Fluid Transport and Storage in the Cascadia Forearc inferred from Magnetotelluric Data

Subduction of hydrated oceanic lithosphere can carry water deep into the Earth, with im- 2 portant consequences for a range of tectonic and magmatic processes. Most ﬂuid is released at 3 relatively shallow depths in the forearc where it is thought to play a critical role in controlling 4 mechanical properties and seismic behavior of the subduction megathrust. Here we present 5 results from three-dimensional inversion of nearly 400 long-period magnetotelluric sites, in- 6 cluding 64 offshore, to provide new insights into the distribution of ﬂuids in the forearc of the 7 Cascadia subduction zone. Our amphibious dataset provides new constraints on the geometry 8 of the electrically resistive Siletzia terrane, a thickened section of oceanic crust accreted to 9 North America in the Eocene, and the conductive accretionary complex, which is being under- 10 thrust all along the margin. Fluids accumulate, over time-scales likely exceeding 1 My, above 11 the plate interface in metasedimentary units, while the maﬁc rocks of Siletzia remain dry. Flu- 12 ids in metasediments tend to peak at ﬁxed slab-depths of 17.5 and 30 km, suggesting control 13 by metamorphic processes, but also concentrate around the edges of Siletzia, suggesting that 14 this maﬁc block is impermeable, with dehydration ﬂuids escaping up-dip along the megath- 15 rust. Our results demonstrate that lithology of the overriding crust can play a critical role in 16 controlling ﬂuid transport and sequestration in a subduction zone, with potentially important 17 implications for mechanical properties.

In Oregon and southern Washington Siletzia is very thick in the east, thinning abruptly to ∼10 71 km in a layer that extends offshore. The conductive accretionary complex is thrust a significant 72 distance (∼50 km, more in places) under this thin layer, terminating at the nearly vertical edge of 73 R1. Offshore, conductivity of C1 is highest from 44-47 • N; however, this may reflect limited off-74 shore data coverage outside of this area (Fig. 1). A deeper, higher amplitude, peak in conductivity 75 (C2) occurs in all sections shown at a slab interface depth of 30-35 km, sometimes shallower. C2 commonly appears as a second peak within a broader conductive layer encompassing C1. Finally, 77 there are conductive upwellings (C3) that rise towards the surface beneath the Cascade arc, beyond 78 the eastern edge of Fig. 2. These arc conductive features [32] can be seen more clearly in Fig.   79 Resistive material extends beyond the southern edge of Siletzia into the Klamath terranes but To provide a more complete view of the geometry of key features, we derive two plan-view 83 images from the 3D resistivity model. First, we compute the depth extent of the resistors (R1-84 R4), defined by the bottom of the 300 ohm-m isosurface. Fig. 3a  tend to concentrate where Siletzia is thin (Fig. 3a). Variations in the western limit of arc vol-     The pattern of conductive features near the plate interface places constraints on fluid pathways, 158 storage, and egress. Fluid inputs at the trench include pore water and hydrous minerals, both in the 159 sediment layer, and in the underlying mafic oceanic crust. Free water in sediments is lost through 160 compaction at depths < 10 km [28,49], and is thus not a likely source for the patches of high con-161 ductivity, which are all deeper. Similarly, dehydration of clay minerals should be largely complete 162 at temperatures of ∼150 • C, which in Cascadia is well above the depth of the imaged conductors.

163
That there is a peak in conductance at a relatively constant slab depth centered at ∼17.5 km sug-164 gests a possible fluid source at corresponding P/T conditions. We suggest this peak may at least 165 in part result from pore collapse in the meta-basalts, which is believed to occur at temperatures of  of ∼ 100 S, and could be present over much of the forearc without violating the MT data (Fig. 3b).   (Fig. 3c), coincident with the thickest parts of Siletzia (Fig. 3a). This has often been      The model grid has a resolution of 11×9.5 km horizontally, with vertical layers logarithmically 276 spaced, starting from very fine layers (20 meters) to discretize the bathymetry and the seafloor sed-277 iments. The resulting core grid is 120×108×57 cells in latitude, longitude and depth respectively.

278
Edges were padded with 7 logarithmically increasing grid cells on all sides, to extend the bound-279 ary of the study area ∼600 km in all directions. We ran the inversion more than 30 times, to test 280 effects of various parameters (e.g., thickness of the slab, effect of offshore sites, effects of seafloor 281 sediments, covariance parameters) and inversion strategies.

282
The seafloor data presented some challenges to the 3D inversion. Lack of short period data 283 makes it difficult to constrain shallow structure such as the seafloor sediments. Inversion tests 284 demonstrated that including a priori a realistic distribution of conductive ocean sediments was 285 essential; anomalous phases of the nominal TE mode (coast-parallel electric field) impedances, 286 both for onshore and offshore sites near the coast could not be fit without imposing this structure.

287
Note that including a sediment layer greatly reduces the resistivity contrast at the seafloor, likely an 288 important factor in improving performance of the inversion. Even with the sediment layer imposed 289 fitting the offshore TE mode data was a challenge, which we found could be mitigated with a two- The data that support the findings of this study are publicly available online at http://ds.iris.edu/ds/products/emtf/.