Fault inversion can accommodate ground deformation above inflating igneous sills

Magma emplacement is commonly accommodated by uplift of the overburden and free surface. By assuming this deformation is purely elastic, we can invert the shape and kinematics of ground deformation to model the geometry and dynamics of 15 underlying intrusions. However, magma emplacement can be accommodated by viscoelastic and/or inelastic processes. We use 3D seismic reflection data to reconstruct how elastic bending and inelastic processes accommodated emplacement of a Late Jurassic sill offshore NW Australia. We restore syn-emplacement ground deformation and compare its relief to sill thickness, showing that: (i) where they are equal, elastic bending accommodated intrusion; but (ii) where sill thickness is greater, inversion of a pre-existing fault and overburden compaction contributed to magma accommodation. Our results 20 support work showing inelastic processes can suppress ground deformation, and demonstrate magmatism can modify fault displacements. Reflection seismology is thus powerful tool for unravelling links between magma emplacement, ground deformation, and faulting.

Although we recognise the need to better understand how magma emplacement translates into ground deformation [e.g., Ebmeier et al., 2018;Magee et al., 2018], most studies examine how intrusions create new structures (e.g., folds and 15 faults) or modify host rock properties (e.g., porosity) [e.g., Hansen and Cartwright, 2006;Magee et al., 2019a;Montanari et al., 2017;Morgan et al., 2008;Reeves et al., 2018]. Here, we use 3D seismic reflection data from offshore NW Australia to investigate how pre-existing faults may affect the mechanics of roof uplift above an intruding sill. With these data we also explore how intrusion-induced deformation may modify fault displacement patterns, cautioning the way we use such patterns to unravel fault kinematics in areas of co-located faulting and magmatism [e.g., Nicol et al., 20 1995;Nicol et al., 1996;Rotevatn et al., 2019;Walsh and Watterson, 1988]. We show that emplacement of the sill in the hanging wall of a major tectonic fault, at a depth of ~0.9 km during the Late Jurassic, was primarily accommodated by roof uplift facilitated by both local elastic bending and fault inversion. Local discrepancies between sill thickness and fold amplitude suggest other inelastic processes (e.g., porosity reduction) may have helped to generate space for the intruding magma. Fault inversion locally reduced throw across the fault, which if not recognised as being intrusion-25 induced may be incorrectly interpreted as evidence of linkage between initially isolated fault segments [cf. Cartwright et al., 1995;Peacock and Sanderson, 1991]. Overall, our work demonstrates that seismic reflection data is a powerful tool for unravelling how intruding magma is expressed at the surface and interacts with faults.   Hocking et al., 1987;Longley et al., 2002;Magee and Jackson, 2020;Tindale et al., 1998]. (C) Uninterpreted and interpreted 2D seismic line across the Exmouth Plateau and 5 Exmouth Sub-basin. See Figure 1A for location.
The earliest phase of rifting on the Exmouth Plateau initiated in the Rhaetian (Late Triassic) and likely ceased towards the end of the Callovian (Middle Jurassic; Fig. 1B) [e.g., Bilal et al., 2018;Black et al., 2017;Gartrell et al., 2016;Tindale et al., 1998]. Late Triassic-to-Jurassic rifting produced an extensive array of ~N-S striking, high-throw (up to 10 ~1 km) normal faults, which offset a thick pre-rift succession primarily consisting of fluvio-deltaic sedimentary rocks (i.e. the Mungaroo Formation; Figs 1B, C, and 2A) [e.g., Bilal et al., 2018;Black et al., 2017;Marshall and Lang, 2013;Stagg et al., 2004]. During the Early Jurassic, the Exmouth Plateau was sediment starved in comparison to the sub-basins located further east, resulting in deposition of a relatively condensed (≲100 m thick) latest Triassic-to-Early Jurassic, syn-rift succession (e.g., Fig. 1C) [e.g., Exon et al., 1992;Karner and Driscoll, 1999]. This syn-rift succession comprises 15 the siliciclastic Brigadier and North Rankin formations, as well as the Murat Siltstone and Athol Formation, and records a transgression from shallow-to deeper-marine conditions (Fig. 1B) [e.g., Hocking, 1992;Hocking et al., 1987;Stagg et al., 2004;Tindale et al., 1998]. Development of a regional unconformity at the end of the Callovian marked the end of this first rift phase (Fig. 1B) [e.g., Bilal et al., 2018;Yang and Elders, 2016]. The unconformity is overlain by the marine Dingo Claystone (Oxfordian-to-Tithonian; Fig. 1B) [e.g., Tindale et al., 1998]. 20 Crustal extension is broadly considered to have continued throughout the Jurassic across the North Carnarvon Basin [e.g., Gartrell et al., 2016;Tindale et al., 1998], although the apparent cessation of faulting during deposition of the Dingo Claystone on the Exmouth Plateau suggests rifting may have been punctuated by a period of tectonic quiescence ( Fig. 1B) [e.g., Magee et al., 2016]. Development of the Base Cretaceous unconformity at ~148 Ma (latest Tithonian) 5 and subsequent rapid subsidence to accommodate a thick succession of deltaic rocks (i.e. the Tithonian-to-Valanginian Barrow Group), mark the onset of a second rift phase across the Exmouth Plateau (Figs 1B and C) [e.g., Paumard et al., 2018;Reeve et al., 2016]. Tithonian-to-Valanginian rifting involved relatively little upper crustal faulting, with this event producing an array of N-S to NE-SW-striking, low-throw (<0.1 km) normal faults. It is thus likely that stretching during this period was dominated by depth-dependent extension or dynamic topography [e.g., Driscoll and Karner, 1998;Reeve 10 et al., 2016]. Rifting culminated in the development of an continent-ocean transition zone and ultimately continental break-up along the western margin of the Exmouth Plateau in the Valanginian-to-Hauterivian (~135-130 Ma; Fig. 1B) [e.g., Direen et al., 2008;Robb et al., 2005;Stagg et al., 2004]. Following continental break-up in the Early Cretaceous, thermal subsidence controlled margin development, resulting in the development of a thick post-rift succession, parts of which have been deformed by tiers of polygonal faults (e.g., Fig. 1C) [e.g., Paganoni et al., 2019;Velayatham et al., 15 2019].
The North Carnarvon Basin records a complex and protracted history of magmatic activity during the Late Jurassic-to-Early Cretaceous (Fig. 1B). A mafic-to-ultramafic, high-velocity magmatic body was emplaced in the lower crust, possibly during the Middle Jurassic, which may have promoted regional uplift and formation of the Callovian 20 unconformity [e.g., Frey et al., 1998;Rey et al., 2008;Rohrman, 2013;Rohrman, 2015]. Extensive sill-complexes across the North Carnarvon Basin (e.g., Fig. 2A), which dating of intrusion-induced forced folds and vent complexes indicate were emplaced at least during the Kimmeridgian, may have been fed by this high-velocity magmatic body [e.g., Magee et al., 2013a;Magee et al., 2017;Rey et al., 2008;Rohrman, 2013]. A transition from sill-complex emplacement to intrusion of an extensive dyke swarm occurred at ~148 Ma, coincident with formation of the Base 25 Cretaceous unconformity [Magee and Jackson, 2020]. The last and main phase of magmatism across the North Carnarvon Basin resulted in development of the continent-ocean transition zones and associated volcanics during breakup (e.g., seaward-dipping reflectors), as well as sporadic sill intrusions within the basin interior [e.g., Hopper et al., 1992;Magee et al., 2013b;Mark et al., 2020;Rey et al., 2008;Symonds et al., 1998]. Here we use the publicly available, high-quality, time-migrated Glencoe 3D seismic reflection survey that was acquired by CGGVeritas in 2007-2008 (Figs 1A and2A). Data were recorded with a line spacing of 25 m using 10, 6 km long streamers, with 480 channels recording to ~8 s two-way time (TWT) at a sample interval of 2 ms. The seismic source had a volume of 3460 in 3 and was fired at shot point intervals of 12.5 m at a tow depth of 7 m. Full-fold, the dataset 5 covers an area of approximately 4042 km 2 ( Fig. 2A). Seismic data were processed to zero-phase and are here displayed with SEG standard polarity, whereby a downward increase in acoustic impedance corresponds to a peak (red-to-yellow on seismic sections) and a downward decrease in acoustic impedance as a trough (blue or black on seismic sections).
We use data from the Briseis-1, Nimblefoot-1, Warrior-1, and Glencoe-1 boreholes to determine ( No boreholes intersect the igneous intrusion within our study area, but data from the nearby Rimfire-1 and Chester-1 ST1 wells ( Fig. 2A), which intersect a ~10 m thick intrusion and an ~18 m wide dyke respectively, suggests intrusions in the region are likely mafic [Childs et al., 2013;Magee and Jackson, 2020;Moig N and Massie, 2010]. Although there is no velocity information available for the thin intrusions intersected by Rimfire-1 and Chester-1 ST1, we consider the 20 sill we study has a seismic velocity of ~5.55(±0.555) km s -1 ; this range is based on velocity data acquired from mafic intrusions in other sedimentary basins [e.g., Magee et al., 2019a;Skogly, 1998;Smallwood and Maresh, 2002]. At the level of the intrusion in our study area, the dominant frequency of the data is ~25 Hz, which coupled with a seismic velocity of ~5.55(±0.555) km s -1 suggests the limits of separability and visibility for the sill are ~56(±5.6) m and ~7(±0.7) m, respectively. Where the sill has a thickness between these limits of separability and visibility, it is expressed in the 25 data as a tuned reflection package; i.e. seismic energy reflected from the top and base intrusive contacts combines on its return to the surface and cannot be deconvolved [e.g., Eide et al., 2018;Smallwood and Maresh, 2002]. Where the sill is thicker than the limit of separability, its top and base reflections can be distinguished, allowing us to use our inferred velocity range to depth-convert the intrusions measured thickness from seconds TWT to metres.

Seismic Interpretation
To define the geometry of the studied sill, which around its outer edges typically appears as a tuned reflection package, we mapped two seismic horizons (Top and Base Sill). We also mapped eight seismic horizons within the host sedimentary sequence to provide a seismic-stratigraphic framework for our analyses. Biostratigraphic data from the 5 Briseis-1 borehole, which is closest to the study area, provides direct age constraints on five mapped horizons ( well-logs only extend between depths of 2563-3548 m TVD (total vertical depth beneath the drill floor) and thus do not 10 intersect the Top Muderong horizon (Fig. 2C). The location of the Top Muderong was instead constrained by using checkshot data from Briseis-1 to convert the measured depth of the horizon in metres to TWT (Fig. 2B). In addition to the five dated horizons we mapped an intra-Mungaroo Formation horizon encountered in Briseis-1, but for which the exact age remains unconstrained (i.e. Intra-Mungaroo; Fig. 2C). We also mapped two horizons above the Top Triassic, but these could not be dated as they were eroded by the Base Cretaceous unconformity and do not extend to the location 15 of Briseis-1; we term the stratigrapically oldest of these horizons Intra-Jurassic R1, and the other Intra-Jurassic R2 (Fig.   2C). All horizons were mapped across the study area, except for the Intra-Mungaroo horizon, which was only identified locally as sub-sill imaging often hindered its recognition. Thickness maps (isochores) between various combinations of the mapped horizons were used to assess deformation of the sedimentary sequence through time.

Forced fold analysis 20
If the emplacement of a sub-horizontal, tabular magma body (e.g., a sill) is fully accommodated by elastic bending, we may expect the intrusion thickness (ST0) to equal the syn-intrusion amplitude (F0) of the resultant fold ( Fig. 3A) [e.g., Bunger and Cruden, 2011;Galland and Scheibert, 2013;Goulty and Schofield, 2008;Hansen and Cartwright, 2006;Pollard and Johnson, 1973]. However, if inelastic processes (e.g., compaction) also contribute to generating space for intruding magma, and thus suppress uplift, we may expect ST0>F0 (Fig. 3B) [e.g., Magee et al., 25 2013a;Magee et al., 2019b;Schmiedel et al., 2017]. We calculate the present-day, vertical fold amplitude (F) every 10 m along a representative seismic line by measuring and depth-converting the distance between the top fold horizon and a projected pre-fold datum (Fig. 3) [e.g., Hansen and Cartwright, 2006]. We also calculate the present-day, vertical sill thickness (ST) every 10 m along the same profile and compare this to F (Fig. 3). Because burial-related compaction likely reduces fold amplitude through time (i.e. F0>F; Figs 3C and D), without affecting the thickness of typically 30 11 incompressible intrusions (i.e. we assume T=T0), we backstrip and decompact F to estimate F0 (Figs 3C and D) [e.g., Magee et al., 2019a]. Airy backstripping of strata involves restoration of its initial porosity (ø0) by removing the effects of overburden loading [e.g., Sclater and Christie, 1980], and thus requires knowledge of: (i) the current porosity (ø) of a given sedimentary sequence; and (ii) the compaction length scale (λ), which is the inverse of the compaction coefficient and estimates the rate of compaction with increasing burial depth. Given that no boreholes penetrate the entire folded sequence, either at 15 the actual fold or elsewhere in the 3D survey, we could not estimate ø and λ from our data. Instead, following the method outlined by Magee et al. [2019a], we computed an envelope of potential backstripped F0 using a range of realistic ø0 (0.7-0.25) and λ (3.7-1.4) values for claystones and sandstones as these rock types dominate the folded siliciclastic sequence studied (e.g., Fig. 1B).

20
There are several limitations to our method for comparing ST0 and F0. First, we assume that the measured sill thickness (ST) is equal to ST0 (Fig. 3), but acknowledge that post-emplacement magma expulsion and/or contraction during crystallisation could mean T<ST0 [e.g., Caricchi et al., 2014;Chaussard, 2016;Magee et al., 2019a]. The use of incorrect seismic velocities may also introduce imprecision into our depth-conversion of ST and F, although we consider that the range of velocities used for the sill and those for the borehole-constrained folded stratigraphy minimises this error.
However, our method does not account for potential lateral variations in seismic velocity across the sill or forced fold, which could reflect changes in lithology. Finally, we note that the Airy backstripping method applied assumes the folded layers had no flexural strength (i.e. elastic thickness) [Magee et al., 2019a]. Yet if the folded rock layer had an elastic thickness, its fold amplitude would have been suppressed [e.g., Hardy and Finch, 2006;Oehlers et al., 1994]. We note 5 that folding of a layer with a flexural strength would promote lengthening of the fold beyond the limits of the underlying forcing mechanism (i.e. the sill); our assumption that the folded sequence has no flexural strength may thus be partially validated by an observed coincidence between the fold outline (i.e. where fold amplitude is zero) and the sill edge [Hansen and Cartwright, 2006;Magee et al., 2019a]. Overall, whilst these limitations my cause F0 to deviate from ST0, we consider their effects are likely negligible and suggest that a difference between ST0 and F0 of >5% can probably be 10 related to syn-emplacement processes, as opposed to post-emplacement modification of ST0 or simply reflecting measurement errors [Magee et al., 2019a].

Fault kinematics
To establish the geometrical and kinematic relationships between the sill, its overlying forced fold, and faulting, we conducted throw analysis of the major, N-S striking, W-dipping fault (Fault 1) that borders the eastern edge of the sill. 15 Throw-depth (T-z) and throw-distance (T-x) profiles are commonly used to assess the geometry and infer the kinematics of normal faults [e.g., Baudon and Cartwright, 2008;Hongxing and Anderson, 2007;Jackson et al., 2017;Rotevatn et al., 2019]. For example, local throw minima expressed in T-x profiles may represent breached relays formed during fault segment linkage [e.g., Cartwright et al., 1996;Mansfield and Cartwright, 1996;Peacock and Sanderson, 1991].
Furthermore, changes in throw gradient (in T-z or T-x profiles) can help identify intervals containing syn-tectonic growth 20 strata, which thus constrain periods of active faulting [e.g., Ferrill and Morris, 2001;Walsh and Watterson, 1989].
Although we were not able to identify piercing points (e.g., channels) either side of Fault 1 to establish whether there was any along-strike offset of strata, we assume faulting was dip-slip and that measured throw patterns reflect displacement distribution. 25 We compiled a throw-depth plot for a representative seismic line crossing Fault 1, near its centre, by measuring the vertical offset of each mapped horizon. Where horizons adjacent to Fault 1 appear deflected, we projected their regional trend to define the fault cut-off and thereby account for both brittle and ductile strains [e.g., Mansfield and Cartwright, 1996]. We extracted expansion indices (EI) from the same line; EI reflect the difference between the hanging wall and footwall thickness of a given stratal package [e.g., Cartwright et al., 1998;Jackson et al., 2017;Thorsen, 1963]. These 30 quantitative fault measurements allow us to constrain the main periods of fault activity, as well as temporal variations in 13 the relative rates of sediment accumulation and fault-throw, at least in two-dimensions [e.g., Jackson et al., 2017]. We analysed along-strike variations in fault throw (T-x) by measuring Top Mungaroo hanging wall and footwall cut-offs every 100 m along Fault 1 on sections oriented normal to the fault; we selected this horizon for T-x analysis because it is well-imaged and occurs at a similar stratigraphic level to the sill. To assess the relationship between sill thickness and fault throw, we also measured sill thickness in the hanging wall of the fault, on the same profiles, every 100 m along-5 strike.

Sill characterisation
The studied sill comprises a >13.4 km long, N-trending, strata-concordant inner sill, bound on its eastern and most of its 10 western flanks by inwardly inclined sheets, and located within the Mungaroo Formation (Fig. 4). The Top Sill contact corresponds to a high-amplitude, positive reflection, marking a downward increase in acoustic impedance across the sedimentary strata-sill interface, and currently occurs at a maximum depth of ~3.81 s TWT (Figs 4A, D, and E). Across much of this inner sill, we identify and map a discrete Base Sill reflection, which has a high-amplitude and negative polarity (i.e. it marks the downward decrease in acoustic impedance; Figs 4B, D, and E). Within the southern and north-15 western sector of the inner sill, we observe no discrete Base Sill contact and the sill is instead expressed as a tuned reflection package; in these areas we map the Base Sill as the lowermost reflection in the tuned package but note this may not correlate to the true base sill contact (e.g., Figs 4B, D, and E). Along its eastern margin and the southern ~7.2 km of its western edge, the sill transitions into transgressive, inward-dipping inclined sheets, which also correspond to tuned reflection packages (Fig. 4). Each inclined sheet extends up into the overlying Jurassic succession, but appear to 20 terminate below Intra-Jurassic R1 horizon (e.g., Fig. 4B). The eastern inclined sheet coincides with a major, N-S striking, W-dipping, tectonic fault (Fault 1; Fig. 4). Where both eastern and western inclined sheets are developed, the entire sill is relatively narrow (up to 4.8 km wide) and the inner sill is ≲3.2 km wide (Fig. 4). North of this zone, where there is no western inclined limb, the sill abruptly widens (up to 6.4 km wide) and has a convex-outwards, lobate western termination (Figs 4A-C and E).  Table 2). We show the inner sill is locally up to ~98 ms TWT (~272±27 m) thick (Fig. 4A). In the eastern half of the intrusion where the inner sill is bound by Fault 1, its thickness varies between ~60-90 ms TWT (~166±17 m to 249±25m), broadly decreasing westwards to 40-60 ms TWT (~111±11 m to 166±17m) 5 (Fig. 4A). The inclined sheets, as well as the southern inner sill tip and its arcuate westwards termination, are expressed as tuned reflection packages, such that their thickness can only be defined as being between the limits of separability (56±5.6 m) and visibility (~7±0.7 m) for the data (e.g., Figs 4B and C). Across the inner sill we recognise discrete and abrupt changes in thickness where (  Intra-Jurassic R2 is only observed in the hanging wall of Fault 1, where it is locally offset by faults within Fault Population A (Fig. 5B). Uplift of supra-sill horizons, relative to their projected regional trends, is clearly demonstrated by Intra-Jurassic R2 where the margins of a dome-shaped fold directly overlie the western and eastern lateral tips of the sill (Figs 4D, E, and 5B). The western margin of the dome-shaped fold is a W-verging monocline, which includes folded 5 strata between the Top Triassic and Intra-Jurassic R2, whereas its eastern margin has a subtle synformal geometry immediately adjacent to Fault 1 that is only expressed between Intra-Jurassic R1 and R2; below Intra-Jurassic R1 there is no apparent folding of horizons adjacent to Fault 1 (Figs 4D, E, and 5C). Above the deepest part of the Top Sill horizon there is a general thickening of strata towards Fault 1 (Fig. 5C); this thickening in part relates to the presence of 15 stratigraphic reflections extending westwards from Fault 1 that onlap onto Intra-Jurassic R1 (e.g., Fig. 4C). There is no apparent change in this regional thickening trend of the Top Mungaroo-to-Intra-Jurassic R2 strata above the western edge of the sill (Fig. 5C). Where depressions are observed at Intra-Jurassic R2 and/or the Top Mungaroo, the intervening strata is locally thinner than adjacent areas (Fig. 5C).

20
The clear dome-shaped fold observed at Intra-Jurassic R2 and its internal variations in elevation are subtly expressed across the Base Cretaceous unconformity (Fig. 5D). Compared to deeper stratigraphic horizons, offset (up to ~0.1 s TWT) of the Base Cretaceous unconformity across Fault 1 is reduced and very minor changes in depth define a polygonal pattern (Fig. 5D). Strata between Intra-Jurassic R2 and the Base Cretaceous unconformity display complex thickness

Comparison between sill thickness and fold amplitude
We compare the present-day vertical sill thickness (ST), which we consider equal to ST0, and fold amplitude (F) at Intra-Jurassic R2 along a selected seismic section (i.e. Fig. 4B) where the sill is ~4.5 km wide ( Fig. 6; Supplementary Tables   2 and 3). We show ST ranges from 56(±5.6) m at the edges of the inner sill to a maximum of ~259(±26) m ( Fig. 6; Supplementary Table 2); note we only measure ST where the top and base sill contact reflections can be defined and thus 5 do not take into account the tuned reflection packages defining the inclined sheets. The ST profile can be sub-divided into four parts [i-iv] where ST is relatively stable, separated by abrupt increases and decreases in ST (Fig. 6); STmean of these parts decreases westwards from ~204(±20) m to ~150(±15) m, respectively (Fig. 6). The marked changes in ST correspond to where steps occur within the sill reflection(s) (Figs 4B and 6). which appear as tuned reflection packages in the seismic data, can only be defined as being between the limits of separability and visibility. The present-day fold amplitude (F) was measured in time and converted using borehole-constrained seismic velocities. We decompacted and backstripped F using a range of parameters to estimate the possible original fold amplitude (F0), which we display as an envelope.
Along the selected seismic section, the fold is ~4.8 km wide, extending slightly to the east and west of the underlying sill tips (Figs 4B and 6). The fold has a relatively flat top and is defined by a W-verging monocline on its western limb 5 (Figs 4B and 6). On its eastern limb, the fold has a synformal geometry (Figs 4B and 6); i.e. at the fault, the present-day fold amplitude (F) is ~75 m but above the eastern edge of the sill, which underlies the synform fold axis, F is ~47 m ( Fig. 6; Table 2). Along the profile, the maximum measured F of ~116 m is ~54(±5)% less than the maximum ST (i.e. 259(±26) m) and the two areas are laterally offset by ~0.59 km in 2D ( Fig. 6; Supplementary Tables 2 and 3). Overall, ST>F by an average of ~37(±7)%, except where the inner sill transitions to the inclined sheets and across one of the sill 10 steps (Fig. 6). We observe no marked variations in F where the sill appears stepped (Fig. 6).
We backstripped our measured F profile, using a range of ø0 and λ parameter values, to define an envelope bounding F0 ( Fig. 6; Supplementary Table 3). Our backstripped F0 envelope mirrors the geometry of the measured F profile, but has a greater magnitude (Fig. 6). For example, backstripping suggests the maximum F0 is between ~112-205 m, which is 15 ~2-95 m greater than F (Fig. 6); these values suggest the maximum F0 was less than the maximum ST by ~57(±7)-21(±7)%. Across most of the fold there is an overlap between ST and F0 envelopes, although T in the eastern section of the sill is locally greater than F0 (Fig. 6). Where the Intra-Jurassic R2 horizon displays a synformal geometry above the eastern sill edge, the backstripped relief of the synform fold axis above the assumed pre-fold datum is ~48-92 m and at Fault 1 it is ~75-139 m (Fig. 6). Fault 1 bounds the eastern edge of the sill and is part of Fault Population A (e.g., Figs 4B and C). There is a minor NNW-30 trending bend in Fault 1 where it is intersected in its hanging wall by a ~3 km long, NW-dipping splay fault (Fig. 5). A T-z analysis of Fault 1 reveals throw decreases upwards from ~426 ms TWT at the Intra-Mungaroo horizon to ~0.02 ms TWT at the Top Barrow, just below its upper tip ( Fig. 7A; Supplementary Table 4). Superimposed on this throw are three zones between ( Fig. 7A; Supplementary Table 4 Throw-length plot (T-x) for Fault 1 measured along the Top Mungaroo horizon, compared to the along-strike variation 20 in T where it abuts the fault. We restore the pre-intrusion throw profile by removing the effects of the sill thickness, assuming the sill has a seismic velocity of ~5.55(±0.555) km s -1 .
A T-x analysis of the Top Mungaroo horizon also demonstrates how the sill affects how throw varies along strike of Fault 1 (Fig. 7B; Supplementary Table 6). For example, the maximum present-day throw is ~606 m, but when the sill 5 thickness is accounted for (i.e. we remove the sill and thus shift overlying hanging wall horizons downwards), throw increases to ~825(±22) m ( Fig. 7B; Supplementary Table 6). The prominent throw minimum (present-day throw of ~269 m) on Fault 1, which disrupts the overall bell-shaped morphology of the T-x profile, is spatially coincident with the branch-point of the NE-SW striking hanging wall splay ( Fig. 7B; Supplementary Table 6).  Lathrop et al., 2020]. Footwall degradation during formation of the Base Cretaceous unconformity means we cannot determine whether this reduction in strain rate locally involved a period fault cessation or not [e.g., . However, kinematic analyses of syn-sedimentary faults elsewhere in the Glencoe 3D seismic survey, where little or no footwall degradation occurred, suggest faulting may have been continuous during formation of the Base 20 Cretaceous unconformity [Lathrop et al., 2020].
The top of the supra-sill fold we mapped coincides with Intra-Jurassic R2 and is onlapped by Jurassic strata beneath the Base Cretaceous unconformity (Figs 4B and 5C). Onlapping of strata onto the fold indicates Intra-Jurassic R2 represented the surface during deformation [Trude et al., 2003]. We interpret that folding occurred in response to sill 25 emplacement and, at least partially, accommodated the intruding magma volume because: (i) strata adjacent to the fold, or beneath the sill, are not folded (e.g., Figs 4B and C), indicating deformation was not driven by regional horizontal shortening but instead by a localised, underlying, forcing process [i.e. it is a forced fold; Stearns, 1978]; (ii) folding was not driven by upwards fault propagation [e.g., Hardy and Finch, 2006], as expansion indices reveal Fault 1 was surfacebreaking in the Late Triassic-to-Jurassic, prior to and likely during sill emplacement (Fig. 7A); and (iii) the lateral edge 30 of the fold broadly overlies that of the sill (e.g., Figs 4D, E, and 5B) [e.g., Magee et al., 2019a]. We also recognise depressions above the sill along Intra-Jurassic R2, which are infilled by overlying strata, and suggest these correspond to hydrothermal or volcanic vents related to sill emplacement and fluid escape (Fig. 5) [e.g., Hansen, 2006;Jamtveit et al., 2004;Planke et al., 2005]. By decompacting and backstripping the sill overburden, we estimate magma emplacement occurred in the Jurassic at a depth of ~0.9 km. Other intrusion-induced forced folds identified within the North Carnarvon Basin have been dated to the Kimmeridgian, and we consider the sill-fold pair studied here are likely a similar age [e.g., 5 Magee et al., 2013a;Magee et al., 2017].

Forced fold mechanics
Roof uplift above intrusions is typically considered to be accommodated by elastic bending of the overburden, implying the volume and amplitude of ground deformation is broadly equivalent to, and thus a proxy for, the emplaced magma 10 volume and thickness (e.g., Fig. 3A) [e.g., Bunger and Cruden, 2011;Hansen and Cartwright, 2006;Pollard and Johnson, 1973;Stearns, 1978]. For example, if the deforming overburden has no flexural strength, there should be no uplift beyond the intrusion edge [e.g., Hansen and Cartwright, 2006;Magee et al., 2019a]. Although inversion of geodetic data that capture ground deformation above intrusions can produce reasonable estimates of emplaced magma volumes by assuming the crust behaves elastically [e.g., Magee et al., 2018;Pritchard and Simons, 15 2004;Sigmundsson et al., 2020], the geometry of modelled intrusions is oversimplified compared to natural examples [Galland, 2012b]. Furthermore, field-, modelling-, and seismic-based studies demonstrate that viscoelastic and/or inelastic deformation of the overburden may accommodate magma emplacement (e.g., Fig. 3B) [e.g., Magee et al., 2013a;Magee et al., 2019a;Morgan et al., 2008;Schofield et al., 2012;Sigmundsson et al., 2020]. For example, inelastic porosity reduction and faulting of the host rock can occur during bending [e.g., Magee et al., 2017;Morgan et al., 2008]. 20 If multiple processes accommodate intrusion, as opposed to simply elastic bending, the volume of ground deformation will underestimate the intruded magma volume [Galland, 2012a]. To determine the structure of syn-emplacement ground deformation, and establish whether elastic bending solely accommodated magma intrusion, we depth-converted and decompacted the top surface of the folded sequence and compare its amplitude to sill thickness (Fig. 6). We note we cannot determine whether lateral variations in compaction degree have modified the fold shape; i.e. the true F0 profile 25 could realistically describe any pattern within the defined envelope.
Where the western limb of the fold overlies a strata-bound inclined sheet in the south of the study area, the sill is overlain by a relatively smooth, asymmetrical forced fold containing little evidence for brittle deformation (Figs 4 and 6). This geometry of the western fold limb implies that here, during sill emplacement, roof uplift was accommodated by elastic 30 bending of the overburden; this is consistent with the broad overlap between the estimated sill thickness (T) and the syn-23 emplacement surface relief (fold amplitude F0) around the western half of the sill (Fig. 6). We note that the fold has a present-day relief of ~50 m above the resolved western sill tip and appears to extend beyond the mapped edge of the sill (Figs 4D and 6). The maximum ST of the western inclined sill limb, which is defined by the limit of separability (56±5.6 m), is also less than the predicted F0 range of the overlying fold (Fig. 6). These sill-fold relationships suggest: (i) the deformed strata likely had some flexural strength, meaning our decompaction method underestimates F0 [e.g., Hansen 5 and Cartwright, 2006;Magee et al., 2019a]; and (ii) the true sill edge is unresolved in our data.
Similar to the western side of the sill, the present-day minimum relief of the fold at Intra-Jurassic R2 above the eastern edge of the fault-hosted inclined sheet is ~47 m (i.e. the synformal fold axis); this corresponds to an estimated decompacted relief of ~48-92 m ( Fig. 6; Supplementary Table 3). The amplitude of this eastern fold is also greater than 10 the potential maximum ST of the underlying inclined sheet, which is expressed in the data as a tuned reflection package (Figs 4B and 6). However, this eastern fold limb has a synformal geometry, with a half-width of ~160 m, and is upturned immediately adjacent to Fault 1. Here, the synform has a present-day relief of ~75 m, which corresponds to an estimated decompacted relief of ~75-139 m ( Fig. 6; Supplementary Table 3); i.e. F0 does not decay to zero as is observed at the western sill limb (Fig. 6). Furthermore, we show that towards this eastern limb, the estimated range of ST is broadly 15 greater than F0 (Fig. 6), although we acknowledge our decompaction analysis likely underestimates F0 as the folded section probably had flexural strength. We consider two scenarios that could produce the observed sill-fold relationships immediately adjacent to Fault 1: (i) projecting the eastward-dipping synform limb down-dip suggests F0 may have decayed to zero at Fault 1 (i.e. the fold was a monocline, similar to that defining the western limb), implying the upturned part of the fold limb could have been generated post-folding due to normal faulting (i.e. frictional fault drag; Fig. 8A) ; 20 or (ii) the synform was generated by collapse of strata during fluid escape via a pipe emanating from the sill tip ( Fig.   8B), similar to hydrothermal vents observed elsewhere [e.g., Hansen, 2006;Jamtveit et al., 2004;Planke et al., 2005].
We discount the frictional fault drag mechanism because the synform shape and magnitude is inconsistent with fault dip [i.e. it is not low-angle; Grasemann et al., 2005] or geometry [i.e. it does not comprise underlapping segments; Childs et al., 2017] (Figs 4D and E). Instead we favour a fluid escape origin for the synform, which implies there may have been 25 no prominent monoclinal limb formed above the eastern sill edge (Fig. 8B). We suggest the potential absence of a monoclinal limb above the eastern edge of the sill could be because uplift was accommodated by inversion (i.e. reverse reactivation) of Fault 1, as opposed to elastic bending (Fig. 8B). Although we favour a model involving little folding above the eastern edge of the sill, the local disparity here between F0 and ST (Fig. 6) implies compaction of overburden strata may also have contributed to accommodating sill emplacement Magee et al., 2013a;Magee 30 et al., 2019a]. Overall, we suggest sill emplacement can be broadly described by a three-stage model: (i) emplacement of a thin, layerparallel sill containing intrusive steps [Magee et al., 2019b], which spreads laterally with little or no roof uplift (Fig. 9A) [e.g., Hansen and Cartwright, 2006;Kavanagh et al., 2015;Magee et al., 2013a;Pollard and Johnson, 1973;Wilson et 10 al., 2016]; (ii) sill inflation, which drove roof uplift and ground deformation, accommodated by elastic bending above the western part of the sill and localised inversion of Fault 1 and overburden compaction above its eastern part (Fig. 9B) [Bunger and Cruden, 2011;Galland and Scheibert, 2013;Goulty and Schofield, 2008;Magee et al., 2017;Montanari et al., 2017]; and (iii) transgression of inclined sheets, which likely exploited fold-related outer-arc extensional fractures or fault opening of Fault 1, and perhaps fluid escape (Fig. 9C) [e.g., Bedard et al., 2012;Magee et al., 2013b;Siregar et al., 15 2019; Thomson and Schofield, 2008]. Subsequent burial-related compaction has modified the forced fold, by reducing its amplitude (F0 becomes F), but not the thickness of the incompressible sill (Fig. 9D) [e.g., Magee et al., 2019a]. Our work confirms that the presence of pre-existing faults can, at least partly, control intrusion-induced deformation and provide pathways for magma ascent [e.g., Gaffney et al., 2007;Magee et al., 2013b;Valentine and Krogh, 2006]. sill [e.g., Hansen and Cartwright, 2006;Kavanagh et al., 2015;Magee et al., 2013a;Pollard and Johnson, 1973;Wilson et al., 2016]. (B) Inflation of the sill causes uplift of the overburden, accommodated by elastic bending, particularly above the western part of the sill, and inversion of the pre-existing fault. Heterogeneous vertical compaction of the 5 overburden also likely accommodates magma emplacement, causing a discrepancy in between ST and F0 above the eastern part of the sill. Tensile fractures may open due to outer-arc stretching during folding and where the fault plane is 26 opened [Bunger and Cruden, 2011;Galland and Scheibert, 2013;Goulty and Schofield, 2008;Magee et al., 2017;Montanari et al., 2017]. (C) Magma exploits open fractures, forming inclined sill limbs [e.g., Bedard et al., 2012;Magee et al., 2013b;Siregar et al., 2019;Thomson and Schofield, 2008]. (D) Burial-related compaction reduces the forced fold amplitude but not sill thickness [e.g., Magee et al., 2019a].

Implications for the geometric and kinematic analysis of normal faults 5
Variations in throw (and displacement) across segmented normal faults is commonly interpreted to reflect their kinematics [e.g., Ferrill and Morris, 2001;Hongxing and Anderson, 2007;Needham et al., 1996;Peacock and Sanderson, 1991;Reilly et al., 2015;Robson et al., 2016;Rotevatn et al., 2019;Tvedt et al., 2013;Walsh et al., 2003;Walsh and Watterson, 1989;Watterson, 1986]. For example, numerical modelling shows the location of throw maxima may coincide with the nucleation site of faults [e.g., Deng et al., 2017]. Recognition of multiple throw maxima across faults 10 have therefore been related to the nucleation, growth, and eventual coalescence of initially isolated fault segments; in these interpretations, throw minima are inferred to represent sites of segment linkage (e.g., relay zones) [e.g., Cartwright et al., 1996;Jackson and Rotevatn, 2013;Mansfield and Cartwright, 1996;Peacock and Sanderson, 1991;Trudgill and Cartwright, 1994]. Our work shows that sill emplacement, roof uplift, and ground deformation in the immediate hanging wall of a fault can drive its inversion (Fig. 8B). To demonstrate the effect such intrusion-induced inversion can have on 15 the distribution of fault throw, we restored the original throw pattern along Fault 1 by removing the depth-converted thickness of the sill (Fig. 7). We show that Fault 1 accommodated ~15-310 m of inversion, locally producing zones of lower throw and higher throw gradients (Fig. 7). In extreme cases, where sill thickness exceeds the pre-intrusion throw on a fault, we envisage that intrusion-induced inversion could locally cause the fault to display a reverse motion.
Correctly identifying where intrusion-induced inversion may have modified fault throw is critical to interpreting the 20 kinematic history of a fault. Identification of intrusion-induced inversion is likely relatively simple where sills have a clear seismic expression and their geometrical relationship to faults can be defined. However, recent studies have shown that an abundance of sub-seismic sills, i.e. with thicknesses below the limit of visibility of a seismic reflection dataset, may not be recognised but can cumulatively over-thicken a sedimentary sequence [e.g., Mark et al., 2019;. If the degree of over-thickening by sub-seismic sills varies across (as well as along) a fault, its throw (and 25 displacement) distribution will be modified and poorly reflect its pre-intrusion kinematic history.
Thickening of stratigraphic packages intruded by sills may also modify expansion indices calculated across a fault. For example, in the south of our study area, the expansion index calculated between an Intra-Mungaroo and Top Mungaroo is 1.59 (Fig. 7A); typically this value, which is >1, would be interpreted to indicate the fault was active and surface-30 breaking during deposition of this sedimentary sequence [e.g., Jackson et al., 2017]. However, if we remove the thickness 27 of the sill, then the expansion index approaches 1 (Fig. 7A). Thickness of intrusive bodies should thus be excluded from analysed of expansion indices, as failure to do so may lead to errors when determining the growth history of normal faults.

CONCLUSIONS 5
Understanding the translation of magma emplacement into ground deformation is critical to volcano monitoring, which partly relies on the inversion of surface uplift data to model intrusion geometries, locations, and volumes. Such inversions assume ground deformation occurs purely via elastic bending, but there is a growing consensus that viscoelastic or inelastic processes may also generate space for intruding magma. Using seismic reflection data from offshore NW Australia, we investigate the form of a forced fold developed above a saucer-shaped sill in the Late Jurassic. We show 10 that elastic bending broadly accommodated emplacement; i.e. sill thickness and fold amplitude are equal across the western half of the sill. However, adjacent to a major pre-existing fault, roof uplift seems to have been suppressed (i.e. fold amplitude is less than sill thickness) and likely involved a combination of fault inversion and overburden compaction. Our results suggest that the presence of pre-existing faults can modify and complicate space generation for shallow-level intrusions, causing the true geometry and location of magma bodies to deviate from the shape and site of 15 their surface expression. Furthermore, we demonstrate that intrusion-induced fault inversion: (i) allowed magma o ascend up the fault; and (ii) modified the displacement distribution of the fault. Given fault displacement is commonly used to unravel fault kinematics, and thereby tectonic histories, caution should be applied when interpreting fault displacement in areas where sub-seismic sills may be present.

AUTHOR CONTRIBUTIONS
JN conducted the analysis and interpretation, wrote the first paper draft, and helped edit the manuscript. CM wrote the manuscript and contributed to analysis and interpretation. CALJ and JK contributed to interpretation and manuscript editing. BL contributed to manuscript editing and understanding of the regional fault setting. 25