5.1 Absence of Oceanic Crust Arrival
In subduction zone environments, RFs are commonly used to investigate
plate interface structure since the method exploits the conversion of
incident P waves from a teleseismic event to S waves at significant
seismic-velocity discontinuities. RFs have identified LVZs along the
plate interfaces in subduction zones globally as negative amplitude
pulses atop positive amplitude pulses at slab depth (Bostock, 2013;
Audet & Kim, 2016). This dipole character has been observed in the
Japan (Kawakatsu & Watada, 2007; Akuhara et al., 2017), Cascadia
(Janiszewski & Abers, 2015; Ward et al., 2018), Costa Rica (Audet &
Schwartz, 2013), Mariana (Tibi et al., 2008), Alaska (Ferris et al.,
2003), and the central Mexico (Pérez-Campos et al., 2008; Y. Kim et al.,
2012) subduction zones. Depending on how far down dip the study area is
located, the negative pulse is typically interpreted as hydrated oceanic
crust or mantle hydrated by fluid expelled from the subducting slab due
to the low S-wave velocities observed, while the positive amplitude
pulse is generally the slab Moho. In Cascadia, Janiszewski and Abers
(2015) interpreted the LVZ as metamorphosed sediments, while Bangs et
al. (2009) interpreted the LVZ in Nankai as high porosity underthrust
sediment. In the northern 1964 segment, Y. Kim et al. (2014) also
observed this typical negative-to-positive character, attributing the
negative arrivals to an LVZ of subducted marine sediments along the
plate interface.
Neither our observed nor the preferred synthetic RFs (Figs 2 and 3)
feature the negative-positive dipole character observed within the
northern 1964 asperity, highlighting a major difference in RF character.
The lack of major arrivals before the positive slab Moho suggests that
beneath Kodiak, the seismic properties at the base of the upper plate
and top of the subducting slab may be similar. late interface material
is commonly inferred from trench sediment input to the subduction zone
(Morgan, 2004; Underwood, 2007) At the trench near Kodiak, both pelagic
sediments and sediment from the Surveyor Fan (von Huene et al., 2012;
Reece et al., 2011; Fig. 1a) comprise the subduction input. We suggest
that the subduction zone environment has altered the properties of any
subducted sediment at the interface, thus suppressing contrast between
the sediment and the surrounding rock. There is ample evidence from
magnetotelluric (Heise et al., 2012), laboratory (Miller et al., 2021)
and field studies of exhumed metasedimentary rocks from subduction zone
forearcs (Rowe et al., 2009; Rowe et al., 2013) pointing to instances of
hundreds of meters of metamorphosed sediments padding the plate
interface. Therefore, the absence of a well-defined LVZ channel at the
plate interface beneath our study area does not necessarily mean an
absence of subducted sediment. The metasedimentary rocks on Kodiak
Island are close enough in velocity (Miller et al., 2021) and density to
the Pacific crust that there is no significant discontinuity at the
interface to resolve with Ps RFs.
While the Ps RFs presented use relatively high frequencies for
teleseismic imaging (1.2 – 2.5 Hz), there may be coherent structural
layers that are too thin to be resolved. For example, using controlled
source seismic reflection data, J. Li et al. (2015) estimated a thin
600-900 m low-velocity channel at ~8-10 km depth along
the plate interface south of Kodiak Island inside the 1938 Mw 8.2 Semidi
rupture zone. A synthetic test of 2.5 Hz Ps RFs showed that a 750 m
thick LVZ would not be well resolved (Fig. S4(a)). We also tested using
higher frequency observations, 4.8 Hz, but the signal-to-noise ratio of
teleseismic sources decreases and the prominent velocity increase
interpreted as the slab Moho is only resolved sporadically across the
array (Fig. S4(b)). In areas where potential slab Moho arrivals are
observed in the 4.8 Hz RF image, we still do not find evidence for an
overlying LVZ (Fig. S4(b)). Thus, we cannot rule out a thin LVZ like
that imaged at shallower depths in the Semidi rupture zone (J. Li et
al., 2015), but we can be confident that a thicker LVZ
(~3-5 km) like that imaged by Kim et al. (2014) in the
Kenai asperity would be resolvable if it existed beneath our study area.
Mann et al. (2022) used scattered P and S coda of teleseismic P waves to
successfully image a continuous ~7-km thick low-velocity
layer lining the top of the subducted Yakutat crust. While we see
reverberations in sections of our profile (e.g., Fig. S5), their quality
is too low to allow for interpretation. The short deployment window
(~25 days) and the limited backazimuth distribution of
the events used in this study limits the usefulness of later arrivals.