Tsunami wave-front detection with oceanographic radar based on virtual tsunami observation experiments

Transcription

1 Session 8 Tsunami

2 Tsunami wave-front detection with oceanographic radar based on virtual tsunami observation experiments Kohei Ogata 1, Ryotaro Fuji 1,2 and Hirofumi Hinata 1 1: Research Institute for Low Temper Department of Civil and Environmental Engineering, Faculty of Engineering, Ehime University, Japan 2: Technology Management Department, Kokusai Kogyo Co., Ltd., Japan Corresponding author s hinata@cee.ehime-u.ac.jp 1 Introduction Since a tsunami generated by the 2011 Tohoku-Oki earthquake was firstly detected by using High-frequency oceanographic radar (HF radar) (e.g., Hinata et al., 2011; Lipa et al., 2011), HF radar system for tsunami detection has been developed. For example, Lipa et al. (2012) proposed a tsunami arrival detection algorithm based on temporal change of radial velocities, which were spatially averaged over bands parallel to the coast. Grilii et al. (2018) proposed a detection method using cross-correlation of the received signals at two points along a tsunami-lay calculated beforehand. The former has a concern for an underestimation of tsunami velocity due to spatially averaging and the latter needs to calculate the tsunami-lay beforehand. The tsunami wave-front detection techniques by HF radar are still open to improvement. Here, we propose a tsunami wave-front detection technique based on the virtual tsunami observation experiment proposed by Fuji et al. (2015). In the experiments, we synthesized actual receiving signals in February 2014 obtained by the NJRC (Nagano Japan Radio Co., Ltd) HF radar system installed on the Mihama coast in Wakayama prefecture and idealized receiving signals produced by a numerical simulation of Nankai Trough earthquake tsunami. The tsunami was detected based on temporal change in the cross-correlation of radial velocities between two observation points along a beam. Performance of the technique was statistically evaluated referring to the work by Fuji and Hinata (2017). 2 Materials and Methods We used actual receiving signals of the NJRC radar installed on the Mihama coast obtained during February 2014, which were the same as those used in

3 Fuji and Hinata (2017). The HF radar coverage with depth contours of the Kii Channel is shown in Figure 1(a). In this study, we simulated tsunami velocities induced by a Mw 9.0 Nankai Trough earthquake using a numerical model. Figure 1(b) shows the tsunami initial sea surface elevation calculated by using Okada s formula. To precisely calculate the tsunami in the shallow water region, this model used the nonlinear long-wave equation as a fundamental equation. We calculated the tsunami velocities for 48 h after the tsunami occurrence. The detailed procedure of a tsunami wave-front detection method is described below: (1) Two cross-correlations of radial velocities between the two different range cells (corobs(b, m, t), corvt(b, m, t)) were calculated every 1 min by using the combinations of the observed and synthesized radial velocities ((v obs(b, m, t), v obs(b, m+2, t)) and (v vt(b, m, t), v vt(b, m+2, t))) in [t-30, t], respectively, where b is the beam number (b =00, 01,, 11), m is the range cell number (m = 1,, 64), t is time in minute. corobs(b, m, t) was calculated for the duration from 00:30 on February 1, 2014 to 23:59 on February 28, 2014 and corvt(b, m, t) was calculated for the duration [t0-330, t0+360] in 660 tsunami events, where t0 is the time of tsunami occurrence; (2) Frequency distributions of cross-correlation (P[corobs(b, m, t)]) was approximated by a normal distribution; (3) When corvt(b, m, t) became larger than top 1 % value of P[corobs(b, m, t)] after tsunami occurrence, the significance function F(m) was set to 1, otherwise F(m) was set to 0; (4) Finally, tsunami arrival was judged at range cell m, when F(m), F(m+1) and F(m+2) are 1. (a) (b) Figure 1. (a) Observation area of the HF radar that had been installed on the Mihama coast with depth contours in and around the Kii channel. The red circle represents the radar station. (b) Initial sea surface elevation induced by the Japan Cabinet Office s fault model case 3. 3 Results and Discussion In the case where the earthquake had occurred at 6:00 on February 1, 2014, the tsunami would have been firstly detected at 4 min after the tsunami occurrence at 31.5 km offshore. On the other hand, if the earthquake had

4 occurred at 13:00 in February 27, 2014, when the significant wave height was greater than 5 m, the wave-front detection would have failed because of the poor signal-to-noise (S/N) ratio (Fuji and Hinata, 2017). These results demonstrate that the detection method should be statistically evaluated because of the dependency of the detection distance on the time-variant S/N ratio. We performed the statistical analysis on the detection probability using 660 tsunami scenarios: in the first-scenario, tsunami occurred at 6:00 on February 1; the second tsunami occurred at 7:00 on February 1; and the last tsunami occurred at 18:00 on February 28. HF radar observation had failed due to intermittent system troubles in 70/660 scenarios. Figure 2 shows temporal evolution of detection-probability at 1, 4, 7, 10, 13 and 16 min after the tsunami occurrence with depth contours in and around of the Kii Channel. The background color represents the detection-probability, and the pink line represents the first leading wave-front of the tsunami. The maximum detection probability was 20 % at 4 min after the tsunami occurrence, which increased to 80 % at 7 min after the tsunami occurrence (9 min before the tsunami arrival to the coast). The 80 % detection-probability line consistently located 3 km behind the wave-front propagating to the coast. (a) (b) (c) (d) (e) (f) Figure 2. Temporal variations of detection probability at (a) 1, (b) 4, (c) 7, (d) 10, (e) 13, (f) 16 min after the tsunami occurrence with depth contours in and around the Kii Channel. The color represents the detection probability, and the pink line represents the first leading wave-front of the tsunami. 4 Conclusion We developed a tsunami detection method and statistically assessed its

5 performance on the detection probability. The maximum probability was 15 % at 4 min after the tsunami occurrence, which increased to 80 % at 7 min after the tsunami occurrence. Since then, 80 % detection-probability line consistently located 3 km behind the wave-front as it propagated to the coast. To obtain comprehensive understanding of the tsunami detection-probability of the radar system, virtual tsunami experiments for the other seasons in 2014, when the sea surface state were different from that in February, and/or for other earthquakes are required. Reference Hinata, H.; Fujii, S.; Furukawa, K.; Kataoka, T.; Miyata, M.; Kobayashi, T.; Mizutani, M.; Kokai, T.; Kanatsu, N., 2011: Propagating tsunami wave and subsequent resonant response signals detected by HF radar in the Kii Channel, Japan. Estuar. Coastal. Shelf Sci, vol.95, pp , doi: /j.ecss Lipa B, Barrick D, Saitoh S, Ishikawa Y, Awaji T, Largier J, Garfield N., 2011: Japan tsunami current flows observed by HF radars on two continents. Remote Sens, pp , doi: /rs Fuji, R.; Hinata, H.; Fujii, S.; Nagamatsu, H.; Ogasawara, I.; Ito, H.; Kataoka, T.; Takahashi, T., 2015: Tsunami detection based on virtual tsunami observation experiment by using oceanographic radar. J Jpn Soc Civil Eng Ser B2 (Coastal Engineering), vol.71, pp (in Japanese with English abstract). Fuji, R. and Hinata, H., 2017: Temporal variability of tsunami arrival detection distance revealed by virtual tsunami observation experiments using numerical simulation and 1-month HF radar observation. J Oceanogr, doi: /s y Lipa, B., Isaacson, J., Nyden, B. and Barrick, D., 2012: Tsunami arrival detection with high frequency (HF) radar. Remote Sens, pp , doi: /rs Guérin, C.A.; Grilli, S.T., Moran, P. et al., 2018: Tsunami detection by high-frequency radar in British Columbia: performance assessment of the time-correlation algorithm for synthetic and real events. Ocean Dynamics, doi: /s

6 Verification of Ocean Radar Tsunami Observation of the 2011 Tohoku Tsunami by Numerical Simulation in Ise Bay and Mikawa Bay, Japan Yu Toguchi 1, Satoshi Fujii 2, and Hirofumi Hinata 3 1: Department of Interdisciplinary Intelligent Systems Engineering, Graduate School of Engineering and Science, University of the Ryukyus, Japan 2: Department of Electrical and Electronics Engineering, University of the Ryukyus, Japan 3: Department of Civil and Environmental Engineering, Ehime University, Japan Corresponding author s k178674@eve.u-ryukyu.ac.jp 1 Introduction After the 2011 Tohoku tsunami, several tsunami detection methods and observations of natural oscillation via ocean radar were investigated (e.g., Fuji and Hinata 2017; Benjamin et al., 2016; Lipa et al., 2011). However, most of the researches used radial velocity for a single radar. By using two or more radars, we can obtain the 2-D current information which is useful for disaster prevention in coastal areas. Thus, we verify HF and VHF radars tsunami observation by a numerical tsunami simulation. Additionally, we show the response characteristics for the tsunami arrival in Ise Bay and Mikawa Bay (Fig. 1). 2 Data and Method 2.1 HF and VHF radars In this study, we used CODAR system HF and VHF radar records. The radars are located in Ise Bay and Mikawa Bay. Those transmitting frequencies are approximately 24.5 MHz and 42 MHz respectively. The specifications are shown in Table 1. To analyze the tsunami, we used five-min interval data obtained by reducing the average number of cross spectra. Because of the decrease in the signal to noise ratio in the received signals, many missing data points and abnormal velocities occurred. To reduce these effects, for radial

7 velocity analysis, we projected the radial velocities onto a beam direction and averaged those velocities over 2 km (1 km) bands ranging 30 km (10 km) from the radar site, 6 km (3 km) along an axis perpendicular to the beam direction for HF (VHF) radar. For total velocity analysis, we used observation points that had 50% or greater data acquisition rates. Table1 System Specifications of the HF and VHF Radar System Radar HF radars (NABE and MATU) VHF radars (MITO and TAHA) Center frequency MHz 41.9 MHz Sweep bandwidth 100 khz 300 khz Range resolution 1.5 km 0.5 km Radar type Transmitting antenna Receiving antenna Time interval FMICW Monopole Monopole and crossed-loop 5 min 2.2 Tsunami simulation The numerical model we used in this study is JAGURS which is developed by Baba et al., (2017). The total calculation area is shown in Fig. 1. The calculation is performed by a finite difference scheme using a staggered grid and the leapfrog method. The calculation is non-linear except for the first level calculation area. To verify the tsunami simulation, we also used tidal gauge records. The observation locations are shown in Fig Results Figure 2 shows the sea surface heights and radial velocities along the beam direction (Fig. 1). The observed data were performed band-pass filter ( min). We selected the beam direction 190 degrees (NABE) and 112 degrees (MATU) from the north for HF radars in Ise Bay, 245 degrees (MITO) and 20 degrees (TAHA) for VHF radars in Mikawa Bay. The beam direction of MATU observes the bay mouth, NABE, MITO, and TAHA observe the bay head for each bay. There is good agreement between the modeled radial velocity and observed velocity for each beam direction. The velocities at MATU, MITO, and TAHA were almost same phase along the beam directions. This is because the beam directions correspond with the major current directions of the natural oscillation mode of 70 min period (Fig. 3) calculated by the method of Loomis (1975). Figure 4 shows that the tsunami current fields for simulated and observed velocity at 18:50 (JST) on 11 March. There is also good agreement between the modeled and observed current velocity. Both the current fields are

8 similar to the natural oscillation mode of 70 min period. This shows the development of the natural oscillation after the first wave. The radars were successfully observed this behavior. Figure 1 Calculation area of the (a) first, (b) second, (c) third, and (d) fourth level domains for the numerical tsunami simulation. 4 Conclusions HF and VHF radars located in Ise Bay and Mikawa Bay observed the tsunami currents and these behaviors successfully even though the velocities were obtained by such short-term spectra. This indicates that HF and VHF radars are expected to provide key information for tsunami monitoring which is useful for observation of continues natural oscillation and tsunami decay. Reference Fuji, R. and H. Hinata, 2017: Temporal variability of tsunami arrival detection distance revealed by virtual tsunami observation experiments using numerical simulation and 1-month HF radar observation. Journal of Oceanography, vol.73, no.6, pp , doi: /s y. Lipa, B., D. Barrick, S. Saitoh, Y. Ishikawa, T. Awaji, J. Largier, and N. Garfield, 2011: Japan tsunami current flows observed by HF radars on two continents., Remote Sensing, vol.3, pp , doi: /rs Loomis, H. G., 1975: Normal modes of oscillation of Honokohau harbor, Hawaii, Hawaii Inst. Geophys. Rep., HIG-75-20, pp Baba, T., S. Allegeyer, J. Hossen, Phil R. Cummins, H. Tsushima, K. Imai, K. Yamashita, and T. Kato, 2017: Accurate numerical simulation of the far-field tsunami caused by the 2011 Tohoku earthquake, including the effects of Boussinesq dispersion, seawater density stratification, elastic loading, and gravitational potential change. Ocean Modelling, vol.111, pp.46-54, doi: /j.ocemod

9 Figure 2 (a) Sea surface height for observation (black) and simulated (red) tide gauges. (b) Radial velocities along the beam direction of NABE, (c) MATU, (d) MITO, and (e) TAHA. The upper shows simulated velocity and the lower shows observation velocity. Figure 3 Eigenmode of 70 min period calculated by Loomis (1975). The color shows velocity potentials and the black bold lines show nodal lines. The vectors are calculated by difference in velocity potentials. Figure 4 (a) Snapshots of tsunami current of (a) numerical simulation and (b) radar observation at 18:50 (JST) on 11 March.

10 WERA Ocean Radar as a Tool for Monitoring of Seismic and Non-Seismic Tsunamis Anna Dzvonkovskaya, Thomas Helzel, Leif Petersen Helzel Messtechnik GmbH, Kaltenkirchen, Germany dzvonkovskaya@helzel.com 1 Introduction High-frequency ocean radar is known to deliver simultaneous wide area measurements of ocean surface current fields far beyond the horizon. The WERA ocean radar system is a shore-based phased-array system to monitor ocean surface in real-time and under all-weather conditions. Recently several WERA systems have been installed in Oman (2015), Chile (2016) and Canada (2016) especially with an option to monitor a tsunami situation as a support of Tsunami Early Warning Systems (TEWS). An ocean radar system does not directly measure the approaching wave height of a tsunami; however, it measures the surface current velocity generated when the tsunami enters the continental shelf. The observed tsunami signatures reflect the known dependency of tsunami currents on bottom topography. An unusual change of the surface current can be detected and tracked by a phased-array radar system in real-time as it has been initially demonstrated by radar measurements during the seismic 2011 Japan tsunami. Afterwards, the requirements for tsunami monitoring by ocean radar have been identified [1] and include certain conditions: The ocean bathymetric data within the radar coverage has to be known in details to plan an ocean radar installation having a maximum effectiveness for tsunami monitoring. The width of the shelf is sufficiently extended to allow time for issuing and transmitting a tsunami alert. The spatial resolution of radar mapping has to be high enough to resolve the tsunami current signatures and thus has high signal-to-noise performance. The temporal resolution of the radar system must be high enough to detect the rapidly changing surface velocity with periods of several minutes. The potential tsunami-affected areas should be monitored in a fast current update

11 mode (e.g. at 30-sec intervals). The radar system should be equipped with an additional uninterrupted power supply unit to account for the possibility that a power outage can happen. The transmission link between a radar site and the central server of TEWS should be stable and independent of local communication networks. For example, during and after strong earthquakes mobile communication networks may fail in the region. Beyond any doubt, an optimization process is necessary for each radar site individually due to different geometries of the continental shelf and radar operating frequency. Nevertheless, by measuring only surface current velocities, ocean radar systems are able to contribute to the development and improvement of TEWS. 2 Detection of Seismic Tsunamis One of the WERA ocean radar systems was in operation at 22 MHz in Rumena, Chile, when the 2011 Japan tsunami waves encountered the Chilean coast after propagating nearly 17,000 km within 22 hours across the Pacific Ocean. After the earthquake in Japan and before the arrival of the tsunami in Chile, the Rumena radar was reconfigured to record time series of signals from the receive antennas in successive 5-minute intervals. The unique opportunity to observe a natural tsunami event on March 12, 2011, using an HF radar showed that such radar systems may be used to measure tsunami surface current velocity; nevertheless, the observations indicated that the measurement update rate needed to be increased. After post-processing intense tsunami signatures of changing surface current velocities were observed by the radar system. Large deviations up to 50 cm/s in ocean current measurements were obtained after detrending the natural tidal component from measured velocities. The tsunami wave train can be clearly seen already tens of kilometers offshore in the radar measurements. The current velocity becomes stronger closer to the coast. Due to the narrow continental shelf (10-20 km) covered by radar, the first appearance of current deviation occurred about 7 min before the waves reached the coast. The current velocities were found to be significantly correlated with the water level measurements from the tide gauge located 50 km south off the radar site. The tsunami wave periodicity was estimated from the radar and tide gauge data. It has values of 14 and 32 min, the same for both instruments. It should be mentioned that the first wave was not the strongest wave as it s usually thought. This indicates a distinguished and important property of HF radar, i.e.

12 it offers the possibility of not only being able to detect the first tsunami wave but also to continuously monitor the full tsunami event and provide a hint of tsunami alarm ending. The estimated radial current residuals were compared with modeled zonal and meridional velocity components calculated specifically for the HF radar coverage in Chile using the NOAA Tsunami Forecast model. The model estimated the first wave arrival about 20 minutes earlier than what actually happened and indicated one wave period less in the tsunami train. The modeled water elevation is only one half of that measured by the tide gauge; therefore, the predicted tsunami current velocity is correspondingly smaller. 3 Real-Time Monitoring of Non-Seismic Tsunamis Since the 2011 Japan tsunami, the WERA system was upgraded to follow the TEWS requirements and was installed as a part of the Ocean Networks Canada (ONC) Tsunami Project, the initiative to develop a near-field tsunami alert network consisting of different types of pressure and seismic sensors as well as ocean radar systems. On 14 October 2016, the 13-MHz WERA system on Vancouver Island, Canada, automatically detected strong changes in measured currents at distances up to 60 km off the coast and triggered an alert immediately. The system tracked the unusual current pattern for 1.5 hour in real-time following the wave propagation coincided with an atmospheric frontal passage. The analysis of the available data records from nearby tide gauges and a meteorological buoy showed that the event may be identified as a meteotsunami (a type of non-seismic tsunami) caused by a powerful extratropical cyclone. A jump in surface current velocity is observed simultaneously with the pressure development (see detailed event description in [2]). On 29 May 2017, another meteotsunami was generated in the North Sea by an air pressure disruption and reached the southwestern coast of the Netherlands (see Fig. 1). There are two 16-MHz WERA radar systems installed in that area, which are operated in a standard non-tsunami mode and provide sea current information around the port of Rotterdam. After re-processing the acquired raw radar measurements from both systems, the original WERA tsunami detection software identified tsunami-like currents more than 40 km offshore and estimated an event duration of 2 hours (see detailed event description and comparison with tide gauges in [3]). For comparison one can observe in Fig. 2 that the wave front has a similar slope to the coastline as it is seen in Fig. 1.

13 Figure 1. A video evidence of the Dutch meteotsunami wave propagation near Zandvoort. (Retrieved from [4]) Figure 2. Meteotsunami signatures (red colour) in WERA tsunami alert maps generated on 29 May 2017 at 03:35 UTC for the radar site at Monster, the Netherlands. 4 Conclusions The remote observations of real tsunami events have shown a good applicability of phased-array radar technology for offshore seismic and non-seismic tsunami monitoring as a valuable tool to support TEWS. The systems can be easily upgraded by modifying the radar hardware and installing a tsunami software package. The presented results using the operational WERA radar systems showed that offshore current measurements may be beneficial to draw the attention of tsunami warning decision makers and authorities and become a useful complement to the already existing water level measurement network. References [1] A.Dzvonkovskaya, 2018: HF Radar for Tsunami Alerting: from System Concept and Simulations to Integration into Early Warning Systems, IEEE A&E Systems Magazine, March 2018, pp. 2-12, doi: /MAES [2] A. Dzvonkovskaya, L. Petersen and T. L. Insua, 2017: Real-Time Capability of Meteotsunami Detection by WERA Ocean Radar System, Proc. of Int. Radar Symposium IRS-2017, Prague, Czech Republic, June 2017, pp. 1-10, doi: /IRS [3] A. Dzvonkovskaya, T. Helzel, H. Peters, 2018: Meteotsunami Observation by WERA Ocean Radar Systems at the Dutch Coast, Proc. of OCEANS'18 MTS/IEEE, Kobe, Japan, May 2018, p [4] The Dutch Meteotsunami

14 Tsunami Wave Front Detection Algorithm using Ocean Surface Radar (Tetsutaro Yamada, Hiroshi Kameda, Yasushi Obata, Tomoyuki Koyanagi, and Toshihiko Arioka) Mitsubishi Electric Corporation, Japan 1 Introduction Tsunami warning systems are required to detect an approaching tsunami as early as possible to maximize the evacuation time. Ocean surface radar can observe the current velocity at every range and azimuth resolution cell in the radar coverage, which is directionally wide and long beyond the horizon. Research on the detection of tsunamis using ocean surface radar has been conducted [1] [2] [3]. However, current velocity measurement error of the ocean surface radar is large [4], and there is a problem that there are many false alarms when detecting a tsunami based on the current velocity measurement every radar resolution cell. As a tsunami detection method, there is a method of reducing the false alarm and detecting a tsunami by averaging the current velocity measurement along the contour line of the seafloor topography. However, the propagation of a tsunami is affected by the seafloor topography from the source seismic epicenter to the radar measurement point. The influence of the seafloor topography in the radar coverage area is limited and the contour line of the seafloor topography and the shape of the tsunami wave front do not always match. 2 Proposed method This paper provides a method that detects the arrival of a tsunami far out at sea using the wave front hypothesis. We propose multiple tsunami wave front candidates within the coverage area. A conceptual diagram of wave front candidate generation is shown in Fig. 1. The wave front hypothesis is a combination of two cells in radar coverage. The wave front is a straight line between the two cells. The cells of the wave front hypothesis are composed of the closest cell from the wave front of every radar beam. The wave front detection algorithm calculates a weighted summation of the current velocity of 1

15 the wave front hypothesis cells to reduce the effect of noise. The weighted summation is described as n vicos i v 2 (1) i 1 i where i is the cell number, vi is the current velocity, θi is the angle between the wave front direction of the hypothesis and radical direction of the cell, and σi is the standard deviation of the current velocity. Lastly, the maximum weighted summation is selected and compared with a detection threshold. Fig.1 Concept of proposed method 3 Main result Numerical simulation evaluation of the proposed method is performed on tsunami simulation data by Sumatra-Andaman earthquake [5] to verify the effectiveness of the proposed method. The initial tsunami water level due to the 2004 Indian earthquake is shown in Fig. 2 In the case of the 2004 Indian Ocean earthquake, tsunami invaded from the northern part of the Straits of Malacca. It is simulated using two-dimensional non-linear shallow-water equation. Assuming that the radar is installed on Pulau Pinang facing the Strait of Malacca in Fig. 2 and the radar can observe the current velocity of a tsunami of each radar cell at 1 minute intervals, the measurement error σi is made uniform at 7 cm / s in all cells. The evaluation time is evaluated on the assumption that the time from the occurrence of a tsunami to the radar coverage area (138 minutes) is 60 minutes. In addition, as a conventional method, a method of comparing the flow rate for each cell with a threshold and detecting a tsunami when there is a current velocity exceeding the threshold value is compared. 2

16 The number of Monte Carlo simulations for calculating false alarm probability is set to times. Figs. 3, 4, and 5 show measurement current velocity and tsunami detection results. The graph on the left shows the measurement current velocity, the color bar shows the magnitude of the current velocity measurement, the middle graph shows the tsunami detection result of the conventional method, and the right graph shows the tsunami detection result of the proposed method. The detection result is indicated by red dots. Fig. 3 shows the tsunami detection result after 183 minutes from the occurrence of a tsunami, and Fig. 4 shows the tsunami detection result after 191 minutes from the occurrence of the tsunami. It can be seen that both the conventional method and the proposed method can detect the tsunami. When the detection probability of the proposed method and the conventional method is evaluated at 60 minutes, the detection probability is 97%. On the other hand, Fig. 5 shows the detection results in the absence of a tsunami. Although the false alarm has occurred in the conventional system of Fig. 5, it can be seen that the false alarm can be suppressed in the proposed system. As a result of performing the Monte Carlo simulation times, the false alarm probability of the conventional method is 0.16 and the false alarm probability of the proposed method is The false alarm suppression effect of the proposed method is verified. Fig. 2 Tsunami initial water level and water depth and radar settings Fig. 3 Simulated tsunami radial current velocity in radar coverage 3

17 Fig. 4 Simulated tsunami radial current velocity in radar coverage Fig. 5 Simulated radial current velocity in radar coverage (no tsunami) Reference [1] K.W. Gurgel, A. Dzvonkovskaya, T. Pohlmann, T. Schlick, and E. Gill, "Simulation and detection of tsunami signatures in ocean surface currents measured by HF radar ", Ocean Dynamics, [2] B. Lipa, D. Barrick, J. Bourg and B. nyden, "HF Radar Detection of Tsunamis", Journal of Oceanography, Vol. 62, pp. 705 to 716, [3] H. Hinata, S. Fujii, K. Furukawa, T. Kataoka, M. Miyata, T. Kobayashi, M. Mizutani, T. Kokai, and N. Kanatsu, "Propagating tsunami wave and subsequent resonant response signals detected by HF radar in the Kii Channel, Japan, Estuar. Coast. Shelf Sci., vol.95 (1). pp , Nov [4] B. Lipa, D. Barrick, J. Bourg and B. nyden, "HF Radar Detection of Tsunamis", Journal of Oceanography, Vol. 62, pp. 705 to 716, [5] Y. Tanioka, Yudhicara, T. Kususose, S. Kathiroli, Y. Nishimura, S. Iwasaki, K. Satake, Rupture process of the 2004 great Sumatra-Andaman earthquake estimated from tsunami waveforms, Earth Planets Space , /BF