5.0. Discussion
Since earthquakes record only the portion of energy released in seismic events (seismic slip) as opposed to that of stable sliding (aseismic slip), the perspective of slip from seismicity limits investigation to seismic energy release, and the related down-dip convergence, which is only a portion of the total convergence associated with the subduction process. The GPS data, however, reflect total lateral convergence across the subduction interface for the duration between campaigns for specific points or benchmarks. Thus, where the seismicity data represent a long term average seismic slip, the geodetic data represent a short term snapshot of total convergence. Thus, the two data sources reflect differing aspects of the localized tectonic convergence, both in time and space. Figure 15 summarizes the results from this investigation.
The initial suggestion in Chatelain et al. (1986) and further work (Prevot and Chatelain, 1994; and Calmant et al., 1997) strongly suggest the existence of a northward dipping transform zone beneath the southern end of Malakula (near -16.75° latitude at the trench axis). Seismicity and geodetic data used here appear to support this view. This area corresponds with odd and transform seismicity extending eastward from the trench axis into the back arc region (bottom panel of figure 15). It also corresponds with a marked change from reverse seismicity concentrated along the trench axis to the south, to less frequent and more dispersed reverse seismicity extending from the fore arc region into the back arc region to the north (middle panel of figure 15). Here the island morphology changes from a single island chain falling roughly 80 to 120 kilometers from the
trench axis to the south, to a triple row of trench parallel features to the north (see figure 1). Espiritu Santo and Malakula sit at or near the trench axis, Maewo and Pentecost sit roughly 120 kilometers to the east of the trench axis, and Ambae, Ambrym and Gaua form an active volcanic chain in the intervening basin. Island sites in the southern section (on Tanna and Efate) exhibit more rapid geodetically determined convergence rates - at or slightly above the far-field convergence rate, whereas sites in the central section (on Espiritu Santo, Malakula, Maewo and Pentecost) exhibit convergence rates that are only a portion of the far-field rate (see top panel of figure 15). The central section also exhibits a relative absence of other seismicity (odd, transform, or normal). The left most gray bar in figure 15 represents the region of cross-arc faulting that separates the southern and central sections, and highlights the correlation of features across the figure.
Taylor et al., (1995), suggests evidence for additional cross-arc faulting north of Espiritu Santo in the Banks Islands vicinity (near -14.3° latitude at the trench axis). Seismicity data appear to support this view. The region corresponds to some odd and transform seismicity extending into the back-arc, a change back to more frequent reverse seismicity concentrated along the trench axis to the north, and a change in island morphology from the triple row of features in the central section. The intervening volcanic island chain continues as the Banks Islands approximately 100 kilometers east of the trench axis but then appears to terminate. The Torres Islands further north occur in closer proximity to the trench axis, approximately 50 kilometers to the east. Unfortunately, there is currently no geodetic data for islands within the northern section. Their convergence rate is likely to return to the higher rates seen in the southern section based on its similar seismic and topographic character, although this is unconfirmed. The right most gray bar in figure 15 denotes this northern region of cross-arc faulting and highlights the correlation of features in the figure.
Comparison of the above delimited sections with the averaged seismic slip rates determined here reveals a complex and unclear relationship. The averaged seismic slip varies from slightly less than 1 centimeter, to 3 centimeters of seismic slip per year for the nearly one hundred year duration of the seismic data sources used (see top panel of figure 15). Over local regions, however, the seismic slip can be as high as 6 cm/year. This indicates that, on average, approximately 10 to 30 percent of the far-field convergence is accommodated as seismic slip - with the majority of the arc being in the 10 to 20 percent range. In more detail, the southernmost part of the arc shows an average of 2 centimeters of seismic slip per year decreasing to less than 1 centimeter at its northern extent. The central section reverses this trend and shows a gradual increase from 1 centimeter of seismic slip per year to 2 centimeters at its northern limit. The northern section continues this trend gradually increasing from 2 centimeters of seismic slip per year to 3 centimeters.
The central section of the New Hebrides Trench also corresponds to the physical extent of the d'Entrecasteaux Zone, an area of high relief having a mean depth of approximately 3500 meters against a surrounding abyssal plain of 4000 meters depth or greater. This region extends west from the trench axis between approximately -15.0° and -16.5° latitude (see "DEZ" in figure 1) and arcs southward eventually linking up with New Caledonia. Within this zone, several unique features exist. Along the northern edge of the d'Entrecasteaux Zone a linear east-west trending feature, the d'Entrecasteaux Ridge, reaches to depths near 2000 meters (Collot, 1985). Towards the southeast, the Bougainville Guyot is a conical feature which rises to within 1000 meters of sea level (Fisher et al., 1991). The eastern margins of these two features currently impinge upon the accretionary wedge west of Espiritu Santo and Malakula, respectively. Some reverse seismicity highlights the underthrusting interaction of the Bougainville Guyot (see "BG" near 167.0° longitude and -16.25° latitude in figure 3). Further to the west, the Sabine Bank is another large feature rising to just a few meters below sea level (see "SB" near 166.5° longitude and -16.0° latitude in figure 3) (Fisher et al., 1991). This feature is even more clearly defined with reverse seismicity. In contrast, the descending plate opposite the southern section of the trench is a relatively smooth abyssal plane in the 4000 to 5000 meter depth range.
Combining these results, the picture becomes clearer. In the central section, despite the existence of currently impinging several thousand meter tall subducting bathymetric features, such as the d'Entrecasteaux Zone containing the d'Entrecasteaux Ridge and the Bougainville Guyot, the analysis here shows little difference in the degree of seismic coupling when compared to the southern section which has essentially no subducting bathymetric features. However, aberrant short term geodetic convergence rates (Calmant et al., 1995, 1997; Taylor et al., 1995) and evidence for episodic coral uplifts (Taylor et al., 1980, 1987) suggest deformation within the overriding plate. The relative absence of seismicity in the central section further suggests any deformation is primarily aseismic with the exception of some convergence accommodation in shallow reverse slip of opposing orientation in the back-arc region along and east of Maewo and Pentecost (Taylor et al., 1995). Utilizing bathymetric features and seismicity to outline significant features, figure 16 shows a tectonic interpretation based on the geodetic and seismic slip results. In this figure, the total lateral convergence for the back-arc region has been estimated using the calculated seismic slip and the 10 to 30% seismic slip relationship. When compared with the far-field convergence rates, the central section exhibits a slip deficit of 3 to 4.8 cm. This is believed to be taken up as deformation within the block.
Others might counter that the seismic recurrence interval along the New Hebrides Trench is greater than one hundred years and major events, therefore, are yet to occur. Indeed, previous estimates for the seismic recurrence interval are generally greater than one hundred years and often several hundred years between major seismic events (Taylor et al., 1980, 1990). Although this may be the case, it would require substantially more frequent and/or larger seismicity at the trench axis to have significant impact on the results here.
Also, the treatment of events here, assumes normal plate convergence at the trench axis. This is not exactly correct. The trench axis runs along a line rotated roughly 19° west of north, an azimuth of 341° clockwise from north. This gives the trench normals an azimuth of 71° to the east of the trench axis or 251° to the west of the trench axis. In a fixed Australian plate reference frame, the plate convergence azimuth based on far-field stations is in the range of 256.0° to 258.6°
(Larson and Freymueller, 1995). Comparing these values shows a difference of 5.0° to 7.6° , suggesting that the convergence is slightly oblique. This difference, however, is relatively small and should have only limited effect since the majority of seismic events occur in reasonable proximity to the trench axis.
Lastly, the investigation here has been limited to the nearly linear portion of the trench axis. The northern and southern ends of the New Hebrides Trench gradually bend to follow more east-west routes - westerly to the Solomon Islands in the north, and easterly towards the Fiji Islands in the south. This is likely to result in an unusually complex tectonic interaction in these regions. For this reason, the investigation here has been limited to the nearly linear portion of the trench axis.