Seismic monitoring of western Pacific typhoons
- First Online:
- Cite this article as:
- Chi, WC., Chen, WJ., Kuo, BY. et al. Mar Geophys Res (2010) 31: 239. doi:10.1007/s11001-010-9105-x
- 911 Downloads
Typhoons inflict large damage to societies, but are usually difficult to monitor in close proximity in real-time without expensive instruments. Here we study the possibility of using seismic waveforms on the seafloor and on land to monitor the turning of a far away or approaching typhoon. Up to 67% of the typhoons making landfall in Taiwan come from the eastern shore, so that we deployed broadband ocean-bottom seismometers (OBSs) offshore eastern Taiwan in 2006 to study ground motion in close proximity to a typhoon. Typhoons generate ocean waves, which generate pressure signals in the water column before being transmitted to the seafloor as seismic waves and recorded by the OBSs. The ground motions on the seafloor correlate with locally increased (ocean) wave heights and wave periods, suggesting that the ground motions are mostly induced by in situ or nearby pressure fields, as shown by coherence function analyses. When a typhoon turns and changes wave-wave interaction near the source region, a new set of en echelon patterns develops which can be observed by OBSs and land stations. Similar features occur when a typhoon crosses a landmass and re-enters the ocean. The energy level ratio between the single-frequency and double-frequency microseisms also changes abruptly when the typhoon turns. These features can potentially help near real-time early warning with little cost to complement other conventional typhoon early warning methods.
KeywordsTyphoon Early warning Seismic method Microseisms Hurricane Cyclone
Microseisms are seismic energy observed between earthquakes as continuous background “noises” of the earth (Kedar and Webb 2005; Bromirski et al. 2005; Kedar et al. 2008; Tanimoto 2008). Microseisms can be ground motions caused by storm-induced ocean waves, which have been studied from distances away from the storm (Tabulevich 1971; Sutton 1996; Suda 1998; Wilson et al. 2003; Kedar and Webb 2005; Webb 2007; Gerstoft et al. 2008; Koper and de Foy 2008; Wilson and Makris 2008).
Theoretically, for typical 0.0625 Hz ocean swells, ocean wave induced pressure in a water column decays exponentially with depth, thus it is easier to excite single frequency seismic waves in shallow water. Nonlinear wave-wave interaction regions occur where typical 0.0625 Hz ocean swells traveling from opposite directions can generate 0.125 Hz standing waves (double-frequency waves), which then couple with seafloor and propagate as seismic waves (Longuet-Higgins 1950; Kedar and Webb 2005; Webb 2007; Gerstoft et al. 2008).
There is debate as to whether microseisms are excited mainly near coastlines in shallow water regions or if it is possible to generate microseisms in deep-sea regions. Some studies propose that ground motions are induced in shallow water regions like continental shelves (Friedrich et al. 1998; Bromirski et al. 1999; Bromirski and Duennebier 2002; Bromirski 2001; Essen et al. 2003; Rhie and Romanowicz 2006; Kedar and Webb 2005; Bromirski and Gerstoft 2009). Other scientists propose that the source region is under or trailing a typhoon (Cessaro 1994; Stehly et al. 2006; Chevrot et al. 2007; Kedar et al. 2008; Gerstoft et al. 2008; Koper and de Foy 2008; Landes et al. 2010; Chi et al. 2010).
Some of the controversy might be due to different types of microseisms used in the studies (e.g. Gerstoft et al. 2006). There is strong evidence showing that some P-wave microseisms can be generated in the deep sea and reach Earth’s core before being recorded on the other side of the earth (Gerstoft et al. 2008; Koper and de Foy 2008). In addition, Bromirski et al. (2005) proposed that there are two kinds of double-frequency waves: short-period double-frequency (SPDF: 0.20–0.45 Hz) microseisms which are excited locally by ocean gravity waves, and long-period double-frequency (LPDF: 0.085–0.20 Hz) microseisms that are excited near coastlines at distant locations. This interpretation is based on much higher SPDF levels compared to LPDF at H2O site in the middle of the open Pacific Ocean. Zhang et al. (2010) also argued that double frequency P wave microseisms can be excited in both deep sea and coastal regions.
So far, all these analyses are based on waveforms recorded a few thousand kilometers away from the cyclones. Waveforms observed directly under a typhoon were not available previously until now. In 2006, we deployed broadband ocean-bottom seismometers (OBSs) around offshore Taiwan for the first time. One of the OBSs was located in a region where many typhoons have turned. The seismographs recorded not only earthquake waveforms but also noises generated by winds, waves, tides and other external forces.
Here we document this unique dataset recorded by the OBSs and other on-land stations in western Pacific regions to study typhoon-induced ground motions in order to improve our understanding of how to seismically monitor a typhoon. In September 2006, Typhoon Shanshan passed over the OBSs, providing an opportunity to study how the microseisms’ wave field evolves with a moving source. We also studied Typhoon Cimaron, which changed course several times during its short life span, to see if a change of course appears in the seismic record.
Saffir-Simpson scale of hurricane intensity
Storm surge (m)
Typhoon Cimaron, a category 5 cyclone (Table 1), changed course several times and caused much damage to China and Hong Kong (Lang and Pierce 2006). However, it also generated many similar but more distinctive signals. In addition, it went across the Philippines, providing a rare opportunity to study the evolution of ground motions when the typhoon made landfall and reduced the wave-wave interaction region, and likewise when the typhoon re-entered the ocean and a new set of wave-wave interaction was generated. The seismic waveforms recorded during this period help us identify the unique seismic signals generated during the development of a new set of wave-wave interaction.
Taiwan wideband ocean bottom seismology (OBS) program and experiment
Technical specifications of the ocean bottom seismographs used for this study
Guralp CMG-3TC broadband three-component seismometer. Flat velocity response from 0.0083 to 50 Hz. Gimbals will level seismometer if tilt <50°
Differential Pressure Gauge V6.3 with low-frequency corner at 0.001 Hz
Quanterra QA330 low-power 24-bit digitizer. Dynamic range: 135 dB
Sample rates: 1–200 Hz
Seascan Timebase. Drift rate is <0.5 ms/day before correction and <0.1 s/year after correction
WHOI-built modem; capable of baud rates of 1 kbit/s or better
Alkaline battery pack will deliver sufficient power for 6 month operation
Lithium battery pack will allow >12 month operation
EGG 8242 commercial acoustic release. Backup burnwire acoustic release housed in dedicated 12″ glass ball
622/544 kg with/without anchor; 5 1/2′ × 4′ and 4′ high
Field operations were challenging but rewarding. Our communication with one OBS was lost owing to a noisy acoustic environment in the region and another OBS remained mired on the seafloor where subsequent volcanic eruptions were reported for the first time (Lin et al. 2007). Two OBSs (S002 and S004, Fig. 1) were successfully recovered from seafloor depths of 1749 m and 4726 m, respectively. They recorded unprecedented ground motions on the seafloor when typhoons passed near them. We also use data recorded by land seismometers (F-net and GSN) arrayed along the western Pacific Ocean. The OBSs and the land seismometers recorded coherent energy synchronized with typhoons (Chi et al. 2010).
Taitung deep sea buoy data
Up to 67% of the typhoons that make landfall in Taiwan do so in eastern Taiwan. To collect critical data before the typhoon makes landfall, the Central Weather Bureau (CWB) of Taiwan deployed a deep sea buoy off the eastern shore of Taiwan in August 2006. The data sets collected include wind speed, wind direction, temperature, pressure, surface water temperature, wave height, wave period, wave direction, and direction spectrum. It collects data 12 times a day, but more frequently when a typhoon is near Taiwan (Underwater Photography Association/ETToday 2006). Such buoys provided rare but important time series at a fixed location during the month of September, 2006. The data can be extended from a point location to a much wider region using Wave Watch III models.
Wave Watch III model data
Wave Watch III (WWIII) is a hindcast system (Tolman 2005) and one of the Marine Modeling and Analysis Branch (MMAB) Operational Wave Models that consists of global and regional nested grids. A hindcast system is a mathematical model that closely estimates past events and enters them into the model to see how well the output matches the observed results. Wave Watch III Model Data are extensively validated and cover a wide region (NOAA/NCEP et al. 2010).
Observations and data analyses
The Shanshan typhoon was recorded at the seismic stations along its course including the OBSs deployed offshore Taiwan. The OBSs were closer to the typhoon, so the signal to noise ratio was much higher in the OBS dataset than in the land stations or in any datasets previously studied. In other words, we used near-source signals in a similar fashion to strong-motion signals for earthquake source studies. The duration of an earthquake is in seconds, whereas the life span of a typhoon is in weeks; therefore, our week-long waveforms recorded signals both near and far from the typhoon.
In addition to the inland and coastal sites, we also studied ground motions on the seafloor. The OBS waveforms have higher amplitudes than those of the land waveforms. Compared with station S004, the S002 station at shallower water depth recorded stronger ground motions (bigger amplitudes) in vertical displacement waveforms (Fig. 4).
Bromirski et al. (2005) propose that SPDF is generated on the seafloor locally by wind seas and the LPDF is excited by ocean waves impacting the coastline far away. Because their dataset was derived from distant storms, they needed to apply a differential scheme to enhance the signals. In this study, the near source wavefields show clear signals without the need for additional data processing and give two similar bands of energy in the double-frequency microseisms.
Typhoon Cimaron provided many clear ground motion patterns (Fig. 6). The far-field land station TATO recorded energy with increasing frequency until Cimaron changed its direction. The envelopes of the single-frequency band have en echelon patterns with increasing amplitude as the typhoon approached the station, wherein signals increased from the lower frequency region (0.04 Hz) to the higher frequency region (0.08 Hz) and jumped back to the lower frequency band. This jumping pattern occurred several times.
We compared the moments of “jumping” with the track of Typhoon Cimaron and found that the “jumping” occurred as Cimaron was passing over the island of Luzon (Fig. 6) and when it was turning. When the typhoon made landfall and passed over Luzon, the microseismic signals started to disappear. Once the typhoon re-entered the ocean, a new set of microseisms started to develop. On Julian day 305, Typhoon Cimaron stalled, and the dominant frequency shifted from 0.07 to 0.065 Hz. Starting from Julian day 306, a new set of microseisms started to develop when the typhoon turned and moved steadily to the southwest. During this period, there were several other storm systems thousands of kilometers away from this region; however, it was difficult to see their effects without more specialized data processing. For example, there are no observed microseism motions at station TATO when Typhoon Cimaron was over the island of Luzon, implying the signals from other distant cyclones were very weak compared with the signals from Typhoon Cimaron, and advanced data processing is needed to study their interaction with the strong signals induced by Cimaron. Based on global satellite imagery over this time span (see supplemental materials), we did not see a new set of microseisms form at station TATO when new storms occurred from these far distances. In other words, we did not see clear evidence of the influence of distance storms, possibly owing to the large signals induced by Typhoon Cimaro nearby. In addition, the jumping features were also observed during Typhoon Shanshan as well as in other typhoons.
Overall, the 0.085–0.2 Hz (SPDF) and 0.2–0.5 Hz (LPDF) bands were the dominant energy in the waveforms, and the frequency-shifting patterns were also observed in these two bands (Fig. 5). They were more energetic so appear to have a longer duration than that of the 0.05–0.085 Hz band.
Discussion and conclusions
The OBSs recorded stronger signals than the land stations. The OBSs sitting on soft sediment can record stronger signals than land stations (Fig. 2). The peak amplitude at S004 occurred a few hours after the typhoon center passed over the OBS. Compared with station S004 at 5 km water depth, station S002 at 1 km water depth recorded stronger ground motions, possibly related to the evanescent nature of the acoustic waves in the water column. However, these amplitude variations might also relate to local site effects.
Compared with station S004, station S002 recorded stronger ground motions (bigger amplitudes) and more long-period noises. However, Typhoon Shanshan passed both of the OBSs in a day. The intensity of Typhoon Shanshan did not differ much during the time period it passed over the OBSs at S002 and S004, respectively. This shows that the strength of a typhoon is not the main cause of the differences of amplitudes in this case. Possible causes include the depth difference and local site effects.
Aster et al. (2008, 2010) and Bromirski and Kossin (2008) have demonstrated that it is possible to use available seismograms to study historic trends of extreme tropical cyclones and their implication in terms of global warming. Lessons learned from this study can also be applied to seismic waveforms recorded before the satellite era to study decadal-scale climate changes.
In sum, we have studied the possibility of seismically monitoring the turning of a typhoon. We observed en echelon patterns, which may signal starting of new sets of swells and also new wave-wave interaction patterns when the typhoon is changing direction or speed. The energy levels of the single-frequency and LPDF microseisms also change when the typhoon is turning, providing additional criteria to detect turning of the typhoon.
Satellite images provide important information on the location of the typhoon when the satellite passes over the typhoon. Seismology provides an independent and complementary method that can be used to study the typhoon in real time. In the future we may use land and cabled ocean-bottom broadband seismometers to monitor such changes in real time. Because all the seismic waveforms are readily available, such a practice will be relatively inexpensive and can provide additional information to the more traditional typhoon alarm systems.
We appreciate the great help from the Chief Editor Amy Draut and three reviewers, so we can better interpret this unique dataset and improve the manuscript. The editorial help from Dr. Amy Draut is exceptional and we are very grateful. We thank Dr. John Collins for his generosity so our OBS program can become a reality. We thank the captain and crew members of the R/V Ocean Researcher I, who helped deploy and recover the OBSs during a difficult operational environment. The Central Weather Bureau of Taiwan is thanked for providing the deep sea buoy data. The Japan Meteorological Agency is thanked for the typhoon and oceanic wave datasets. We thank the National Institute of Informatics (NII) of Japan for the typhoon buoy data. We thank F-net for providing excellent seismic waveforms for this study. Prof. Junkee Rhie is thanked for providing the program for us to analyze Wave Watch III datasets. Dr. Eleanor Willoughby is thanked for showing us how to calculate coherence functions. This project is partially funded by NSC (grant number NSC99-2116-M-001-020) to WCC and by the Central Geological Survey to WCC. The TEC contribution number for this article is 0073. IES contribution number is IESAS1504.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.