Figure 4. (a) Viscosity structure of minimum contribution of
viscoelastic relaxation models. The blue region indicates the elastic
lithosphere, the elastic slab and the code nose, the red region
indicates the mantle wedge, the green region indicates the oceanic
mantle and the rest of the continental mantle (b)Viscosity structure of
maximum contribution of viscoelastic relaxation mode. The blue region
indicates the elastic lithosphere and the elastic slab, the red region
indicates the mantle wedge, the green region indicates the oceanic
mantle and the rest of the continental mantle. (c), (d) The minimum and
maximum predicted 3-month viscoelastic relaxation only horizontal
displacements. The white barbed line shows the plate boundary between
the Pacific plate and the North American plate.
We use the biviscous Burgers body to model the viscoelastic relaxation
of the mantle wedge, as this model has been shown to improve fit to the
postseismic data in many past studies. Following past studies (e.g.,
Tian et al., 2020; Huang et al., 2020), we assume that the viscosity of
the Kelvin element of the Burger’s body is 1/10 of the viscosity of the
Maxwell element. We vary the Maxwell element viscosity in the range
(1-5)*1019 Pa-s. Huang et al. (2019) found the value
of the Maxwell viscosity of the mantle wedge to be
3*1019 Pa-s for the nearby 1964 Alaska earthquake. A
higher value of viscosity of the mantle wedge will result in lower
predicted displacements (Figure S1 a).
For the lithosphere thickness, the multichannel seismic (MCS) line ALEUT
3 (Kuehn, 2019) suggests an approximately 40km Moho depth for the
overriding plate at the region of this earthquake. The mantle
lithospheric thickness is not known, but needs to be added to the
crustal thickness. We thus vary the lithospheric layer thickness between
40km and 50km given previous postseismic models for Alaska (e.g., Huang
et al, 2020). A thicker lithospheric layer will result in lower
predicted displacements (Figure S1 b).
Many studies have shown the significance of considering a cold nose in
the viscoelastic relaxation modeling of subduction zone earthquakes
(e.g., Sun et al.,2014; Lou et al., 2021). According to the thermal
modeling of Syracuse et al. (2010), it is to assume the existence of an
essentially elastic cold nose, lthough the extent of the cold nose is
uncertain. Applying a cold nose will result in lower predicted
displacements (Figure S1 c).
Nearly all of the predicted postseismic displacements result from
relaxation of the mantle wedge material. We assumed a sub-oceanic mantle
Maxwell viscosity of 1020 Pa-s, based on Huang et al.
(2020) and Tian et al. (2020). However, if we made the sub-oceanic
mantle to be elastic (infinite viscosity), the predicted signal does not
change notably (Figure S1 d). Assuming a much lower viscosity for the
sub-oceanic mantle mainly affects the vertical model prediction for
sites near the updip end of the rupture, with little change to the
horizontal predictions. Therefore, in this study we do not further
consider variations in the sub-oceanic mantle viscosity.
4 Results