T-waves [Linehan, 1940] are phases containing at least some acoustic paths through water bodies that are usually recorded by hydrophones, and sometimes by seismometers (Fig. 1).
T-waves are important for studying abyssal earthquakes and also as an important monitoring tool for the Comprehensive Nuclear-Test-Ban Treaty. For the acoustic leg of the path, the energy in the ocean is mostly trapped at about a depth of 1000 m in the SOFAR (Sound Fixing and Ranging) channel, a low acoustic-velocity zone in the water column formed due to the decreasing temperature and increasing pressure at depth. Okal [2008] pointed out that rays can be trapped in the SOFAR-channel, which acts as a 2D wave-guide, instead of a typical 3D propagation in solid earth paths. In the impedance
Figure 1: We have defined three different kinds of T-waves for this study. For the Type 1 T-wave, the first leg is the elastic wave from the earthquake to the conversion points at the 1000 m water depth. The second leg is traveling in the SOFAR-channel, a low velocity zone in the water column at around 1000 m water depth. See a cartoon of the water velocity profile to the right of the figure. The third leg is from the SOFAR-channel sub-vertically to the OBS. For the Type 2 T-wave, the first conversion point is outside of the 1000 m depth, or the second conversion point is far away from the OBS. For the Type 3 T-wave, the elastic wave propagates from the earthquake to the seafloor near the OBS. Then there is a shortest acoustic path to the OBS.
SOFAR-channel
2
data near to our S004 site provided by NCOR [National Center of Ocean Research] of Taiwan showed the slowest velocity in SOFAR-channel at 1000 m from water depth, and which heterogeneity might help to leak the T wave energy out of SOFAR-channel.
There is much less attenuation when energy travels in the SOFAR-channel, and T-waves of very smaller events from great distances can still be detected. The slow propagation velocity and the small attenuations in the SOFAR-channel makes it possible to better locate these events from afar. Fox and Dziak [1998] have used T-waves to successfully locate volcanic activity on Gorda Ridge, and helped a response cruise to the correct magmatic activity site. The detecting threshold for ridge events using T-waves is usually one magnitude below the seismic network [J A Hanson and Bowman, 2006].
Because earthquakes occur in solid earth materials, this implies that T-waves need to have at least one conversion between an elastic wave in solid earth and an acoustic wave in fluid, i.e. P or S to acoustic [Okal, 2008]. At the same time, acoustic to P [D H Shurbet, 1954] or S wave [Okal and Talandier, 1997; D H Shurbet, and M. Ewing, 1957]
conversions have also been documented. Those studies have helped us better understand the T-waves.
However, there are debates on which part of the T-wave should be picked for relocation. Previously it was considered that beginning of the T-phase should be picked to satisfy the principle of causality, even though the emergent features of some T-waves made it difficult. On the other hand, using well-located events through typical seismological methods, it has been found that the traveltimes estimated using T maximum phases gave closer fits [Fox and Dziak, 1998; J Hanson et al., 2001; Slack et al., 1999].
In addition, there is no systematic study of how the radiation patterns of the earthquake affect the T-wave amplitudes. Without a better understanding of these issues, we can not
3
study the duration and the amplitudes of the T-waves in greater details. We also need to know the paths of the different parts of the T-waves before we can better relocate the events.
Accurate locations of earthquakes in remote abyssal regions are usually difficult to derive using other methods, thus it is usually difficult to calibrate the accuracy of the T-wave derived earthquake locations. The situation is also complicated by the trade-offs between origin time, depth, and location [Butler, 2006].
It will be helpful to study an earthquake which is well located by other methods, then pick different parts of the T-waves to see which phase in T-waves should be picked for relocation purpose. In addition, if there are many earthquakes with well-constrained focal mechanisms, we can also study the effects of earthquake radiation patterns on the T-wave attributes.
Whether P- or S-waves are more effective to be converted into T-waves is still under debate. de Groot-Hedlin and Orcutt [2001] suggest that, at low angles of incidence, the shear waves are mainly reflected back into the crust. Compressional waves are mainly reflected back into the crust at high incidence angles, with a smaller portion refracted into the sediments. Intuitively, the effective of converted T-waves are depended on seismic P- and S-waves normal vibration vector on seafloor. Balanche et al. [2009]
suggest such a result using wave propagation simulations [Park et al., 2001]. They found that S-waves are more efficient than P-waves in producing energetic T-waves.
Here we systematically studied 60 Mw > 3.3 earthquakes in Taiwan region, all of which have excited T-waves. We used an ocean bottom seismometer (OBS) and a broadband seismic station on an island (Fig. 2) to study these earthquakes. For this study we use many earthquakes with well-constrained focal mechanisms to study the effects of
4
earthquake radiation patterns on the T-waves attributes. We study the effects of spatial gaps in conversion points on reducing the amplitude of the T-waves. We also propose a method to determine if the arriving T-waves energy is coming from water column or from the crust.
Figure 2: (a) The location map of Taiwan. Coast lines are shown with back color and 1000 m bathymetric contour lines with blue color. The red triangles are the locations of the OBS station (S004) and the island seismic station (LYUB); (b) Regional topography and bathymetry represented in color scheme and we also plotted the 60 earthquakes used in this study.
5