Variable In Situ Stress Orientations Across the Northern Hikurangi Subduction Margin

We constrain orientations of the horizontal stress field from borehole image data in a transect across the Hikurangi Subduction Margin. This region experiences NW‐SE convergence and is the site of recurrent slow slip events. The direction of the horizontal maximum stress is E‐W at an active splay thrust fault near the subduction margin trench. This trend changes to NNW‐SSE in a forearc trench slope basin on the offshore accretionary wedge, and to NE‐SW in the onshore forearc. Multiple, tectonic, and geological processes, either individually or in concert, may explain this variability. The observed offshore to onshore stress rotation may reflect a change from dominantly compressional tectonics at the deformation front, to a strike‐slip and/or extensional tectonic regime closer to the Taupo Volcanic Zone, further inland. In addition, the offshore stress may be affected by topography and/or stress rotation around subducting seamounts, and/or temporal stress changes during the slow slip cycle.

Episodic, shallow (<15 km) slow slip events (SSE), spanning timescales ranging from days to years, are recorded at several subduction zones, including the Nankai margin, offshore Japan (Araki et al., 2017;Hirose et al., 1999;Obara et al., 2004), the Costa Rican margin (Davis et al., 2015;Dixon et al., 2014), and the Hikurangi Subduction Margin (HSM) . Despite the recognized importance of stress in earthquake dynamics, data constraining stress states and variability on, and in the near field of, faults that host SSEs is limited (Chang et al., 2010;Huffman et al., 2016;Warren-Smith et al., 2019). This study provides new data on contemporary stress orientation variability across an area of recurring slow slip in the northern HSM. We characterize shallow crustal stress orientations from boreholes along a transect from the actively deforming frontal accretionary wedge, through the landward, offshore Tuaheni Basin (a forearc trench slope basin), and onto the onshore forearc ( Figure 1). We then discuss possible geological processes responsible for the observed stress orientation variations.

State of Stress at the Hikurangi Subduction Margin
The HSM accommodates westward subduction of the Pacific plate beneath the North Island of New Zealand at rates of ∼2-3 cm/year in the south, and ∼6 cm/year in the north (Figure 1a) (Wallace et al., 2004). Northward convergence rate increase at the trench is due to forearc clockwise rotation, which also creates changes in upper plate tectonics along-strike of the margin. Pacific-Australia plate convergence is oblique, with the margin-perpendicular component accommodated primarily along shallow subduction thrust faults, and the margin-parallel component accommodated by a combination of forearc rotation and strikeslip in the onshore North Island Dextral Fault Belt (Beanland & Haines, 1998;Wallace et al., 2004). Episodic SSEs are observed at shallow depths of 2-15 km at the northern HSM every 1-2 years, and involve ≤20 cm of slip on the plate boundary occurring over a few weeks (Wallace et al., 2010. There have also been moderate-sized subduction thrust earthquakes in the north Hikurangi SSE source region, including two M w 7.0-7.2 events in 1947 (Figures 1b and 1c) (Doser & Webb, 2003).
Aspects of the HSM stress state have been reported previously. In the onshore northern HSM, focal mechanism-derived S Hmax directions are NE-SW (oblique to relative plate motion, and approximately parallel to HSM margin strike), with suggested local variation to ENE-WSW S Hmax orientations in the area east of Gisborne (subparallel to relative plate motion, and HSM strike-oblique) (Townend et al., 2012). Borehole image logs (acquired within the upper 3 km of the crust) from three onshore wells (Kauhauroa-2, Kauhauroa-5, and Tuhara-1A) also show NE-SW S Hmax orientations in the Hawke's Bay region (Lawrence, 2018). Investigations of the principal contraction (strain) direction from GPS measurements along the HSM are highly variable, with dominantly east-west directions near the east coast, and with some rotation to NE-SW further inland (Dimitrova et al., 2016;Haines & Wallace, 2020). P p measured from drill stem tests and repeat formation testing, and inferred from mud weights in onshore wells, reveal shallow overpressures within basins of the upper plate (Burgreen-Chan et al., 2016;Darby & Funnell, 2001). The southern Wairarapa and south Hawke's Bay regions display lower overpressures than the northern Raukumara and north Hawke's Bay regions of the northern HSM. Overpressures are found within Cretaceous-Paleogene, smectite-rich seal rocks of the Wanstead Formation, are spatially variable within Neogene, low-permeability, mudstone units, and are uncommon within Quaternary units (Darby & Funnell, 2001). These shallow overpressures have been attributed to both disequilibrium compaction, where pore water flow is restricted during sediment compaction, and to porosity reduction associated with high horizontal compressive stresses related to plate convergence (Burgreen-Chan et al., 2016;Darby & Funnell, 2001). Finally, in the Hawke's Bay region, σ 3 magnitudes determined from leak-off tests performed in onshore wells are less than (though in places close to) vertical stress (S v ) magnitudes, suggesting that normal, strike-slip, and potentially reverse faulting stress regimes are experienced here (Burgreen-Chan et al., 2016).  frontal accretionary wedge, and the bedded sediments that comprise its hangingwall and footwall (Fagereng et al., 2019;Saffer et al., 2019). Hole U1519A is located within an upper continental slope sedimentary basin (Tuaheni Basin) and intersects bedded mudstones, siltstones, and sandstone packages, as well as mass transport deposits . Finally, Hole U1520A/B is located on the incoming subducting plate and intersects trench-wedge clastic sediments and a lower sequence of carbonates and volcaniclastics (Barnes et al., 2020).

GVR Resistivity Image Log Processing and Analysis
GVR resistivity image logs are analyzed in this study to identify and quantify the properties of stress-induced borehole features such as borehole breakouts. GVR logging uses rotating resistivity buttons to provide a 360° resistivity image of the borehole wall. This study reports borehole breakout orientation results determined from a high-resolution, post-expedition reanalysis of the GVR image logs which has resulted in a more detailed and accurate data set than the preliminary data provided from the shipboard analysis (Wallace et al., 2019b). Raw data was processed from the GVR tool following the procedure detailed in Wallace et al. (2019a). For this study, all GVR images (those generated from the shallow, medium, and deep resistivity button) were analyzed. The shallow button GVR image logs are the preferred image for data acquisition as they have a higher potential of recording features close to the borehole wall, including stress-induced features. The GVR image logs were statically and dynamically normalized to enhance resistivity contrasts for improved feature identification, with a 1 and 0.5 m normalization window used for the latter. Feature classification is based on criteria set out in McNamara et al. (2019). Quality rankings and circular statistics for stress induced borehole features identified from the GVR image logs follow World Stress Map criteria (Heidbach, 2016).

Borehole Image Stress-Induced Features
From all GVR image logs collected during IODP Expedition 372, 86 distinct borehole breakout pairs are identified in Holes U1518A/B and U1519A (Table 1; Figure 2); none were identified within other Holes. No drilling-induced tensile fractures, or petal centerline fractures are observed on any GVR image logs. Borehole breakouts, were observed, are more common at deeper depths. No systematic correlation between the occurrence of borehole breakouts and well logging units is confidently observed. Individual borehole breakouts are observed to cross multiple individual bedding layers in both U1518 and U1519 Holes.
Borehole breakout azimuths are variable between the two drillsites and with depth in each individual Hole ( Figure 2). Stress indicator quality ranking for both holes is "D" meaning the stress orientations observed are likely more representative of local stress states than regional stress states (Tingay et al., 2008). Eight of the 10 borehole breakout pairs identified in Hole U1518B occur between 518 and 591 mbsf (meters below seafloor) and have an average orientation of 003° ± 6°/180° ± 4° (Figure 2a). These lie in the footwall of the Pāpaku fault that was intersected at 315-348 mbsf (Cook et al., 2020). From these breakout orientations, we report an S hmin orientation of N-S, and infer an S Hmax orientation of E-W (Figures 2a and 2b). Orientation trend variations between borehole breakouts <10 m apart ranges from 3° to 16° within this interval (Figure 2a). In the hanging wall (∼210-220 mbsf), two borehole breakout pairs (low resistivity zones, ∼180° apart around the borehole, associated with increased caliper values) are identified and have 065°/252° and 056°/236° azimuths (Figure 2a), from which a NE-SW S hmin and NW-SE S Hmax orientation is inferred.

Discussion
The stress orientation data presented here are the first direct measurements made across the offshore northern HSM (Figure 3). At the borehole scale, S Hmax rotates from an E-W orientation at the Pāpaku fault, an active splay thrust fault near the deformation front (Site U1518), to a NNW-SSE orientation above the plate interface where peak slip in SSEs occurs (Site U1519), and finally to a NE-SW orientation in onshore wells ( Figure 3). Recent investigation of seismic anisotropy across this HSM offshore region shows fast azimuths align with observed U1518 and U1519 borehole-derived S Hmax orientations (Arai et al., 2020;Zal et al., 2020). Geological and tectonic influences on stress direction variability at both the well scale and across the northern HSM are discussed here.

Well-Scale Stress Orientation Variability at Sites U1518 and U1519
The rotation of S Hmax within Hole U1518B from NW-SE direction in the Pāpaku Fault hanging wall to E-W in the footwall is indicated by a small number of data points. This depth-related stress rotation, if real, may be explained by a number of processes acting in combination or separately. First, as the Pāpaku Fault is an active splay fault near the trench, recent slip on this structure may have perturbed the stress field. Second, the NW-SE S Hmax orientation is observed within a depth interval of core-scale hanging wall folds and fractures (Fagereng et al., 2019). LWD resistivity image logs of this same depth interval also reveal the presence of fracturing and bedding orientation patterns consistent with large-scale folding . Recent slip on faults and fractures associated with this deformed depth interval may have generated a local rotation on the stress directions in this region. Third, east of Hole U1518B, there is an abrupt change in topography which may influence the formation and orientation of the shallower stress-induced wellbore failures observed here (Figure 1c; M. L. Zoback et al., 1989). This topography may also have a broader effect on the overall observed E-W S Hmax orientation observed at Site U1518. Modeling of the spatial influence of the MCNAMARA ET AL.  (Mardia, 1975;Tingay et al., 2008) and show values generated for both pairs of borehole breakout sets. N = number of borehole breakout pairs. Stress indicator quality ranking determined from Tingay et al. (2008). mrsl, meters relative to sea level; mbsf, meters below seafloor. a Average and standard deviation values for Hole U1518B are calculated from the eight borehole breakouts observed in the Pāpaku fault footwall due to low confidence in the two observed in the hangingwall.

Table 1 Stress Indicators From Analyzed LWD Resistivity Image From IODP Expedition 372
topographical feature on local stress conditions at Site U1518, and more than two borehole breakout measurements are required in this region to confirm and explain the causes of stress rotations with depth here.
The small-scale variation in borehole breakout orientation within Hole U1519A is likely due to stress field perturbation from localized deformation in these depth intervals. This is supported by the large number of MCNAMARA ET AL.
10.1029/2020GL091707 6 of 11 fractures and observation of deformed bedding within these depth intervals, as noted from LWD resistivity borehole imaging .

Stress Orientation Variability Across the Northern HSM
A number of geological and tectonic phenomena, either in isolation or in concert, could explain the observed S Hmax orientation variability across the northern HSM, from the trench through to the onshore MCNAMARA ET AL.
10.1029/2020GL091707 7 of 11 forearc. The E-W S Hmax in the Pāpaku fault footwall (Site U1518), is consistent with the location of this site in a region characterized by horizontal compression subparallel to relative plate convergence. An E-W S Hmax orientation is further supported by the N-S strike of thrust splay faults imaged in seismic and bathymetric data from this area (Barker et al., 2018;Barnes et al., 2020), and E-W shortening directions inferred from sediment magnetic susceptibility anisotropy measured from the Pāpaku Fault footwall (Greve et al., 2020).
NNW-SSE orientations reported from Hole U1519A are not aligned with NE-SW S Hmax (margin parallel) orientations reported from onshore wells and derived from focal mechanisms, nor with E-W S Hmax orientations reported from Hole U1518B (Figure 3). Several factors may explain this discrepancy. First, Hole U1519A is located ∼5 km above a zone of the plate interface that experiences regular large SSEs, and is locked in-between these events (Wallace et al., 2010 (Figure 1c). Hole U1519A was drilled and logged during a time in-between SSEs, when the subduction plate interface was locked, and elastic strain accumulation was occurring, similar to past inter-SSE periods (Wallace et al., 2010). As stress orientations from existing onshore industry wells are located above the steadily creeping plate interface region (with no locking and minimal slow slip; Figures 1 and 3), the observed difference in S Hmax orientations may be related to differences in plate interface behavior (creeping beneath the onshore region; locking and episodic SSEs offshore). This assumes that stress states at the plate interface remain broadly consistent from the plate interface depth to shallow crustal levels. We also note that the broadly NE-SW orientation of the onshore S Hmax observations may also be related to the position of the northern Hikurangi forearc to the east of an actively extending intra-arc rift (the TVZ). Here, the forearc is likely to be under margin-normal (southeast-directed) extension, due to the transmission of slab rollback forces across the forearc and into the extending arc region .
The observed NNW-SSE S Hmax orientation at Hole U1519A may also result from the influence of a subducting seamount which has been inferred between Sites U1519 and U1518 (Barker et al., 2018). It has been inferred that observed fast azimuth directions around this seamount are likely related to effect the subducting seamount has on local structure and stress patterns (Arai et al., 2020;Zal et al., 2020). Numerical modeling suggests that enhanced lateral compression (S Hmax magnitude), rotation of S Hmax , and increased fault-normal stresses can be expected ahead of the landward flank of the subducting seamount, while creating an extensional stress shadow behind it (Sun et al., 2020). Given that Hole U1519A is located on the landward side of the proposed subducting seamount, and in the context of the approximate seamount morphology that is elongate parallel to the margin (NE-SW), this may explain the observed NNW-SSE compressional direction of S Hmax here ( Figure 3). This theory, however, would require a morphological feature at the plate interface to generate a perturbation large enough to affect the stress field at shallow levels where borehole breakouts are being measured. Furthermore, this effect would have to only apply to the leading flank and not the extensional stress shadow, in order to explain the compressional S Hmax orientation at Site U1518 (which is located seaward of the seamount, in the expected stress shadow).
The borehole breakout-derived S Hmax orientations recorded from Hole U1519A may represent a transient stress state associated with the current interseismic SSE period, which may change during SSE. If so, stress states during SSEs in this region could either rotate the S Hmax orientation, change the stress state from compressional to strike-slip or extension as a result of the small, typically several to tens of kPa (Ikari et al., 2020), or allow a combination of both. Analysis of fracture orientations from GVR image logging at Site U1519 (Data Set 2) shows a dominant NNE-SSW strike consistent with the interseismic NNW-SSE compressional S Hmax direction, though multiple strike orientations seen for subordinate fractures may be more consistent with alternative stress states to the one observed at the time of drilling. These may reflect temporal rotations in S Hmax orientation. If SSE stress drops reduce S Hmax below the S V magnitude, becoming σ 2 to create an extensional stress state, then from calculated total S v magnitudes of the region (Data Set 1), the interseismic S Hmax magnitude would fall in the region of ∼19-28 MPa. Until stress orientations can be quantified during SSE, further fracture orientation analysis is performed, and/or quantification of the interseismic S Hmax magnitude from existing data is carried out, this idea remains conjectural, though worthy of further investigation.
Overall, offshore borehole breakout-derived S Hmax orientations from IODP wells (NNW-SSE, and E-W) differ from the documented NE-SW S Hmax for the northern HSM from shallow depths to plate interface (Lawrence, 2018;Townend et al., 2012). This likely reflects a transition from a thrust fault regime in margin-normal compression at the northern HSM deformation front, to extensional (normal-faulting) and strike-slip tectonics in the onshore forearc of the northern HSM forearc, adjacent to the actively extending Taupo Volcanic Zone. We suggest that the offshore IODP stress orientations presented here are influenced by processes occurring near the deformation front, including topographical effects, possible temporal effects associated with the SSE cycle, and spatial effects associated with the presence of a subducting seamount in the region.

Conclusions
Changes in horizontal stress orientations are observed in a transect across the northern HSM. Borehole image log S Hmax orientations imply margin-normal (E-W), maximum compression in the Pāpaku thrust fault footwall, a compressional NNE-SSW orientation in the forearc offshore Tuaheni Basin, and a NE-SW orientation consistent with strike-slip faulting further landward, within the onshore forearc. The NNW-SSE S Hmax observed ∼40 km east of the coast at Site U1519 may reflect a number of offshore subduction-related processes including temporal variations in subduction interface locking and elastic strain accumulation associated with the SSE cycle, and/or spatial controls associated with subducting seamounts in this region. Well-scale variations in stress orientation are likely caused by a combination of topographic effects and recent activity on active fractures and faults.
Data Availability Statement and Employment's Endeavor fund (contract CO5X1605 to GNS Science). A. Fagereng acknowledges funding from ERC Horizon 2020 Starting Grant 715836. K. E. Petronotis and L. J. LeVay received funding from the IODP JRSO (NSF Grant 1326927). This work was carried out in part using the Techlog Wellbore Software provided for research purposes by Schlumberger Limited.