large earthquakes (Zhang, Fu, and Gui 2001). Period of seismic activity is often used to predict the future trend of earthquakes. Internationally, Geller and colleagues (1997) and Sykes, Shaw, and Scholz (1999) hold somewhat opposite views. According to the self-organized critical (SOC) phenomenon, Geller and colleagues believe that earthquakes cannot be predicted. But Sykes, Shaw, and Scholz think that at a certain scale large earthquakes can be predicted. Predicting the future trend of earthquakes according to the quasi-period of historical earthquakes clearly involves great uncertainties. In practice, prediction according to the causative rules of historical earthquakes is one of the common methods for predicting middle- and long-term earthquakes (Wang 2009). But it is unavoidable for such a method to fail in predicting great earthquakes with an especially long causative cycle. This article discusses the limitations of predicating earthquakes based on historical earthquakes by analyzing the foreshock characteristics and historical record of earthquakes in eastern Japan. On 9 March 2011, a M7.3 earthquake occurred in northeastern Japan off the Sanriku coast; it was accompanied by active aftershocks including a M6.8 event the next day. These events were located just north of the Pacific Ocean center of the Tohoku Earthquake, which took place two days later on 11 March 2011 (Figure 1a). Since the M7.3 earthquake was a significant earthquake event in its own rights, based on their experience seismologists took it as the principal earthquake. Because it caused no damage, the M7.3 earthquake received little attention from the research community, government, and general public. But the cruel fact was that this earthquake and the subsequent earthquake sequence associated with it were actually a foreshock sequence of a larger magnitude earthquake rather than a typical aftershock sequence to the M7.3 earthquake (Figure 1b). Figure 1c shows that the energy had already been in a gradual attenuation situation more than half a day after the M7.3 earthquake, and the maximum magnitude of quakes did not surpass M6.0. However, three earthquakes stronger than M6.0 happened during the more than three hour period from 18:06 to 21:22 on March 9, which made the earthquake energy release rate increase rapidly. From that point on another five earthquakes stronger than M5.0 took place, although with a somewhat reduced energy release rate. The tail end of the curve in Figure 1c is the occurrence time of the Mw9.0 earthquake. Figure 1c shows that although the energy release rate before the great earthquake was lower compared with immediately after the M7.3 quake, it was absolutely not completely calm (He, Zhou, and Ma 2011). Compared with the foreshocks, there were intensive strong aftershocks for several days after the Mw9.0 great earthquake (Figure 2), which were embodied in crowed lines in the M-T diagram (Figure 1b), while the frequency of strong aftershocks decreased gradually (Figure 3a). The variation tendency of the energy release rate is basically degraded without obvious fluctuation in the foreshock sequence (Figure 3b). These earthquake sequences show that the M7.3 earthquake was not the principal earthquake, but part of a foreshock sequence. Through analyzing sequences of historical earthquakes and probabilities of future earthquakes predicted by Japanese seismologists prior to the great earthquake based on historical earthquakes, this article explains why seismologists mistook the M7.3 earthquake as a principal earthquake. The article also discusses the limitations of predicating earthquake according to the data provided by foreshocks and the historical earthquake record. The Tohoku Earthquake happened in the sea off the coast of Sanriku in northeastern Japan, which belongs to the Pacific seismic and volcanic activity zone (He, Zhou, and Ma 2011). Northeast Japan is located at the subduction zone of the Pacific Plate as it approaches the Japanese archipelago, with the subduction zone forming the Japan Trench (Figure 4). Southeast Japan is located at the subduction zone of the Phillipine Plate as it sinks under the Japanese archipelago, which stretches southward from Izu Peninsula to Shikoku Island. The Nankai Trough of Japan is formed at the boundary of the subduction zone (Figure 4a). The boundary between these two subduction zones is called the Sagami Trough and the Lzu-Ogasawara Trench. According to the characteristics of historical seismicity, the eastern coastal area of the Japanese islands can be divided into four earthquake zones from the north to south, that is, the Sanriku area and its near seas earthquake zone, Miyagi area and its adjacent marine earthquake zone, Kanto area and its offshore earthquake zone, and the Nankai trough earthquake zone. Based on the spatial and temporal patterns of historical earthquakes in these areas, Japanese seismologists make predictions about the probability of occurrence of future earthquakes (Okada 2011). The oldest strong earthquake in history in Sanriku happened in 869, and the next strong earthquake occurred in 1611. Thereafter, three M8.0 or greater earthquakes with accompa- nying tsunami happened in 1677, 1896, and 1933 respectivel y (Figure 5a). So the tectonics in the offshore areas of both Sanriku and Fukushima provide the necessary conditions for the occurrence of M8.0 or greater earthquakes. According to the predictions based on historical earthquakes, the probability of occurrence of M8.0 earthquakes in northern and central Sanriku and its nearby seas is about 0.5–10 percent in the 97 years after 1933. In contrast, the probability of the occurrence of M7.7 earthquakes centered offshore from southern Sanriku is 80–90 percent in the future 105 years (Okada 2011) (Figure 5a). According to the energy release diagram of historical earthquakes (Figure 5b), this prediction corresponds with the basic patterns of energy release. The time sequence and energy release of the last four earthquakes suggest that the interval between the first two earthquakes (one cluster) and the last two earthquakes (another cluster) since 1611 is short, about a few decades, while the interval between the two clusters is about 200–300 years. This implies that there may exist two periods for the seismicity in this area: one is short, about 50–100 years, and the other is long, about 200– 300 years. Both periods are important for understanding the occurrence of earthquakes in the area. A M8.2 earthquake occurred in 1793 in Miyagi and nearby seas, and a series of M7.4 and above earthquakes followed this quake until 1978. The average interval between these quakes was about 37 years (Figure 6a). When the earthquake of M7.2 happened in 2005, Japanese seismologists thought that the energy did not reach the historical earthquake level. When the M7.3 earthquake broke out on 9 March 2011, almost all seismologists believed that it was related to the earthquake in 2005 (Figure 6b). Therefore the experts thought the energy had been basically released. Based on historical earthquakes, Japanese seismologists estimated that the probability of a M7.5 earthquake reoccurring in Miyagi and offshore was 99 percent. But if Miyagi marine areas are linked to South Sanriku, M8.0 earthquakes with the same probability of 99 percent are possible (Okada 2011). The evaluation and prediction were reasonable from the perspective of historical earthquakes. The truth, however, was that the Mw9.0 great earthquake broke out just two days after the M7.3 earthquake. The Mw9.0 earthquake was not only the manifestation of the linkage of Miyagi and near seas with South Sanriku, but it also revealed the linkage of the entire northeast ocean trench of Japan because these areas formed a crack of 400 km along the ocean trench (Ozawa et al. 2011). The ...