Spatial Variation of Shallow Stress Orientation Along the Hikurangi Subduction Margin: Insights From In‐Situ Borehole Image Logging

Knowledge of the contemporary in‐situ stress orientations in the Earth's crust can improve our understanding of active crustal deformation, geodynamic processes, and seismicity in tectonically active regions such as the Hikurangi Subduction Margin (HSM), New Zealand. The HSM subduction interface is characterized by varying slip behavior along strike, which may be a manifestation of combined variations in both stress state and the mechanical properties of faults and their hanging walls. Alternatively, these variations in subduction thrust slip behavior may drive heterogeneity in the stress state in space and time. In this study, we analyze borehole image and oriented four‐arm caliper logs acquired from 13 boreholes along the HSM to present a comprehensive stress orientation data set from borehole data within the shallow (<3 km) upper plate of the subduction thrust. Our results reveal a 065°/245° SHmax orientation within the central HSM (Hawke's Bay region) which rotates to 112°/292° and 140°/320° in the southern HSM. This rotation of SHmax orientation correlates spatially with along‐strike variations in subduction interface slip behavior, characterized by creep and/or shallow episodic slip events in the central HSM and interseismic locking in the south. The borehole SHmax orientations suggest that contemporary stress orientations may be caused in part by along‐strike variation in deformation style imposed by clockwise rotation of forearc. In the southern HSM, borehole‐derived SHmax orientations are inconsistent with SHmax orientations derived from focal mechanism solutions in the subducting plate, implying some degree of mechanical decoupling between the shallow hanging wall and subducting slab.

Analysis of earthquake focal mechanism solutions reveals that the maximum horizontal stress (S Hmax ) orientations (≤60 km depth) are NE-SW in the northern and central HSM and ENE-WSW orientation in the southern HSM ( Figure 1c). These stress indicators are largely from earthquakes within the subducting slab, with most located at depths >25 km (Townend et al., 2012). In contrast, seismic anisotropy fast orientations (which are often assumed to be parallel to S Hmax ) determined from shear wave splitting methods that sample the upper ∼40 km (Figure 1c) suggest a dominant S Hmax orientation of NE-SW for most of the HSM forearc, while the northern HSM forearc displays variable fast orientations, with a more dominant ENE-WSW inferred S Hmax (Illsley-Kemp et al., 2019).
Shallow (<3 km) S Hmax orientations have been determined from limited analysis of borehole image logs from boreholes drilled onshore and offshore along the HSM (Griffin, 2019;Griffin et al., 2021;Heidbach et al., 2018;Lawrence, 2018;McNamara et al., 2021). Analysis of borehole image data from four onshore boreholes show NE-SW to ENE-WSW S Hmax orientations in the central HSM (Heidbach et al., 2018;Lawrence, 2018), and an E-W to NW-SE S Hmax orientation is determined from three borehole image logs in the southern HSM (Griffin, 2019; Griffin et al., 2021;Heidbach et al., 2018). Boreholes offshore the northern HSM drilled as part of the International Ocean Discovery Program (IODP) Expeditions 372 and 375 show an E-W S Hmax orientation close to the Hikurangi trench, and an NW-SE S Hmax orientation in the offshore forearc (McNamara et al., 2021), indicating strong variations in stress orientations across the forearc.
In this study, we provide a detailed analysis of shallow (<3 km) S Hmax orientations from stress-induced borehole failures, and assess their variability within the upper plate of the HSM. We analyze six borehole image logs and oriented four-arm caliper logs (not previously used for stress orientation studies), and provide a reanalysis of the seven borehole image logs investigated in previous studies (Griffin, 2019;Griffin et al., 2021;Heidbach et al., 2018;Lawrence, 2018), with a focus on acquiring higher resolution measurements (length, width, orientation) of induced borehole failures. We then discuss spatial variations in contemporary S Hmax orientations and their relationship to far-field stresses and long-term patterns of tectonic deformation, and their potential links to along-strike variations in subduction megathrust slip behavior.

Geological Setting
The Hikurangi Subduction Margin (HSM) lies along the Pacific-Australian plate boundary at the southern end of the Tonga-Kermadec-Hikurangi Trench, off the east coast of the North Island, New Zealand ( Figure 1a). The Hikurangi Subduction Margin accommodates westward subduction of the Hikurangi Plateau (a Cretaceous large igneous province) beneath the continental crust of North Island at the Hikurangi Trough (B. W. Davy, 1992). The Hikurangi Plateau is ∼10-15 km thick and transitions to a more typical 5-7 km thick oceanic plate further north at the Kermadec Trench (B. Davy et al., 2008;B. W. Davy, 1992;Ghisetti et al., 2016;Mochizuki et al., 2019). The southern termination of the HSM is located somewhere beneath New Zealand's northeastern South Island, where oblique convergence is transferred to the Marlborough Fault System and Alpine Fault (Barnes et al., 1998;Little & Roberts, 1997).
Neogene to present tectonic deformation across the HSM is complex and includes contributions from shortening associated with subduction at the Hikurangi Trough, clockwise rotation of the East Coast forearc, strike-slip faulting along the North Island Dextral Fault Belt (NIDFB), and back-arc extension in the Taupo Volcanic Zone (TVZ) (Beanland & Haines, 1998;Wallace et al., 2004, Figure 1a). The East Coast forearc has rotated clockwise for at least the last few Myr at the rate of 3°-4°/Myr relative to the Australian plate (Nicol et al., 2007). This rotation results in back-arc rifting in the central North Island's Taupo Volcanic Zone (TVZ), transpression in the southern North Island, and creates a large along-strike change in convergence rate at the Hikurangi 10.1029/2021JB023641 4 of 19 Trough, from ∼20 mm/year in the south to ∼60 mm/year at the northern Hikurangi Trough (Wallace et al., 2004, Figure 1a). Wallace et al. (2004) suggest that an along-strike change from subduction of the large igneous province (Hikurangi Plateau) at the Hikurangi Trough to normal oceanic crust along the Kermadec Trench exerts a torque on the forearc, producing clockwise rotation of the eastern North Island. Overall, relative motion between the Pacific and Australian plates occurs through this region at ∼40 mm/yr, and is oblique to the orientation of the plate boundary. The oblique relative motion is partitioned into a margin-perpendicular component and a margin-parallel component. The margin-perpendicular component occurs along the Hikurangi subduction interface and active upper-plate thrust faults within the accretionary wedge and overriding plate (Barnes et al., 1998;Nicol & Beavan, 2003). The margin-parallel component of Pacific-Australia relative plate motion is largely accommodated by a combination of right-lateral strike-slip in the North Island Dextral Fault Belt (NIDFB) and vertical-axis clockwise rotation of the North Island forearc (Beanland & Haines, 1998;Nicol et al., 2007;Wallace et al., 2004).

Data and Methodology
We analyze borehole image logs acquired from 11 boreholes using a range of tools including; the Schlumberger Fullbore Formation Microimager (FMI™; Figure 2a) and Oil-Based Mud Imaging tool (OBMI™), Baker Atlas Simultaneous Acoustic and Resistivity Imager (STAR™), Tiger Energy Services Acoustic Formation Imaging Technology (AFIT; Figure 2b), and two orientated four-arm caliper logs (Figure 2c). The tool types and their borehole wall coverage for each borehole are summarized in Table S1 in Supporting Information S1. Data processing and quality assessment are performed on all image logs, the details of which are documented in the supplementary material (Text S1 and S2 in Supporting Information S1).
From the borehole image logs, we identify stress-induced borehole failures, including borehole breakouts (BOs; Figures 2a and 2c), drilling-induced tensile fractures (DITFs), and petal-centerline fractures (PCFs; Figure 2b). BOs and DITFs are well-known indicators of horizontal in-situ stress orientations in vertical to semi-vertical boreholes, under the common assumption that one of the principal stresses is vertical stress (S v ). BOs and DITFs develop parallel to the contemporary minimum (S hmin ) and maximum (S Hmax ) horizontal stresses, respectively (Aadnoy & Bell, 1998;Bell, 2003;Bell & Gough, 1979). BOs and DITFs can also be used to determine S Hmax orientations in boreholes deviated ≥20°, as long as corrections are applied to address the impact of vertical stress (S v ) on their development (Peška & Zoback, 1995).
BOs form as enlargements of the borehole diameter on opposite sides of the borehole wall where the circumferential hoop stress, induced by non-uniform horizontal principal stress magnitudes, is large enough to exceed the rock strength (Bell & Gough, 1979;Zoback, 2007). Borehole breakouts typically appear on resistivity image logs as a pair of wide, out-of-focus zones, conductive (in water-based mud; Figure 2a) or resistive (in oil-based mud, such as OBMI tool; King et al., 2010) zones. In acoustic televiewer logs, they appear as low amplitude and long travel time zones. In both types of logs, BOs are located ∼180° from each other around the circumference of the borehole wall. BOs often correlate with borehole enlargement and are associated with large caliper values as the result of the borehole failure ( Figure 2c; Tingay et al., 2016). In this study we use these criteria to identify BOs from available image logs, only identifying features as BOs if there are two present at the same depth interval and they are oriented 180° ± 10° from each other around the borehole circumference.
Oriented four-arm caliper data are also used to infer the presence of BOs along boreholes. To reliably distinguish BOs from other non-stress related enlargements, such as keyseats and washouts, we apply the criteria presented by Reinecker et al. (2016). They recommend that all four-arm caliper data, borehole deviation and azimuth, and pad 1 azimuth must be controlled carefully to interpret borehole breakouts. The following criteria are considered: (a) one pair of caliper arms reads very close to the bit size while the opposite pair measures a larger diameter; (b) caliper differences must exceed the bit size by 10%; (c) rotation of the tool ceases in breakout zones; and (d) the enlargement zone must extend more than one meter vertically.
DITFs develop on the borehole wall where there is a significant difference between the two horizontal principal stress magnitudes and the local stress concentrations around the borehole wall lead to hoop stresses that overcome the tensile strength of the rock (C. Barton et al., 2009;Brudy & Zoback, 1999;Zoback, 2007). DITFs typically appear as narrow, conductive (on resistivity image logs) or low amplitude and longer travel time (on acoustic image logs) pairs, ∼180° from each other around the circumference of the borehole wall. DITFs are generally parallel or slightly inclined to the borehole axis in vertical to semi-vertical boreholes (C. A. Barton et al., 1998;Bell, 2003;Rajabi et al., 2016aRajabi et al., , 2016bTingay et al., 2016). In this study, DITF selection criteria follows these considerations and only accepts pairs of DITFs located within the same depth interval that are 180° ± 10° from each other around the borehole circumference.
Here, all BOs and DITFs are reported as individual feature lengths and widths, such that a single BO or DITF measurement does not span a number of separate individual BOs or DITFs, similar to what has been done in previously analyzed image logs along the HSM (Griffin, 2019;Lawrence, 2018). This is an important aspect of quantifying induced features from borehole image logs because geological properties, such as varying strength associated with variably bedded lithologies, impact the development and growth (both width and length) of borehole breakouts (Fellgett et al., 2019;Kingdon et al., 2016). It is also important to capture each induced feature individually for accurate statistical considerations of borehole stress orientations.
PCFs are induced fractures that form within the bedrock ahead of the drill bit in response to stress concentrations at the bottom of the borehole during drilling and propagate inward toward the borehole (Davatzes & Hickman, 2010;Y. Li & Schmitt, 1998;Wenning et al., 2017). PCFs appear as conductive (resistivity image logs) or low amplitude (acoustic image logs) partial sinusoids that merge into discontinuous borehole axial centerline fractures (Figure 2b, Kulander et al., 1990). The average of the centerline fracture orientations or dip orientation of the partial sinusoids of a PCF is parallel to the orientation of S hmin (Davatzes & Hickman, 2010). In contrast to the DITFs, the centerline portions of PCFs are often less than 180° from each other around the circumference of the borehole wall. We identify PCFs when the entire structure (petal fracture and both centerline fractures) are visible on the image log, and where the centerline fractures do not fall into the criteria used for identifying DITFs (180° ± 10°).
The quantitative World Stress Map (WSM) quality ranking system (where A-quality data is a reliable indicator of regional stress orientations and E-quality is more likely representative of localized stress orientation) and circular statistical analysis is used here to provide quantified uncertainty for all reported stress orientations from borehole image and oriented four-arm caliper logs (e.g., Heidbach et al., 2016). A detailed explanation of these calculations can be found in the supplementary material (Text S3 in Supporting Information S1). Borehole locations, image log intervals, mean S Hmax orientation, standard deviation, and WSM quality classification based on the length-weighted method (Heidbach et al., 2016) for individual boreholes are summarized in Table 1 and Table 2.

Spatial Variation of Shallow S Hmax Orientations
Our results show that contemporary S Hmax orientations in the HSM upper plate change from 065°/245° ± 10° (ENE-WSW) along the central HSM to 112°/292° ± 20° (WNW-ESE) in the southern HSM (with a local S Hmax orientation of 140°/320° ± 22° (NW-SE) from D-quality data) ( Figure 5). The 065°/245° ± 10° borehole-derived S Hmax orientation in the central HSM is rotated modestly (∼17°) anticlockwise with respect to the Pacific-Australian plate motion (82°; Beavan et al., 2002). This implies that far-field plate boundary forces exerted at the HSM may be the primary control on the contemporary stress orientations within the upper plate in this region. In contrast, the 112°/292° ± 20° borehole-derived S Hmax orientation in the southern HSM is rotated ∼30° clockwise (bearing ∼58° rotation from localized D-quality 140°/320° ± 22° S Hmax orientations) relative to the Pacific-Australian plate motion ( Figure 5). This observed 47°-75° along-strike rotation in S Hmax orientation may result from a number of possible influences, including (a) along-strike variations in subduction interface coupling behavior, (b) HSM forearc kinematics and long-term tectonic deformation, (c) lateral variations in basement topography along strike, or (d) regional changes in recent slips on local active faults, surface topography, and geomechanical stratigraphy along strike.
The spatial pattern of rotation in borehole-derived S Hmax orientations along the HSM is consistent with the location of changes in subduction interface coupling from north to south (Figures 1b and 5). In the northern and central HSM, the subduction interface is largely creeping and experiences shallow (<15 km), episodic slow slip events that extend offshore and possibly to the trench. At the southern HSM the plate interface is strongly interseismically coupled to ∼30 km depth, and is currently accumulating elastic strain in the surrounding crust (e.g., Wallace, 2020, Figure 1b). Accrual of elastic compressional strain and stress resulted from interseismic coupling in the southern HSM could impose a NW-SE component (parallel to the convergence direction between the Hikurangi forearc and subducting plate) to the approximately E-W Pacific-Australian plate motion. This may result in the observed WNW-ESE S Hmax orientation (112°/292°-140°/320°) in the shallow upper plate.
The HSM experiences a dominantly compressional tectonic regime due to subduction; however rapid clockwise rotation of the forearc, which accommodates the margin-parallel component of oblique Pacific-Australian plate motion, results in significant tectonic transitions along strike (Nicol et al., 2007;Wallace et al., 2004). Deformation resulting from clockwise rotation of the Hikurangi forearc drives strike-slip and/or normal faulting within the onshore portion of the northern and central HSM, and transpressional faulting in the southern HSM ( to explain S Hmax orientation variations in the Nankai subduction zone (Chang et al., 2010;Lin et al., 2010;Tobin et al., 2009;Wu et al., 2012) and the Costa Rica margin (Malinverno et al., 2016).
Basement topographic variation, such as uplifted basement blocks or seamounts, can introduce gravity and density changes within the crust creating geomechanical inhomogeneities (e.g., Bassett et al., 2022;Chow et al., 2022). Such geomechanical inhomogeneities can add horizontal compressional stress around the uplifted basement margins and extensional stress above them, changing the stress orientations (Artyushkov, 1973;Bott, 1991;   Enever et al., 1999;Gale et al., 1984;Rajabi et al., 2016a;Ruh et al., 2016;Sun et al., 2020). This may partly produce our observed S Hmax rotation. For example, the SEEBASE™ depth to basement model of New Zealand (Figure 6a; FROGTECH, 2013) shows that the central HSM basement is deeper (∼9-15 km) than the southern HSM basement (<5 km) (Figure 6a). Most southern HSM borehole-derived S Hmax orientation measurements are located on basement topographic highs while the central HSM measurements are located in areas where the basement is deep. Furthermore, localized basement highs can be observed within the central HSM (Figure 6a). Two of the central HSM borehole-derived S Hmax measurements (boreholes 4 and 5 in Figure 6a) are located close to the margins of these basement highs and as such their orientation may be influenced by them. However, as these S Hmax orientations align with other measurements made away from these basement topographic features, we consider this to be unlikely.
Linked to the central to southern change in basement topography are surface topography (Figure 6b) and geomechanical heterogeneities introduced by variations in rock and sediment physical properties, both of which can influence S Hmax orientations. The variable basement topography means surface topography is more pronounced in the southern HSM than the central HSM (Figure 6b). This may influence S Hmax orientations in shallow depth intervals (<500 m below sea level) such as boreholes Te Mai-2 and Orui-1A (boreholes 8 and 10 in Figure 6b). However, given that the borehole breakouts used to measure S Hmax orientations from this region are mostly located at depths 600-2700 m below sea level, we infer that surface topography is not a primary factor controlling the observed S Hmax rotation. Finally, due to basement uplift in the southern HSM, borehole breakouts are measured in a wider variety of lithologies (Miocene to Cretaceous sediments) than in the northern HSM (dominantly Miocene sediments). If the wider ranges of lithologies in the southern HSM are geomechanically different from those in the central HSM, this lateral heterogeneity in mechanical rock properties could in part be responsible for the S Hmax rotation.
Recent slips on active faults are also known to reorient stress, creating variations in S Hmax orientations both along the region and with depth (e.g., D. Li et al., 2018;Provost & Houston, 2001;Rice, 1992). However, it cannot feasibly explain the large-scale S Hmax orientation rotation observed along the HSM strike, beyond what has already been discussed with respect to deformation style changing from strike-slip/normal faulting in the central HSM, to the dominantly reverse and transpressional faulting in the southern HSM. However, recent faulting may explain some of the localized S Hmax rotations observed within the central and southern HSM regions. For instance, at borehole Whakatu-1 (boreholes 6 in Figure 5) a 144°/324° ± 11° (NW-SE) S Hmax orientation is noted, which is rotated ∼ 80° from the dominant 065°/245° ± 10° S Hmax orientation in the central HSM ( Figure 5). The NW-SE S Hmax orientation in Whakatu-1 is perpendicular to the HSM margin and to observed active, NNE-SSW striking reverse faults in this region ( Figure 5, Hull, 1987;Langridge et al., 2016;Litchfield et al., 2014). Similarly, Griffin (2019) suggests the activity of nearby faults as the cause of the observed S Hmax rotation along depth in borehole Ranui-2.

Shallow S Hmax Orientations and Maximum Contraction Directions
Geodetic measurements over the last 25 years have been used to determine New Zealand's contemporary surface strain field (Dimitrova et al., 2016;Haines & Wallace, 2020). Maximum contraction directions change from dominantly E-W in the northern central HSM (Hawke's Bay region), to a mix of both NE-SW and WNW-ESE in the southern central HSM, to WNW-ESE and NW-SE directions in the southern HSM (Figure 7a, Haines & Wallace, 2020). The majority of S Hmax orientations are within 40° of GPS maximum contraction directions (Figure 7b), indicating that the two datasets are broadly compatible, although there is one outlier with a misalignment of ∼57° (borehole 6 in Figure 7b). The broad agreement between shallow S Hmax orientations and geodetic maximum contraction directions may suggest that stress orientations in the upper plate are influenced by elastic strain arising from interseismic coupling on the Hikurangi subduction interface.
The observed variable misalignment of maximum contraction directions and S Hmax orientations (Figure 7b) may reflect variations in the shallow S Hmax orientations (as explained in 5.1) or temporal changes in stress orientation at the HSM. This is unlikely for the HSM as the borehole data and GPS measurements were collected within the same time period, 1994-2013 and 1995-2013, respectively. Similar widespread stress-strain direction misalignment is also noted at the Nankai and Cascadia subduction zones (e.g., Townend & Zoback, 2006;Wang, 2000;Wang et al., 1995). Wang (2000) explains the misalignment of maximum contraction directions with respect to S Hmax as a result of geodetic strain signals reflecting temporal changes in the stress state associated with the subduction earthquake cycle. Townend and Zoback (2006) explained this mismatch in SW Japan by showing that the stress orientations determined from focal mechanisms within the upper plate are more strongly influenced by long-term tectonic processes (convergence between northeastern Honshu and Amuria plate), while the geodetic strain orientations are dominated by interseismic strain accumulation between large earthquakes. This could certainly be the case at HSM, although observed differences between stress orientations and maximum contraction directions are not as dissimilar as those reported for the SW Japan case, making it difficult to disentangle the influence of long-term plate motion versus earthquake cycle processes on the HSM upper plate stress field. Townend et al. (2012) derived S Hmax orientations from earthquake focal mechanisms (≤60 km depth) between 2004 and 2011, and observed an average S Hmax orientation of 060°/240° ± 17° for the central HSM, and 066°/246° ± 22° for the southern HSM (Figure 8a; calculation of average S Hmax orientation is explained in supplementary material (Text S3 in Supporting Information S1)). Comparing focal mechanism-derived to borehole-derived S Hmax orientations shows that in the central HSM they are closely aligned (borehole: 065°/245° ± 10°; focal mechanism: 60°/240° ± 17°). In contrast, the southern HSM shows an apparent stress field rotation of 46°-74° with depth (borehole S Hmax : 112°/292° ± 20°-140°/320° ± 22°; focal mechanism S Hmax 66°/246° ± 22°) (Figure 8b). Only one compressional focal mechanism S Hmax orientation (149°/329° ± 33°) in the southern HSM (latitude 40.5°S), located at ∼25 km depth (near the subduction interface), aligns with the borehole-derived S Hmax orientations for the southern HSM (112°/292° ± 20°-140°/320° ± 22°) (Figures 8a and 8b).

Variation of S Hmax Orientations With Depth
Stress rotations with depth can be generated by a number of mechanisms, including (a) topography, (b) shallow fault activity, and (c) mechanical anisotropy or heterogeneity. Topography is ruled out as the majority of borehole breakouts come from depths deeper than the amplitude of local topography. Active faulting in the shallow crust has been used to explain S Hmax orientation rotation with depth at the borehole scale in the southern HSM  (Ranui-2; Griffin, 2019). Faults in the region are both reverse and strike-slip, and predominantly striking NE-SW to ENE-WSW (Langridge et al., 2016;Litchfield et al., 2014). Slip on these shallow reverse faults may rotate the S Hmax into the observed NW-SE orientations recorded from borehole data here, while the overall subduction S Hmax at depth remains in a NE-SW orientation driven by relative plate motion. High fluid pressure can also encourage slip on deeper faults, although in the southern HSM, magnetotelluric data (Heise et al., 2019), seismic tomography studies (Bassett et al., 2014;Eberhart-Phillips et al., 2017), and borehole analysis (except for offshore boreholes Titiahoa-1 and Tawatawa-1; Burgreen-Chan et al., 2016;Darby & Funnell, 2001) do not provide evidence for elevated fluid content or pressure.
Considering borehole-derived WNW-ESE/NW-SE S Hmax orientations and focal-mechanism derived NW-SE S Hmax orientation, located at ∼25 km depth, the rotation in the S Hmax orientation may occur across the plate interface. In this case, the hanging wall would be mechanically decoupled from the subducting plate, generating the observed rotation in S Hmax above (from borehole data) and below it (from focal mechanism data). Such decoupling may imply a low shear stress on the plate interface. Mechanical decoupling has been suggested to occur on subduction plate interfaces elsewhere, though these findings often show principal stresses changing from vertical to horizontal, not a rotation in the horizontal stress orientation as observed here (T. Byrne & Fisher, 1990;Flemings & Saffer, 2018). Further information on the likely strength of the plate interface is required to investigate this theory. Linking the idea of weak faults creating geomechanical heterogeneities and the activity of shallow faults together may provide a further explanation for the shallow S Hmax in the southern HSM.
More recently, Illsley-Kemp et al. (2019) measured seismic anisotropy using shear wave splitting fast orientations from earthquake data based on the assumption that they align to S Hmax orientation if other significant crustal anisotropies are not present (e.g., fracturing, faulting, grain, and crustal preferred orientations) (Araragi et al., 2015;Boness & Zoback, 2006;Johnson et al., 2011;Pastori et al., 2019;Yang et al., 2011;Zinke & Zoback, 2000). Illsley-Kemp et al. (2019) report a dominant NE-SW (030°-060°) S Hmax orientation for the Hikurangi forearc, with some areas showing a more ENE-WSW orientation (060°-090°), such as the Hawke's Bay area. In comparison to borehole-derived S Hmax orientations presented here (Figure 8a), the two datasets agree in the central HSM shallow crust but not in the southern HSM, with a NE-SW shear-wave splitting and a borehole-derived S Hmax of NW-SE. Illsley-Kemp et al. (2019) report that their data is generated from a depth range of 0-40 km which at the Hikurangi forearc captures the upper plate, the plate interface, and the subducting plate. The discrepancy between the two S Hmax orientations is thus likely due to (a) the shear wave splitting-derived data are sampling the subducting plate stress field, which seems likely as those directions broadly match focal-mechanisms subducting plate S Hmax orientation (Figure 8a), (b) the borehole-derived S Hmax orientations are only consistent for the upper few km of the upper plate crust, or (c) the fast directions are controlled by NE-SW striking faulting in the region, which is suggested to be the case by Illsley-Kemp et al. (2019) or other crustal anisotropies.

Conclusions
This paper presents a comprehensive analysis of contemporary S Hmax orientations from borehole data along the HSM, and discusses stress field orientation variability within the context of variable tectonics and slip behavior of this subduction margin. S Hmax orientations in the central HSM are predominately 065°/245°, which rotates to a dominantly 112°/292° ± 20° S Hmax orientation in the southern HSM. Our borehole-derived S Hmax orientations vary along the HSM suggesting that the observed stress orientations are likely caused by along-strike variation in interseismic coupling behavior of the Hikurangi subduction interface. The long-term tectonic deformation arising from rapid rotation of the Hikurangi forearc, causing reverse faulting and strike-slip in the southern part of the margin and a combination of extension and strike-slip in the northern and central margin, may in part cause the along-strike variations in observed stress orientations. The basement uplifts may also be at play in influencing the along-strike variations in observed stress orientations by imposing a horizontal compressional stress component near the uplifted margin, introducing surface topography and geomechanical heterogeneities. However, the extent to which basement topography may change S Hmax orientation along the HSM requires further data and modeling. S Hmax orientations measured from boreholes are mostly within 40° of maximum contraction directions, suggesting that the observed stress orientations could be influenced by elastic strain arising from interseismic coupling on the subduction interface. In the southern HSM, borehole-derived S Hmax orientations are 46°-74° to S Hmax orientations derived from focal mechanism solutions in the subducting plate, implying that the shallow upper plate is mechanically decoupled from the slab, potentially reflecting low shear strength along the plate interface. Further interpretation of HSM stress state could be achieved by constraining stress magnitudes, which will be the focus of future studies.

Data Availability Statement
This research used data provided by the New Zealand Petroleum and Minerals group (NZPM) within the Ministry for Business, Innovation and Employment (MBIE). The borehole image logs and oriented four-arm caliper data used in this paper can accessed through MBIE's online free database (https://data.nzpam.govt.nz/GOLD/system/ mainframe.asp). All additional data used in this article were collected from published sources referenced in the text. Borehole breakout measurements presented in this study can be accessed at https://github.com/Behboud-iEffatGeo/StressCharacterization_HSM.git and https://doi.org/10.5281/zenodo.6402741 (Behboudi et al., 2022).