Next, research was done on how a tsunami, which propagated into both channels changes in a channel. A tsunami that propagates into channels, it is deformed through refraction, wave shoaling and reflection. Also, through the relationship with the natural period of the bay or the sea area, resonance or decay will occur, sometimes causing huge tsunamis at localized areas. Therefore, we carried out research on the period response characteristics of a tsunami in both channels. First, the tsunami profiles at the points shown in Figure 8 were acquired by numerical simulation. Then we analyzed the spectrum of the acquired tsunami profile and acquired the amplification rate ((1)/(2)), where (1) is the spectrum at each point and (2) is the spectrum at the tsunami source. In addition, from this we examined the response characteristics of each periodic wave. The numerical simulation of the tsunami was done by analyzing nonlinear long waves theory equation using the LeapFrog difference method, with complete reflection at the land area. The calculation-grating interval was 600 m, calculation time was 6 hours and the calculation time interval was 1 second. The wave profiles of the tsunamis that we used are represented as sine waves of 2 m in wave height and 1 m in wave amplitude. The range of wave period is from 10 to 60 minutes with a 5-minute interval.

　

　

Figure 9 shows the distribution of the amplification rate of periodic waves in the Kii channel. From this figure, it can be seen that in K2, K4, K5, the peak of amplification appears at the time of 25 and 50 minutes, and that these are the natural period of the Kii channel. In K2, the rate of amplification for 25 minutes becomes larger than the rate of amplification for 50 minutes. In the Naruto straits where it is located in the inner part of a bay, and width is narrower than the Kitan straits, the rate of amplification for 50 minutes becomes larger than the rate of amplification for 25 minutes. We have done spectral analysis on the observation record of Kamchatka earthquake tsunami observed at K8: Nasa bay in 1977 and found that the resonant period at Nasa bay is around 21 and 11 minutes. This verifies the tendency that outside of the K2: inlet of the channel, periodic waves of 20-25 minutes is amplified.

　

　

In Figure 9, in order to examine the response characteristics of the tsunami at coastal areas of the Kii channel, we showed the amplification rates of each periodic wave at K6 and K7. The amplification characteristics of K6 and K7 are similar to that of the Naruto and Kitan straits. But the amplification rate of periodic waves for 10 to 25 minutes is about twice as large as that of the Naruto and Kitan straits. Here, we have analyzed the spectra of the secondary undulation of tides at Komatsushima port in 1979 and found that the natural period of Komatsushima port is about 14 minutes. When the results are compared, it is clear see that resonance occurs at this period. Moreover, a larger resonance occurs at period of 50 minutes, which is the natural period of the Kii channel, and the amplification rate is larger. From this we found that even in each of the bays along the coast of the Kii channel, the characteristic that the amplification rate of periodic waves of 25 and 50 minutes are large does not change. In addition, the amplification rate is larger at the natural period of the Kii channel, 50 minutes, than the natural period of the bay. This means that the resonance of the whole Kii channel contributes more to the amplification of the tsunami height than the natural period of the bays. It was found that the amplification characteristics in the Kii channel and outside of the channel is different and that while outside of the channel the natural period of 25 minutes is predominant, inside of the channel the natural period of the Kii channel, 50 minutes, is predominant, even though amplification is also seen at 25 minutes, which is the natural period of the bays, and also the natural period of the Kii channel. In addition, here it should be emphasized that when a tsunami propagates into the Kii channel the amplification rate is larger than one, even at periods other than resonance periods. This means that if a tsunami propagates into the Kii channel it will inevitably amplify and the tsunami height will become higher as a result.

　

　

　

Figure 10 shows the distribution of the amplification rate of each periodic wave in the Bungo channel and in this figure, in order to see the affection of the shape of the bay, we picked out B4 as an example and showed the variance of the amplification rate. When we look at B3 located at the deep inner part of the channel, it can be seen that even though the amplification rates at 25 and 50 minutes are at the extreme value, it is only as large as the incident wave, and that other periodic waves have decreased. However, since in B4, which is located at the deep inner part of the channel, amplification rates at 50 and 22 minutes are at the extreme values, it appears that this is the natural period of the Bungo channel.

　

　

These results were examined and it was concluded that the Hoyo straits have a wider opening width compared to the Kitan and Naruto straits and thus the energy of a tsunami may propagate more easily to the Seto inland sea. There may be an affection of this fact. In addition, if we obtain the natural period of Uwajima bay using the Median wave period, the results are 55 and 19 minutes respectively. This period is almost the same as the natural period of the Bungo channel seen in the figure. However, related to this point, we feel that it is necessary to analyze it in more detail oscillation mode and we intend to carry with such research in the future.

　

　

The results from this research are summarized as follows. First, the Kii, Hidaka and Tomita submarine canyons affect tsunami propagation when a tsunami enters the Kii channel, and if the tsunami source is located to the north of the submarine canyons, the wave rays of tsunamis more easily propagate into the Kii channel. Secondly, in the Bungo channel, the wave rays of a tsunami more easily propagate into the Bungo channel if the center of the tsunami source is located to the west of Ashizuri cape or a tsunami occurs in the same tsunami source as the Hyuganada earthquake tsunami. Thirdly, tsunami heights in the Kii channel are easily amplified irrespective of the period of incident waves and the Kii channel has resonant periods of 25 and 50 minutes. Fourthly, tsunami heights along the coast of the Kii channel are strongly influenced by the natural period of 50 minutes of the Kii channel rather than the natural period of the bay. Fifthly, in the Bungo channel, tsunami heights are not easily amplified for all periodic waves, but resonate by the specific periodic waves for 22 and 50 minutes.

　

　

Murakami, H. and H. Takatani. 1977. "Harbor oscillation characteristic of Nasa bay", Proc. Annual Conference, JSCE, Vol.32, pp.63-64. (In Japanese)

　

Murakami, H. and T. Shimada. 1979. "Study on the characteristic of the secondary undulation of tides based on field data and survey", Proc. Coastal Eng., JSCE, Vol.26, pp.129-133. (In Japanese)

　

Murakami, H., Y. Ito, N. Yamamoto and H. Sato. 1995 "Study on the tsunami risk in the coast of Shikoku Island", Proc. Coastal Eng., JSCE, Vol.42, pp.361-365. (In Japanese)

　

Murakami, H., T. Shimada, Y. Ito, N. Yamamoto and J. Ishizuka. 1996a. "Review of the height of historical tsunami (1605 Keicho, 1707 Houei, 1854 Ansei) in Shikoku island", J. JSNDS, 15-1, pp.39-52. (In Japanese)

　

Murakami, H., Y. Kozuki, K. Kurata, T. Sugimoto and K. Yoshida. 2001. "Propagation characteristics of the 1854 Ansei Nankai earthquake tsunami in the Kii and Bungo channels", Historical Earthquake, Vol.17, pp.110-116. (In Japanese)

　

Shimada, T., Y. Mori and T. Sakai. 1984. "Development of the numerical analysis model of deformation of tsunamis in Osaka Bay" Proc. Annual Conference, JSCE, Vol.38, pp.351-352. (In Japanese)