Controls on Bending‐Related Faulting Offshore of the Alaska Peninsula

Oceanic plates experience extensive normal faulting as they bend and subduct, enabling fracturing of the incoming lithosphere. Debate remains about the relative importance of pre‐existing faults, plate curvature and other factors controlling the extent and style of bending‐related faulting. The subduction zone off the Alaska Peninsula is an ideal place to investigate controls on bending faulting as the orientation of the abyssal‐hill fabric with respect to the trench and plate curvature vary along the margin. Here, we characterize faulting between longitudes 161°W and 155°W using newly collected multibeam bathymetry data. We also use a compilation of seismic reflection data to constrain patterns of sediment thickness on the incoming plate. Although sediment thickness increases over 1 km from 156°W to 160°W, most sediments were deposited prior to the onset of bending faulting and thus should have limited impact on the expression of bend‐related fault strikes and throws in bathymetry data. Where magnetic anomalies trend subparallel to the trench (<30°) west of ∼156°W, bending faults parallel magnetic anomalies, implying that bending faults reactivate pre‐existing structures. Where magnetic anomalies are highly oblique (>30°) to the trench east of 156°W, no bending faults are observed. Summed fault throws increase to the west, including where pre‐existing structure orientations are constant (between 157 and 161°W), suggesting that another factor such as the increase in slab curvature must influence bending faulting. However, the westward increase in summed fault throws is more abrupt than expected for gradual changes in slab bending alone, suggesting potential feedbacks between pre‐existing structures, slab dip, and faulting.


Introduction
Bending and loading of the subducting oceanic lithosphere at subduction zones causes the crust and upper mantle to flex, forming a bulge seaward of the trench that has been termed the outer rise (Bodine & Watts, 1979;Caldwell et al., 1976;Garcia et al., 2019).Flexure of the incoming plate and negative buoyancy of the downwelling slab puts the upper portion of the lithosphere under extension and results in normal faulting in the incoming plate (Chapple & Forsyth, 1979;Faccenda, 2014;Ranero et al., 2003).These normal faults, known as bending-related faults, are found at subduction zones around the globe and occur between the trench axis and outer-rise (Hilde, 1983;Masson, 1991).Bending-related faults form between the trench and outer-rise, and slip occurs on these faults both prior to subduction as the plate approaches the plate interface and post-subduction as the plate bends, and eventually unbends and dehydrates (e.g., Ranero et al., 2005).
In this study, we use a compilation of multibeam bathymetry data, including recently acquired data from the Alaska Amphibious Community Seismic Experiment (AACSE) in 2018-2019 (Barcheck et al., 2020) to characterize bending-related faults in detail, including orientations, lengths, spacing, and throws.To test models for controls on faulting, we compare these bending-fault characteristics with the orientations of pre-existing abyssalhill faults from magnetic data, orientation of the trench, and changes in dip of the incoming plate along strike.

Tectonic Background
The subduction zone offshore of the Alaska Peninsula is an ideal location to examine controls on the formation of bending-related faulting (Figure 1).The subducting plate has an age of ∼55 Ma throughout the study area (Lonsdale, 1988) and is subducting nearly orthogonally at a rate of 63 mm/yr (Sella et al., 2002).The strike of the trench axis also remains relatively uniform at an azimuth of ∼70°through the study area, which spans longitudes 155-161°W.The consistent plate age, trench axis strike, and convergence rate lead to a nearly constant thermal structure of the subduction zone.
Although the age and convergence direction are constant, the dip of the slab and orientation of pre-existing structures vary along the strike.The dip of the slab steepens from the Gulf of Alaska west to the Aleutians, including steepening between the eastern Semidi segment, longitudes ∼155-159°W, and the western Shumagin Gap, longitudes ∼159-162°W (Hayes et al., 2018;Kuehn, 2019; Figure 1).One possible cause for the eastward shallowing of the slab is the subduction of an oceanic plateau (the Yakutat block) in the easternmost part of the Alaska subduction zone; the buoyancy resulting from the thickened crust is thought to contribute to shallow slab subduction there (Worthington et al., 2012).Another possible cause for the westward increase in the slab dip is a transition from oceanic/continental subduction to oceanic/oceanic subduction; west of the primary study region, the subduction zone transitions to an oceanic/oceanic margin, which may promote stab steepening (Holt et al., 2015;Sharples et al., 2014).
The spreading history of the incoming oceanic crust also varies along the strike, separated by a remnant triple junction marking the relict Kula, Pacific, Farallon triple junction (Engebretson et al., 1985;Lonsdale, 1988).The remnant triple junction appears as a T-shaped feature in the magnetic data at ∼158°W (Figure 1).Cessation of spreading at this triple junction occurred between ∼43 and 44 Ma (Engebretson et al., 1985;Lonsdale, 1988).Oceanic crust formed from Kula-Pacific spreading is currently subducting in the Shumagin Gap (∼159-162°W) and western Semidi segment (∼155-159°W).This crust formed at fast spreading rates (half rates of ∼74 mm/yr, Engebretson et al., 1985), and magnetic anomalies trend slightly oblique to the trench axis by ∼20°.Oceanic crust in the eastern Semidi segment formed from Pacific-Farallon spreading at intermediate rates (half rates of ∼28-34 mm/yr, Engebretson et al., 1985).Magnetic anomalies in this crust trend ∼N-S and highly oblique to the trench (∼70°).Tectonic reconstructions by Fuston and Wu (2020) suggest the possible existence of a Resurrection plate and Kula-Resurrection ridge striking N-S that would have subducted beneath the Alaska Peninsula.The proposed Kula-Resurrection ridge would have been active from ∼60 to 40 Ma.
Previous work, based on lower-resolution bathymetry data along the Alaska-Aleutian trench that largely focused on a region further west than our primary study area, identified a connection between the trends of magnetic anomalies and strikes of bending-related faults (Masson, 1991;Mortera-Gutiérrez et al., 2003).Masson (1991) showed that the angle between pre-existing abyssal-hills inferred from magnetic anomalies plays a key role in whether bending reactivates abyssal-hills or forms new faults, including in the Alaska-Aleutian subduction.In the western Aleutians, between 179°E and 169°W, analysis of bathymetry data shows that fault strikes closely follow the oceanic spreading fabric, which is near parallel to the trench (<10°difference) (Masson, 1991;Mortera-Gutiérrez et al., 2003).Between 157°W and 169°W, bending faults show two strikes: one primary set following the inherited spreading fabric, and a secondary set parallel to the trench axis (Masson, 1991).The angle between the trench and abyssal-hill faults is up to 30°in that region, and bending fault orientations suggest both reactivation of inherited weaknesses and the formation of new bending faults paralleling the trench.Shillington et al. (2015) used 2D active-source seismic transects to show that the incoming plate outboard of the Shumagin Gap (∼159-162°W) is more pervasively faulted than outboard of the Semidi segment (∼155-159°W).They also found that the upper mantle of the slab has a larger area of reduced seismic velocities seaward of the Shumagin Gap compared to the Semidi segment, which they attributed to an increase in hydration and associated serpentinization to the west.The incoming plate outboard of the Shumagin Gap also has greater total seismicity, which could suggest a greater number of bending-related faults (Shillington et al., 2015).
In addition to possible along-strike changes in bending faulting and resulting hydration, the Alaska peninsula also exhibits changes in coupling (Drooff & Freymueller, 2021;Li & Freymueller, 2018), great earthquake history (Davies et al., 1981), seismicity at a range of depths (Shillington et al., 2015;Wei et al., 2021), and arc chemistry (Buurman et al., 2014;Wei et al., 2021), all of which have been proposed to be influenced by faulting and hydration of the incoming plate.The Semidi segment has a history of generating great (M > 8.0) earthquakes with a recurrence interval of ∼50-75 years (Davies et al., 1981), including the recent M8.2 Chignik earthquake in July 2021 in the western part of the Semidi segment (e.g., Elliott et al., 2022;Liu et al., 2022).GPS measurements show that the Semidi segment is highly locked overall and that locking increases to the east (Drooff & Freymueller, 2021;Li & Freymueller, 2018;Zhao et al., 2022).The Shumagin Gap, however, is only weakly coupled (<30% coupled).Great earthquakes along the megathrust appear to be less common in the Shumagin Gap, with the last occurring in 1847 or possibly 1788 (Davies et al., 1981).However, the eastern part of the deep Shumagin Gap recently ruptured in a M7.8 earthquake in July 2020 (Liu et al., 2022;Xiao et al., 2021) and hosted an intraplate M7.6 earthquake in October 2020 (Zhou et al., 2022).Greater roughness at the top of the subducting plate due to increased bending faulting (e.g., Li et al., 2018;Wang & Bilek, 2014) and fluids from the hydrated lithosphere (Cordell et al., 2023;Li & Freymueller, 2018) have been proposed to contribute to changes in locking and earthquake history.
There are also along-strike changes in intermediate depth earthquakes (Shillington et al., 2015) and calculated bvalues (Wei et al., 2021).Florez and Prieto (2019) showed that subduction zones with high b-values (a comparatively greater ratio of small earthquakes to large earthquakes) suggest a greater extent of dehydration reaction, and thus more water stored in the downgoing plate.The Semidi segment is characterized by a doubleseismic zone with moderate b-values and few earthquakes extending deeper than 100 km (Abers, 1992;Wei et al., 2021).This suggests that that the volume of water stored in the downgoing slab through bending faults at this segment is less than other subduction segments to the west.High b-value earthquakes in the Shumagin Gap extend to depths >200 km, implying greater amounts of water stored here (Wei et al., 2021).Finally, trace element geochemistry at volcanic centers in the Shumagin and Semidi segments shows that sediment input of source magmas is higher at the Semidi segment and water input is less (Wei et al., 2021).

Data and Methods
We map and characterize bending faults on the incoming plate offshore of the Alaska Peninsula using a compilation of existing bathymetry data.Recently deposited sediments have the potential to mask the bathymetric expression of bending faults; therefore, where it is available, we use existing seismic reflection data to map the total sediment thickness and thickness of sediment deposited near the trench axis during bending faulting, which we call trench fill.Finally, to examine possible controls on bending-related faulting, we use the trends of magnetic anomalies to estimate the strike of pre-existing faults and fractures, and use profiles of depth to top of igneous crust and bathymetry to calculate the dip of the incoming plate outboard of the trench.

Characterizing Bending Faulting in Bathymetry Data
We map bending-related faults between longitudes 161-155°W (Figure 2) using new high-resolution multibeam bathymetry data collected as part of AACSE (Barcheck et al., 2020) combined with other existing bathymetric data (Ryan et al., 2009; Figure 2).The combined data provide nearly complete, continuous coverage of an area spanning roughly 100 × 300 km with a 125-m grid resolution, forming an excellent basis for systematic identification and characterization (geometry and displacement) of faults based on their surface expression.While bathymetry data are limited due to their inability to quantify faults in the subsurface, the seismic data sets that enable subsurface quantification of faults (Section 3.3) are too incomplete to allow for a comprehensive characterization of the entire region.
The mapped region encompasses the incoming plate subducting in the Shumagin Gap (159-162°W) and Semidi segment (155-159°W).Previous efforts to examine bending faults offshore of the Alaska Peninsula used lower resolution (∼200 m) GLORIA and Seabeam swath data (Scanlon & Masson, 1992) or spatially limited swath data from modern sonar systems (Shillington et al., 2015).The availability of a new high-resolution bathymetry area provides the opportunity to map faults in greater detail and extent.Faults are mapped by hand using a processed bathymetry grid that is detrended and demeaned to produce relative elevation, which removes longwavelength variations and highlights faulting (Figure 2, Figure S8 in Supporting Information S1).Detrending is done by applying a cosine bandpass filter with corners at 10,000/3,000/1,000/100 m.These values were chosen based on the average and minimum spacings between bending-related faults.Only faults with minimum scarps of ±5 m in the detrended grid are included in this mapping effort.
Each mapped fault trace is resampled to 100-m intervals along-strike.We measure fault strike, fault dip direction (seaward or trenchward), and distance from the trench for each 100 m segment.We examine variations in bending fault strike along the subduction zone by calculated histograms of fault azimuth in 1°longitude wide bins (i.e., 156-157°W, 157-158°W, etc., Figure 3).
The maximum throw for each fault is estimated using a 4-km-long bathymetric profile orthogonal to the fault.The position of this profile was selected at the location along the fault estimated to have the maximum scarp height.Linear regression of hand-picked points from the bathymetric profile is used to calculate slope and 95% confidence intervals for the hanging wall and footwall seafloor surfaces.These surfaces are then used to calculate the vertical separation by subtracting elevations of the hanging wall surface from the footwall surface at a range of positions that could represent the intersection of the fault plane with the scarp face.The throw is calculated from vertical separation by assuming a dip of 60°(Figures S1 and S2 in Supporting Information S1).We use this approach because it accounts for variations in the bathymetric slope of the footwall and hanging wall, which is particularly important within 30 km of the trench where bending is the most severe.To examine along-strike variations in the total amount of fault slip, maximum throw estimates are summed within 0.5°-wide bins.We utilized 0.5°-wide bins to achieve relatively high resolution in the along strike variations in summed fault throws while still having enough faults in each bin.We note that varying the bin size does not impact the observed trends.(Barcheck et al., 2020;Ryan et al., 2009).Gray areas show regions without swath bathymetry coverage.The trench axis (Bassett & Watts, 2015) is shown by the thick black line in all panels.(b) Interpreted bathymetry data with mapped bending-related faults (yellow).Red lines in the eastern part of the study area that are oriented N-S are interpreted to show remnant abyssal-hill structures and not active bending-related faults.Rupture patches are shown on the overriding plate with colors matching those shown in Figure 1.(c) Mapped bending-related faults (thin black lines) overlain on a magnetic anomaly grid (Bankey et al., 2002).Note the rotation in the spreading direction from N-S west of ∼156°W to E-W east of ∼156°W.White dashed lines and boxed annotations show magnetic chrons based on interpretations from Lonsdale (1988).A remnant triple junction can be observed at 158°W.Green dots show magnetic anomaly picks from Elvers et al. (1967) used as a guide for the magnetic anomaly picks presented in this study (larger purple points).Notice the apparent absence of bending faults at the seafloor east of ∼156°W (and not at the triple junction), where the orientations of magnetic anomalies change.Magnetic anomaly orientations do not change at the triple junction due to a plate reorganization that occurred ∼53 Ma (Lonsdale, 1988).

Estimating Strikes of Pre-Existing Structures From Magnetic Anomaly Data
Inherited abyssal-hill fabric within oceanic plates can be estimated from magnetic anomalies, where abyssal-hill faults are expected to form parallel to the mid-ocean ridge due to normal faulting (e.g., Macdonald et al., 1996), and thus parallel to the trend of magnetic anomalies.Elvers et al. (1967) map isochrons in detail on the subducting Pacific plate in our study area.To estimate the likely orientation of pre-existing faults in our study area, we resample picks made by Elvers et al. (1967) (which closely follows magnetic anomaly peaks of the North American magnetic anomaly map created by Bankey et al. (2002)) to 1 km.Based on these magnetic anomaly picks, we estimate the strikes and standard deviations of abyssal-hill faults that may be reactivated during slab bending in 1°wide bins using four distinct anomalies in the study area (Figures 2c and 3).We compare these data with the trend of the trench axis of Bassett and Watts (2015).

Sediment Thickness and Basement Offsets From Seismic Reflection Data
We compile seismic reflection data collected on the incoming plate from the National Archive of Marine Seismic Surveys (NAMSS, Triezenberg et al., 2016), the ALEUT experiment (Bécel et al., 2015(Bécel et al., , 2017;;Kuehn, 2019;Li et al., 2015) and the AACSE experiment (Barcheck et al., 2020;Bécel et al., 2019) to determine total sediment thickness and the thickness of sediments within the trench throughout the study area (Figure 4).Previous studies have documented along-strike changes in sediment thickness on the incoming plate, with larger sediment thickness offshore of the Semidi segment (Straume et al., 2019).The thickness of trench fill sediments also varies along the margin (e.g., von Huene et al., 2012), and this portion of the sedimentary section has the highest potential to mask bending faults because it was deposited during the time of bending fault formation and development.We pick arrival times for the seafloor, base of trench fill, and top of igneous crust on time migrated profiles (Figure 5).The top of the igneous crust is marked by an irregular bright reflector arising from the contrast between sediments and the basalts of Layer 2A.Sediments deposited on the incoming plate outboard of the trench consist of pelagic sediments and terrigenous fan sediments of the Zodiac fan (Creager et al., 1973;Stevenson et al., 1983;von Huene et al., 2012).These sediment layers generally parallel the underlying oceanic crust on the incoming plate and reflectors within these deposits can be traced out of the trench onto the incoming plate.These Subtracting the seafloor from the base of trench fill and from the top of igneous crust provides thickness in the two-way travel time of the trench fill and total sediment section, respectively.We convert time to depth using a velocity of 1.8 km/s; this average velocity is based on seismic processing of ALEUT reflection data (Bécel et al., 2015).We use a single velocity for depth conversion because of uncertainty in both spatial distribution and depthdependent velocities for each of the three sediment packages on the incoming plate: pelagic sediments, terrigenous fan sediments, and Quaternary trench fill (Creager et al., 1973;Stevenson et al., 1983;von Huene et al., 2012).These uncertainties obviate the benefit of using a depthdependent velocity for conversion.To create a grid of total sediment thickness and trench fill on the incoming Pacific plate, we grid the resulting sediment thickness values using a nearest neighbor algorithm with a 100-km radius and 0.1°grid spacing (Figure 4).

Estimating Bending Angle From Bathymetry Data and Seismic Reflection Profiles
The dip of the slab at depth increases along strike (e.g., Hayes et al., 2018;Kuehn, 2019), and we seek to evaluate the contribution of changes in slab bending to observed patterns of faulting.We create a grid of the depth to the top of igneous crust by subtracting the grid of sediment thickness described in Section 3.3 from the bathymetry grid.We calculate the dip of the incoming plate near the trench by applying linear regression of 50-km-long trench-perpendicular profiles of the top of the igneous crust (Figure 6) following the method of Nishikawa and Ide (2015).For comparison, we also estimate the dip of the seafloor along the same profiles using the same bathymetry data as the fault mapping analysis (Figures 7 and S6 in Supporting Information S1).

Results
We map 255 bending-related faults offshore of the Alaska Peninsula between longitudes 161°W and 155°W.Bending faulting is observed progressively farther from the trench outboard of the eastern Semidi segment compared to outboard the western Shumagin segment.In the western part of the study area (west of 158°W), most bending faults are concentrated within 50 km of the trench, but farther east faults are observed up to 75 km from the trench (Figures 2 and S3 in Supporting Information S1).We do not observe any active bending-faults east of 156°W in the newly acquired bathymetry data; all the potential structures that can be mapped here are roughly oriented N-S (Figure 2b) and likely represent differential compaction over relict abyssal-hill spreading faults or associated structures.
Individual mapped faults have lengths of ∼10-20 km, with a maximum length of 48 km, are spaced ∼3 km apart, and have maximum estimated throws of up to 423 m.Fault length varies along the trench, with average lengths of 15.6 km, 13.9 km, 20.6 km, 19.7 km, and 10.6 km, in 1°bins from 161°W to 156°W.There are similar numbers of faults dipping toward the trench and away from the trench.However, faults that dip trenchward generally exhibit larger throws than those dipping seaward (Figure S3 in Supporting Information S1).Cumulative fault throw summed across all mapped faults increases markedly from east to west, with the eastern portion having both fewer total number of faults and smaller estimated fault throws than the west (Figure 7).
A comparison of bending fault azimuths with the orientations of the trench and magnetic anomalies shows that bending faults generally parallel pre-existing structures (Figure 3).Bending-related faults in our study area primarily have strike azimuths between 80 and 100°(Figure 3).West of 157.5°W, magnetic anomalies parallel inferred abyssal-hill faults, which dominantly strike E-W (azimuths of 80-90°).Here faults strike 10-20°oblique to the trench axis, which has a relatively uniform azimuth of ∼70°(Figures 2c and 3).Between 157.5 and 156°W, magnetic anomalies are less continuous and have more variable orientations, including some that strike ENE-  (Triezenberg et al., 2016) and multi-channel seismic reflection lines from ALEUT (Bécel et al., 2015(Bécel et al., , 2017;;Kuehn, 2019;Li et al., 2015) and AACSE (Bécel et al., 2019).WSW, nearly parallel to the trench axis.East of 156°W, magnetic anomalies strike ∼N-S and no bending faults are observed.
There are two key areas where fault strikes are not subparallel to magnetic anomaly trends.The first is at ∼159.25-160°W,where bending fault strikes show one dominant trend centered at ∼85°and a broad secondary trend ranging from ∼35 to 80°.This area also contains small mounds (Figure 2b) that we tentatively interpret as petit spot volcanic constructs due to their morphological similarity to petit spot volcanoes found at other subduction zones, such as the Japan trench (Hirano, 2011;Hirano et al., 2006).The second is located between 156 and 157°W where fault strike azimuths span a wide range from ∼35 to 100°(Figure 3).This region is just west of the  (Bécel et al., 2017) and (b) Outboard of the Semidi Segment, with few to no bending fault expressions in the sediments or at the top of the crust (Shillington et al., 2015).Insets show seismic line locations highlighted in red.Topography in the top of crust outboard of the Semidi segment (panel b) is largely caused by the formation of crust at moderate spreading rates, which creates a more faulted crust surface at formation than fast spreading rates offshore of the Shumagin Gap, and these features are likely not active bending-faulting.There is little evidence of faulting-caused deformation in the sediments in this region and observable deformation may be caused by differential compaction.(c) Roughly trench-parallel seismic reflection profile F988-19 (Triezenberg et al., 2016) showing a westward decrease in sediment thickness.Note the rougher crust surface to the east, also mentioned in panel B, created by slower spreading rates.magnetic anomaly trends rotating from N-S to ENE-WSW.In both of these regions, fault lengths are generally shorter than in the surrounding areas.
The dip of the incoming plate at the top of the basement near the trench gradually increases from east to west, from a dip angle of ∼1.6°seaward of the eastern portion of the Semidi segment between longitudes 156°W and 157°W to 2.2°seaward of the westernmost part of the Shumagin Gap between longitudes 160°W and 161°W (Figure 6).This increase in dip at the top of the basement near the trench mirrors the dip at the seafloor in the outer rise (Figure S6 in Supporting Information S1) and the gradual increase in slab dip and curvature observed at greater depths (e.g., Figure 1; Hayes et al., 2018;Kuehn, 2019).(a) Regional map showing the locations of profiles used to estimate the outer rise dip outboard of the trench and the grid of depth to the basement based on bathymetry data (Figure 1) and sediment thickness (Figure 4).Colored trenchperpendicular profiles match those in panels b-f.(b-f).Linear regressions (black dashed lines) through two bathymetric profiles (colored lines, see panel a for location) ∼0.5°apart.Dip angles (alpha) for each longitudinal bin labeled on the plot.For comparison, we also estimated dip of the incoming plate at the seafloor (Figure S6 in Supporting Information S1), which also shows a general westward increase in incoming plate dip.
Sediment thickness also varies along strike, with the thinnest sediment cover offshore of the Shumagin Gap (∼400 m) in the western portion of the study area and the thickest sediment cover (∼1,100 m) offshore of the Semidi segment in the east (Figure 4).The increase in total sediment thickness between ∼155°W and 157°W can largely be attributed to the precense of the Oligocene-Miocene aged terrestrial Zodiac fan, contributing >500 m of pelagic sediment and terrigenous turbidites (Creager et al., 1973;Stevenson et al., 1983;von Huene et al., 2012).This fan formed off the coast of the Pacific Northwest of the United States (off the coast of present-day Washington and Oregon) and was transported on the Pacific plate through its northward migration over the last 32-40 Ma (Stevenson et al., 1983).Sediment thickness varies from an average of 250 m to a maximum thickness of ∼1,600 m.The thickest sediments are observed in the central part of the Zodiac fan and near the head of the fan, and sediment thickness decreases toward the edges of the fan (Stevenson et al., 1983).Smaller-scale variations may be caused by filling in irregular seafloor topography.For example, this may occur in the eastern part of the study area where oceanic crust was accreted at intermediate spreading rates and is rougher than in the west.The area of thin sediments east of the study area between ∼152 and 148°W occurs in the area of the Patton Murray seamount chain (Figure 4).Pelagic sediments and terrigenous Zodiac fan sediments were deposited on the oceanic plate prior to bending and are thus expected to be offset by these younger bending faults.
We also quantify the thickness of trench-fill sediments as these sediments are deposited during bending and subduction and thus have the greatest potential to mask bending faulting at the seafloor.Trench fill sediments are deposited primarily by along-trench transport (von Huene et al., 2012).Although trench fill sediments are up to 1.5 km thick, they are generally confined to a narrow (∼10-20 km) region near the trench, and bending faulting extends more than 50-70 km from the trench (Figures 4 and 5).

Estimating Bending Fault Throw and the Impact of Sediment Cover on Fault Mapping
Cumulative bending-related fault throw is greatest outboard of the western Shumagin Gap between 159 and 161°W and decreases eastward from 158°W to 156°W; east of 156°W outboard of the eastern Semidi region, no bending faulting is apparent in the bathymetry or seismic data (Figures 2, 5 and 7).Faults observed in the seismic data east of 156°W in the igneous crust are oriented roughly N-S (Figure 2) and most likely features created during crust formation.The crust in this region formed at slower spreading rates than crust to the west (Engebretson et al., 1985), creating a rougher crust surface (Buck et al., 2005;Buck & Poliakov, 1998;Carbotte & Macdonald, 1994).Possible faulting that may exist in the sediment cover offshore of the eastern Semidi segment (Figure 5b) is therefore not likely caused by active faulting but may instead be caused by differential compaction of sediments over a fractured crust surface (Carvers, 1968).These results are consistent with previous seismic reflection imaging of bending faults in widely spaced seismic reflection profiles and with a documented westward increase in the frequency of outer rise earthquakes (Matulka et al., 2022;Shillington et al., 2015).
One possible contribution to the apparent eastward decrease in cumulative fault throw at the seafloor could be masking by sediments, which increases in thickness to the east (e.g., Shillington et al., 2015;Li et al., 2018, von Huene et al., 2012; Figure 4).However, the Zodiac fan and other pelagic sediments on the oceanic crust are older than active bending faults in the present-day outer rise.Active faulting due to plate bending should therefore cut these older sediments, and fault scarps should still be evident on the seafloor even in the region covered by the fan.Quaternary trench fill sediments, on the other hand, were deposited during bending faulting and thus could mask the seafloor expressions of these faults, but we find that these sediments are confined to a narrow (∼10-20 km) region near the trench and vary significantly in thickness along strike from existing seismic data, and thus cannot account for the systematic eastward decrease in bending faulting observed in bathymetry data.

Controls on Bending Faulting Strike Orientations and Throw
New constraints on bending faulting from this study offer the opportunity to examine controls on the orientations of bending faults and along-strike variations in cumulative outer rise fault throws.

Bending Fault Strike Orientations
Previous analyses on controls on bending fault formation (e.g., Billen et al., 2007;Masson, 1991) found that abyssal-hill faults are reactivated when they are oriented <25-30°of the trench and bending forms new faults when the angle between the spreading fabric and trench is >25-30°.Bending-related faults that reactivate preexisting abyssal hill faults are expected to strike parallel to magnetic anomalies generated by seafloor spreading, and newly formed faults are expected to strike parallel the trench axis.In our study area, most bending faults strike parallel to magnetic anomalies, suggesting that they formed by reactivation (Figure 3), consistent with the previous studies of this area (Masson, 1991;Shillington et al., 2015).This is also consistent with observations at other subduction zones where pre-existing structures are near parallel to the trench, including the western Aleutians (Masson, 1991;Mortera-Gutiérrez et al., 2003), offshore Nicaragua (Ranero et al., 2003;Van Avendonk et al., 2011), and in the Kuril subduction zone (Fujie et al., 2018;Kobayashi et al., 1998).
The average throws (∼300 m) and average spacing (∼3 km) of bending faults offshore of the Alaska Peninsula are similar to the characteristics of bending faults in other locations where bend faulting occurs primarily by reactivation of abyssal hill faults (e.g., Fujie et al., 2018;Ranero et al., 2003).Fault spacing is ∼5 km at the Kuril trench (Fujie et al., 2018) and ∼2 km at the Middle America Trench (Faccenda et al., 2009;Ranero et al., 2003) where abyssal-hill faults are reactivated.When new bending faults form and cut across pre-existing fabrics, they often have larger throws and are more widely spaced compared to reactivated faults, as is observed at the Chilean and northern Japan trenches (e.g., Fujie et al., 2018;Geersen et al., 2018).In the eastern part of our study area (<156°W), where pre-existing structures are oblique to the trench, bending faults are not observed (Figure 8).We discuss a possible explanation for the absence of bending faulting in the east in Section 5.2.2.
There are two regions within our study area (156-157°W and 159-160°W, Figure 3) that exhibit bending faults with a broader range of strikes that do not parallel magnetic anomalies.The region between 156 and 157°W occurs north of the relict ridge-ridge-ridge triple junction and thus may have more complicated pre-existing structures.Complex and evolving abyssal-hill faulting is observed near modern ridge-ridge-ridge triple junctions and other areas of spreading changes (Smith et al., 2011), and the same may have been true in this area.Large topographic features such as the structure near 156°W could locally influence stress state and fault orientation (Geersen et al., 2022).However, we assert that complex variations in pre-existing structures are the primary controlling factors between 156 and 157°W.We discuss the region between 159 and 160°W in Section 5.3.

Cumulative Bending-Related Fault Throw
We observe a significant westward increase in the summed throw of all mapped faults within 0.5°bins (Figure 7).Slab curvature is thought to be a primary control on the amount of slip on bending faults, where higher degrees of slab bending are expected to be associated with larger magnitudes of cumulative fault slip (Faccenda, 2014).In our study area, the bending angle of the incoming plate estimated from the dip of the top of the basement crust in seismic reflection data (Figure 6) steepens to the west, consistent with westward steepening of the slab dip and increase in slab curvature at depth (Figure 1; Buffett & Heuret, 2011;Hayes et al., 2018;Kuehn, 2019).This correlation suggests that the increase in slab dip could contribute to the westward increase in cumulative bending faulting.However, given the relatively modest changes in slab curvature along-strike of ∼1°over 5°of longitude, it is surprising that we observe such a large and abrupt along-strike change in the amount of bending faulting: from no discernible bending faulting east of 156°W to significant bending faulting between 159 and 161°W.Therefore, while changes in slab dip likely contribute to the westward increase in observed cumulative bending faulting, other factors appear necessary to explain the relatively abrupt along-strike change in summed fault throws.
An abrupt change in pre-existing fabric in the subducting plate provides one possible explanation for the abrupt observed change in observed summed fault throws between 156 and 158°W.Magnetic anomaly patterns suggest that pre-existing structures west of 158°W are E-W striking and thus near parallel to the trench and favorable for reactivation, while east of 156°W they strike ∼N-S up to 70°oblique to the trench (Figure 2c; Shillington et al., 2015) and are unfavorable for reactivation.In the transition between these domains (156-158°W), the neartrench magnetic anomalies generally trend E-W, but they become weaker and less linear as one moves east, perhaps due to proximity to the relict triple junction directly to the south, and the transition to orthogonal (N-S) fossil ridge orientation directly to the east, as discussed above (Figures 2c and Engebretson et al., 1985;Lonsdale, 1988).Abundant faults are observed within this transition (Figure 3), but their cumulative slip is relatively modest (Figure 7), perhaps due to the complicated spreading fabric.Reactivation of remnant structures, which are estimated to be ∼30% weaker than the surrounding crust (Billen et al., 2007), could allow extensional strain in the upper lithosphere to be accommodated by faulting west of ∼158°W.In contrast, to the east, where favorably oriented weaknesses diminish and eventually disappear, bending stresses may not exceed the yield strength of the upper lithosphere and thus limited faulting occurs.This interpretation implies that bending stresses alone may be insufficient to promote the formation of outer rise faults in the oceanic lithosphere in locations that have modest slab dips and that lack inherited weaknesses.For comparison, slab dip is significantly steeper (by ∼2°) in other subduction zones where new bending faults form without reactivating pre-existing structures (e.g., Japan and Chilean subduction zones (Fujie et al., 2013;Nishikawa & Ide, 2015;Ranero et al., 2005).
It has been hypothesized that weakening of the oceanic plate by faulting at the outer rise and associated hydration and serpentinization could provide a positive feedback to induce additional slab bending and outer rise faulting (Billen & Gurnis, 2005;Contreras-Reyes & Osses, 2010;Faccenda et al., 2012;Hyndman & Peacock, 2003;Ranero et al., 2003).In Alaska, favorably oriented pre-existing structures may be important for this feedback to initiate as they allow faulting and hydration even at modest bending angles.(Dziewonski et al., 1981;Ekström et al., 2012) for the Oct 2020 M7.6 intraplate event and earthquakes two weeks after the M7.6 (points colored by depth; from the Alaska Earthquake Information Center).Highlighted between the dashed blue lines is an area of complex bending faulting between ∼159.25 and 160°W, which lies immediately updip of the 2020 M7.6 intraplate event.Interpreted petit-spot volcanism (white triangles) at ∼159.25°W occurs on the eastern edge of the region of complex faulting.Large megathrust rupture patches are shown on the overriding plate colored the same as in previous figures.
The long wavelength over which slab dip steepens along the Alaska subduction zone primarily reflects the transition from flat-slab subduction in the Gulf of Alaska to the east (Davis & Plafker, 1986;Petersen et al., 2021) to normal ocean-ocean plate subduction to the west.The rapid transition in bending-related faulting observed in our region may induce an additional short-wavelength transition in plate weakening that may enhance westward slab steepening.A westward reduction in plate strength at the outer-trench slope is also consistent with a decrease in the distance from the trench where bending faulting initiates: up to ∼75 km from trench at ∼157°W but confined to <50 km from trench farther west (Figure 2).Bending and faulting are expected to occur over larger wavelengths for stronger plates.Similar relationships between pre-existing structures, outer rise faulting, and slab bending angles are observed in the Middle America subduction zone offshore Costa Rica and Nicaragua (Ranero et al., 2003).
In summary, we propose that the combination of along-strike changes in slab dip and the orientation of preexisting structures with respect to the trench best explains a relatively abrupt along-strike change in the amount of faulting.The steeper slab dips and favorably oriented pre-existing structures, which are weaker than the surrounding crust, allow pervasive bending faulting in the west where larger faults with greater throws form.In the east, bending stresses associated with modest slab bending may not exceed the yield strength of the lithosphere, and pre-existing structures are highly oblique to the trench, so limited faulting is observed.Feedbacks between pre-existing structures, bending faulting, and plate weakening due to bending faulting may further promote faulting in the west (e.g., Billen & Gurnis, 2005;Contreras-Reyes & Osses, 2010).

Complex Faulting and Possible Linkage to October 2022 M7.6 Intraplate Event
Between ∼159 and 160°W, we observe an area of relatively complex bending-related faulting (Figures 3 and 8).Outside this area, fault orientations generally exhibit a single dominant peak in the azimuth centered around the average trend of magnetic anomalies (∼85-90°).Within the complex zone, a peak is still observed at ∼85°, but with an additional broad plateau with abundant faulting spanning orientations between 35 and 80°(Figure 3).At ∼159.25°W, we also observe a series of features that we interpret as small volcanic constructs (Figure 8).Similar features are recognized in the outer rises of other subduction zones (e.g., Japan trench; Fujie et al., 2020;Hirano, 2011;Hirano et al., 2006) and categorized as petit-spot magmatism.Off Japan, petit-spot volcanic provinces do not geochemically resemble mid-ocean ridge melts or occur near hotspot centers and are thus hypothesized to be caused by partial melting of the asthenosphere induced by plate bending and fracturing (Hirano et al., 2006).
The complex faulting in this region suggests comparable complexity in the pre-existing structures or stress state of the incoming plate in this region.Magnetic anomalies are relatively continuous through this region, and thus, there is no evidence for the former here.Given the short length scales associated with the complexity, we require a mechanism that can produce relatively abrupt changes in plate stress.Geodetic observations indicate a relatively abrupt change in megathrust coupling between the Shumagin Gap (∼159-162°W) and the Semidi segment (∼155-159°W; Drooff & Freymueller, 2021;Xiao et al., 2021), with a transition approximately coincident with the region of complex faulting.Along-strike variations in megathrust coupling and coseismic slip have been invoked to explain differences in incoming plate seismicity in many subduction zones (e.g., Christensen & Ruff, 1988;Emry et al., 2014), and it is possible that changes in coupling could also cause complexities in stress in the incoming plate over long time periods and thus explain the complex bending-related faulting observed.
Complex patterns of stress and faulting within the incoming plate could also promote the generation of small amounts of melt and intraplate volcanism.Valentine and Hirano (2010) and Hirano (2011) suggest that localization of tensional tectonic forces can enable partial melting in the lower lithosphere, producing intraplate volcanic provinces.Thus, varied stresses in the downgoing plate caused by differential coupling of the slab interface may localize tectonic stresses in the downgoing plate such that melts are generated and accumulate in this transition region and extrude as petit-spot volcanic features.
The region of complex faulting and petit-spot volcanism that we observe between 159 and 160°W lies updip from the Oct 2020 M7.6 intraplate event, raising the possibility that they could have related origins.Two causes have been proposed for this enigmatic earthquake, which appears to have ruptured a steep fault in the subducting plate that strikes ∼15°and thus orthogonal to the trench: (a) reactivation of remnant spreading features produced at the Kula-Resurrection ridge, which is now subducted (Fuston & Wu, 2020;Jiang et al., 2022); (b) accumulated shear stresses caused by lateral variability in slab dip and coupling (Herman & Furlong, 2021).In the first case, the 2020 M7.6 event is modeled as right-lateral strike-slip motion on an N-S striking fault dipping steeply to the east (Jiang et al., 2022), with slip distribution and associated aftershocks extending to within 30 km laterally from the zone of complex faulting.If this event reactivates hypothesized pre-existing Kula-Resurrection fabric just north of the trench (Fuston & Wu, 2020), then it is plausible that persistent slip on this feature has induced static stress changes in the incoming plate just up-dip of the fault tip (Yang et al., 2023) that are sufficient to perturb the bending stresses and associated fault orientations.In the second case, accumulated stresses in the subducting plate arising from lateral variability in coupling between the Shumagin Gap and Semidi segment could explain both the occurrence of the M7.6 intraplate earthquake (Herman & Furlong, 2021) and the complexities observed in bending-related faulting patterns outboard of the trench.Although more work is needed to evaluate the influence of changes in megathrust coupling on long-term deformation in the incoming plate, the spatial proximity of the earthquake and complex faulting imply a common origin.

Implications for Hydration
One major importance of bending-related faulting is its role in allowing ingress of seawater and hydration of the crust and upper mantle of the incoming plate (Cai et al., 2018;Faccenda, 2014;Faccenda et al., 2009;Grevemeyer et al., 2018;Ivandic et al., 2008;Korenaga, 2017;Nishikawa & Ide, 2015;Peacock, 2001;Ranero et al., 2003;Shillington et al., 2015).Extensional faults on the incoming plate are thought to act as conduits for seawater to percolate several kilometers through the crust and into the upper mantle of the incoming plate.The water may reside as fluid-filled cracks in the crust and upper mantle (Miller et al., 2021) as well as react with the peridotites to form serpentinite (Carlson & Miller, 2003;Faccenda, 2014;Faccenda et al., 2009;Grevemeyer et al., 2018;Korenaga, 2017;Peacock, 2001;Ranero et al., 2003).Seismic velocity models from subduction zones around the globe show the reduced velocities in the crust and upper mantle of the subducting plate near the trench, which are interpreted to represent the presence of hydrous minerals and/or fluid-filled cracks (Cai et al., 2018;Contreras-Reyes et al., 2007, 2011;Fujie et al., 2018;Ivandic et al., 2008;Shillington et al., 2015;Van Avendonk et al., 2011).P-wave velocity models in the study area exhibit a more pronounced reduction in seismic velocity in the upper mantle of the incoming plate outboard of the Shumagin Gap (western portion of the study area; west of ∼157°W) than in the Semidi segment (eastern portion of the study area; east of ∼157°W), suggesting greater hydration in the west (Shillington et al., 2015).Likewise, high-resolution P-wave models from streamer tomography also show a reduction in velocity in the upper crust off the Shumagin Gap, interpreted to arise from a combination of faulting and alteration (Acquisto et al., 2022).
The combination of preferentially oriented pre-existing structures and an increase in slab dip outboard of the Shumagin Gap (west of ∼157°W) promote more bending faulting and are thus expected to produce increasing hydration of the incoming lithosphere, in comparison to the region outboard of the Semidi segment (east of ∼157°W ) where the number and size of bending faulting is lower, A westward increase in hydration can be inferred from the westward increase in number of faults and larger fault throws and is consistent with observations of a doubleseismic zone in the downgoing plate in the Shumagin Gap (Wei et al., 2021), prominent conductors at depth in magnetotelluric data (Cordell et al., 2023), and geochemical signatures consistent with fluids in arc volcanism in the western Shumagin Gap (Wei et al., 2021).
Faults to the west of our study area also appear to have more irregular fault orientations (Figures 2b and 3) and may form interconnected fault networks as they grow and merge near the trench, thus creating a larger and more pervasive damaged zone.This interconnected fault network to the west could also promote increased hydration.

Implications for Plate Boundary Properties
Geodetic studies of the Alaska Peninsula show that the western Shumagin Gap is <30% coupled, whereas the eastern Semidi segment is almost entirely locked (Drooff & Freymueller, 2021;Li & Freymueller, 2018).Changes in subduction zone inputs could influence changes in coupling here (Shillington et al., 2015).Westward increases in the total number and throws of bending-related faults on the incoming plate, combined with the westward decrease in sediment thickness, can influence the heterogeneity of the megathrust interface once subducted.Possible petit-spot features at ∼159.25°W may also locally contribute to a rougher plate interface in the western part of our study area.Rough seafloor is proposed to promote creeping of the megathrust interface and numerous small to medium (<M7.5)events (Wang & Bilek, 2014) and potentially contribute to slow slip events (e.g., Saffer & Wallace, 2015).Higher degrees of bending faulting to the west are also expected to result in greater hydration of the crust and upper mantle, stored in hydrous minerals or as free water.Dehydration and migration of fluids at depth could also influence megathrust properties and behavior (e.g., Cordell et al., 2023;Saffer & Tobin, 2011;Saffer & Wallace, 2015).Thus, increased bending faulting and thinner sediment cover on the incoming plate in the western part of the study area may lead to a heterogeneous and fluid-rich megathrust interface and promote creep.The lack of bending faulting and larger amounts of sediment entering the subduction zone in the eastern Semidi segment (Li et al., 2018) could contribute to greater megathrust homogeneity.At shallow depths, the thicker subducted sediments in the Semidi segment appear to be overpressured (Li et al., 2018) and may contribute to the observed aseismic slip (e.g., He et al., 2023).At greater depths, thicker subducting sediments could contribute to a relatively homogenous plate boundary, allowing for increased locking and propensity to rupture in earthquakes in the eastern portion of our study area (Li et al., 2018;Ruff, 1989).
On the other hand, thinner sediment and rougher seafloor subducting in the Shumagin segment may promote shallow seismic slip and tsunamigenesis (Bécel et al., 2017).These properties are commonly associated with the incoming plate in subduction zones with shallow, tsunamigenic earthquakes, including tsunami earthquakes (Polet & Kanamori, 2000).Subduction zones with low coupling can host infrequent tsunamigenic earthquakes.For example, the 1946 M8.6 tsunami earthquake ruptured a weakly coupled region to the west of our study area (López & Okal, 2006).Bécel et al. (2017) proposed that the properties of the Shumagin sector, including the rough seafloor and paucity of sediments, may make it prone to tsunamigenic shallow slip when infrequent large earthquake rupture into this weakly coupled segment, as may have occurred in 1788 (Davies et al., 1981).
We acknowledge that there are uncertainties associated with projecting features observed outboard of the subduction zone to seismogenic depths.The position of features at depth is influenced by slab geometry and convergence direction (Harmon et al., 2019).In the study area, there are significant uncertainties in the continuation of the Zodiac fan at depth (von Huene et al., 2012), and competing reconstructions for the spreading history in this region predict different configurations of pre-existing structures at depth (e.g., Fuston & Wu, 2020).Nonetheless, the correlations between incoming plate properties, changes in locking and seismicity at a range of depths, and arc lava chemistry imply that subduction zone inputs contribute to patterns of plate boundary behavior (Shillington et al., 2015;Wei et al., 2021).

Conclusions
Analysis of new high-resolution bathymetry data collected outboard of the Shumagin Gap and Semidi segment provides new insights into controls on formation and patterns of bending-related faulting in the outer-rise of the incoming Pacific plate.
1. Bending-related faults strike dominantly parallel to magnetic anomalies, indicating that bending primarily reactivates relict abyssal-hill faults originating at oceanic plate formation.The angle between magnetic anomalies and the trench controls the bending fault strike, where reactivated faults parallel magnetic anomalies and newly formed faults parallel the trench.2. The plate bends more steeply to the west in the Shumagin Gap region, where observed faulting is more extensive and where larger faults with greater throw form.These observations suggest that increased bending of the downgoing plate is likely a contributing factor to the westward increase in summed fault throws.However, feedbacks between pre-existing structures, slab weakening, and bending and faulting appear necessary to explain the relatively abrupt along-strike changes in the amount of bend faulting.3. The subducting plate updip of the M7.6 intraplate earthquake (between 159 and 160°W) exhibits relatively complex bending faulting and petit spot volcanism.Variations in coupling and slab dip could contribute to both bending faulting patterns and the M7.6 earthquake here.4. The westward increase in bending faulting has important implications for incoming plate weaknesses and the ability for bending-faulting to pervasively hydrate the incoming plate at the western Shumagin Gap. 5. Thin sediment cover and pervasive bending-related faulting on the incoming plate outboard of the western Shumagin Gap promote a heterogeneous fluid -rich plate interface and creeping megathrust behavior at depth.Thick sediment cover and nearly absent bending-related faulting on the incoming plate outboard of the eastern Semidi segment promote a homogeneous megathrust, contributing to recurring great earthquakes.

Figure 2 .
Figure 2. (a) New high-resolution bathymetry data(Barcheck et al., 2020;Ryan et al., 2009).Gray areas show regions without swath bathymetry coverage.The trench axis(Bassett & Watts, 2015) is shown by the thick black line in all panels.(b) Interpreted bathymetry data with mapped bending-related faults (yellow).Red lines in the eastern part of the study area that are oriented N-S are interpreted to show remnant abyssal-hill structures and not active bending-related faults.Rupture patches are shown on the overriding plate with colors matching those shown in Figure1.(c) Mapped bending-related faults (thin black lines) overlain on a magnetic anomaly grid(Bankey et al., 2002).Note the rotation in the spreading direction from N-S west of ∼156°W to E-W east of ∼156°W.White dashed lines and boxed annotations show magnetic chrons based on interpretations fromLonsdale (1988).A remnant triple junction can be observed at 158°W.Green dots show magnetic anomaly picks fromElvers et al. (1967) used as a guide for the magnetic anomaly picks presented in this study (larger purple points).Notice the apparent absence of bending faults at the seafloor east of ∼156°W (and not at the triple junction), where the orientations of magnetic anomalies change.Magnetic anomaly orientations do not change at the triple junction due to a plate reorganization that occurred ∼53 Ma(Lonsdale, 1988).

Figure 3 .
Figure 3. Histograms comparing fault segment strike azimuths (blue bars) in 1-degree bins with the average trend of the trench (red dotted line, Bassett & Watts, 2015) and average trend of magnetic anomalies (black dotted line).The one-standard-deviation range of magnetic anomalies is shown by the gray box in the background.Dominant peaks of fault strikes primarily follow the trend of magnetic anomalies (panels a, c, d).Between 159 and 160°W (panel b), fault strike azimuths show one dominant trend following magnetic anomalies, and a secondary trend ranging from ∼35 to 80°.Between 156 and 157°W, there are fewer faults with a broader distribution of strike azimuths (panel e).Note that the vertical axis varies between panels.

Figure 4 .
Figure 4. Map of gridded sediment thickness based on legacy single-channel USGS seismic reflection lines(Triezenberg et al., 2016) and multi-channel seismic reflection lines from ALEUT(Bécel et al., 2015(Bécel et al., , 2017;;Kuehn, 2019;Li et al., 2015) and AACSE(Bécel et al., 2019).Seismic profiles are indicated by dashed white lines.The thickness of trench fill is indicated by open circles sized by thickness.The absence of a trench fill circle for trench-perpendicular lines represents a thickness of zero.Primary study area shown by dashed red box as in Figure 1.
Figure 4. Map of gridded sediment thickness based on legacy single-channel USGS seismic reflection lines(Triezenberg et al., 2016) and multi-channel seismic reflection lines from ALEUT(Bécel et al., 2015(Bécel et al., , 2017;;Kuehn, 2019;Li et al., 2015) and AACSE(Bécel et al., 2019).Seismic profiles are indicated by dashed white lines.The thickness of trench fill is indicated by open circles sized by thickness.The absence of a trench fill circle for trench-perpendicular lines represents a thickness of zero.Primary study area shown by dashed red box as in Figure 1.

Figure 5 .
Figure 5. Examples of seismic reflection profiles (a) outboard of the Shumagin Gap (ALEUT line 6) showing extensive bending faulting at the seafloor and top of the crust(Bécel et al., 2017) and (b) Outboard of the Semidi Segment, with few to no bending fault expressions in the sediments or at the top of the crust(Shillington et al., 2015).Insets show seismic line locations highlighted in red.Topography in the top of crust outboard of the Semidi segment (panel b) is largely caused by the formation of crust at moderate spreading rates, which creates a more faulted crust surface at formation than fast spreading rates offshore of the Shumagin Gap, and these features are likely not active bending-faulting.There is little evidence of faulting-caused deformation in the sediments in this region and observable deformation may be caused by differential compaction.(c) Roughly trench-parallel seismic reflection profile F988-19(Triezenberg et al., 2016) showing a westward decrease in sediment thickness.Note the rougher crust surface to the east, also mentioned in panel B, created by slower spreading rates.

Figure 6 .
Figure6.(a) Regional map showing the locations of profiles used to estimate the outer rise dip outboard of the trench and the grid of depth to the basement based on bathymetry data (Figure1) and sediment thickness (Figure4).Colored trenchperpendicular profiles match those in panels b-f.(b-f).Linear regressions (black dashed lines) through two bathymetric profiles (colored lines, see panel a for location) ∼0.5°apart.Dip angles (alpha) for each longitudinal bin labeled on the plot.For comparison, we also estimated dip of the incoming plate at the seafloor (FigureS6in Supporting Information S1), which also shows a general westward increase in incoming plate dip.

Figure 7 .
Figure 7. Along strike variations in panel (a) summed maximum throws on bending faults within 0.5°-wide bins; (b) dip of the subducting plate at the outer rise, estimated from bathymetric seafloor slope (light blue, Figure S6 in Supporting Information S1) and from the dip of the top of the crust based on a structure contour map of the base of incoming plate sediments (dark blue, Figure 6); (c) difference between the expected strike of pre-existing structures from magnetic anomalies and the trench; (d) incoming plate sediment thickness on the incoming plate at distances of 10-50 km from the trench.

Figure 8 .
Figure 8. Bathymetric map of bending-related faulting (yellow lines) with abyssal-hill structures (red lines).Also shown is the CMT solution(Dziewonski et al., 1981;Ekström et al., 2012) for the Oct 2020 M7.6 intraplate event and earthquakes two weeks after the M7.6 (points colored by depth; from the Alaska Earthquake Information Center).Highlighted between the dashed blue lines is an area of complex bending faulting between ∼159.25 and 160°W, which lies immediately updip of the 2020 M7.6 intraplate event.Interpreted petit-spot volcanism (white triangles) at ∼159.25°W occurs on the eastern edge of the region of complex faulting.Large megathrust rupture patches are shown on the overriding plate colored the same as in previous figures.