Fractures and faults across intrusion-induced forced folds: a georesource 1 perspective

Abstract

vertices for input into FracPaQ are available at XXXX.

Introduction
Space for the emplacement of tabular magma bodies (sills and laccoliths), particularly at shallow-levels, is commonly created by uplift of the overlying rock and free surface (e.g.Pollard and Johnson 1973;Segall 2013).This uplift bends the overburden to produce domelike forced folds, within which localised extension and compression along outer-and innerarcs, respectively, drives internal deformation (Fig. 1a) (e.g.Pollard and Johnson 1973;Magee et al. 2013;Wilson et al. 2021); these forced folds are similar to periclinal folds generated above faults or salt bodies, and by differential compaction (e.g.Stearns 1978; Cosgrove and Hillier 1999;Lisle 1999;Meng and Hodgetts 2020).Like these other types of four-way dip closure, intrusion-induced forced folds in sedimentary basins have been targeted for petroleum exploration, with some found to host hydrocarbon reserves (e.g.Schutter 2003;Rodriguez Monreal et al. 2009;Jackson et al. 2020).As intrusion-induced forced folds can thus trap fluids, it is worth considering their suitability as potential targets for water, geothermal, or mineral/metal exploration, as well as CO2 storage (e.g.Weis 2012;Scott et al. 2015;Montanari et al. 2017;Wilson et al. 2021).
Drivers of forced folding have limited lateral extents and underlie the deforming rock volume, meaning elastic bending dictates fold development as opposed to buckling, which involves (sub-)horizontal compression (e.g.Pollard and Johnson 1973;Cosgrove and Ameen 1999;Goulty and Schofield 2008).This bending occurs radially about a central point or axis, producing a non-developable surface fold with non-zero Gaussian curvature (e.g. a dome); most buckle folds form developable surfaces (Figs 1a and b) (e.g.Lisle 1999;Mynatt et al. 2007).Being a non-developable surface means forced fold growth induces radial and circumferential tension along outer-arcs, locally instigating internal fracturing and normal faulting of the bending rock volume (Figs 1a and c) (e.g.Stearns 1978;Cosgrove and Ameen 1999).These tensional stresses become compressional in inner-arc sections, where deformation bands and compaction may occur (Figs 1a and c) (e.g.Ramsey 1968;Pollard and Johnson 1973;Wilson et al. 2021).Critically, the development of tensional and compressional structures within a rock volume can markedly change its porosity and permeability, influencing fluid flow (e.g.Fossen and Bale 2007;Sanderson and Nixon 2015;Dimmen et al. 2020).Yet few studies have explored how bending-related deformation within intrusion-induced forced folds may affect fluid flow (e.g.Wilson et al. 2021).To assess the suitability of intrusion-induced forced folds as potential exploration or storage targets, for a variety of fluids critical to the energy transition, we need to constrain how their deformation history impacts porosity and permeability (Wilson et al. 2021).
Here, I build on previous work examining fracture development in forced folds (Cosgrove and Ameen 1999;Pearce et al. 2011;Cosgrove 2015;Wilson et al. 2021) by analysing fracture networks across the top surface of intrusion-induced forced folds.Specifically, I use satellite imagery and bathymetry data to map faults and fractures across: (1) long-lived forced folds within the Erta'Ale Volcanic Segment, in the Danakil Depression sedimentary basin, Ethiopia (Magee et al. 2017);and (2) at Cordón Caulle (Chile) and West Mata (Lau Basin, SW Pacific), where recent individual intrusion events produced new forced folds (e.g.Castro et al. 2016;Chadwick Jr et al. 2019).I compare these mapped fold and fracture geometries to forced folds recognised in seismic reflection data (Hansen and Cartwright 2006) and generated in physical experiments (Montanari et al. 2017;Henriquet et al. 2019;Poppe et al. 2019;Montanari et al. 2020;Warsitzka et al. 2022).With these data, I aim to constrain fracture network characteristics of intrusion-induced forced folds.This work will help inform predictions of subsurface fracture networks within forced folds, and contribute to the assessment of forced folds in georesource exploration and storage.

Erta'Ale Volcanic Segment
Situated in the Danakil Depression, a Pleistocene-Recent sedimentary basin comprising a thick sequence of evaporites, the Erta'Ale Volcanic Segment (EAVS) marks one of the final phases of continental break-up along the Red Sea rift (Fig. 2a) (e.g.Bastow et al. 2018).The EAVS contains a series of volcanoes surrounded by basaltic to silicic lavas, which primarily emanate from fissures oriented NW-SE (Fig. 2a) (e.g.Watts et al. 2020).Spatially associated with several volcanoes in the EAVS are dome-like features that are heavily fractured and faulted: (1) the Alu and Alu South domes are close to and partially underlie the composite Dalafilla stratovolcano, respectively (Fig. 2b) (Pagli et al. 2012;Magee et al. 2017);(2) adjacent to the Borale'Ale stratovolcano is a sub-circular dome, previously interpreted to be a shield volcano, containing an elliptical central graben that itself hosts a small volcanic vent (Fig. 2c) (Barberi and Varet 1970;Watts et al. 2020); and (3) a broad area of uplift beneath the Gada'Ale volcano (referred to as Gada'Ale East) associated with an adjacent, complex dome (referred to as Gada'Ale West), both of which are inferred to be formed due to underlying salt movement (Fig. 2d) (Barberi and Varet 1970).Another dome-like structure is present ~30 km north of the EAVS at the Dallol volcano (Fig. 2e) (e.g.López-García et al. 2020).Mapping of lava flows that deflect around these domes suggest they formed over the past <80 Kyr (e.g.Fig. 2b) (Magee et al. 2017;Watts et al. 2020).Ground deformation geodetically detected at these sites, except Borale'Ale, over the past 35 years indicates the domes continue to periodically uplift and subside, likely linked to subsurface magmatism (Amelung et al. 2000;Nobile et al. 2012;Pagli et al. 2012;Albino and Biggs 2021).For example, a 23.2 × 10 6 m 3 lava eruption from a NW-trending fissure ~2 km NW of the Dalafilla stratovolcano summit in 2008 was: (1) preceded by ~9 cm of uplift across Alu over three months; and (2) accompanied by ~1.9 m and ~1 m of subsidence at the Alu and Alu South domes, respectively (Pagli et al. 2012).Modelling of this ground deformation suggest Alu and Alu South are underlain by a sill at 1 km depth, possibly with a saucer-shaped geometry, and a larger magma reservoir at 4 km (Pagli et al. 2012;Magee et al. 2017).Based on their morphology and relation to magmatic or slat movement events, it is plausible that these dome-like features at Alu, Alu South, Borale'Ale, Gada'Ale, and Dallol are forced folds (Barberi and Varet 1970;Magee et al. 2017).

Cordón Caulle
On 4 th June 2011, the rhyolitic volcano Cordón Caulle, Southern Chile, produced an explosive sub-Plinian eruption followed by lava effusion beginning on 15 th June (Fig. 3a) (e.g.Castro et al. 2016;Wadsworth et al. 2022).Between the ~8th June and 3 rd July 2011, surface elevations in a ~12 km 2 area around the vent site increased by up to ~240 m; these elevation changes can partly be attributed to eruption of a ~35-60 m thick lava, but primarily relate to intrusion-induced surface uplift of previous tephra layers (i.e.forced folding; Fig. 3a) (Castro et al. 2016).Modelling this ground deformation suggests uplift was driven by emplacement of a laccolith, with a 0.8-2 km radius and ~0.8 km 3 volume, at a depth of 20-200 m and pressure of 1-10 MPa (Castro et al. 2016).Development of the forced fold was accompanied by surface fracturing and faulting (Fig. 3a) (e.g.Castro et al. 2016;Wadsworth et al. 2022).Subsidence of up to 40 m occurred across the forced fold from August 2011 onwards, and has been related to magma migration out of the laccolith (perhaps coupled with thermal contraction) (Zheng et al. 2020) or sintering of pyroclasts during intrusion growth (Wadsworth et al. 2022).

West Mata
The West Mata submarine volcano is located between the NE Lau Spreading Centre and the Tofua Volcanic Arc near Fiji and Samoa in the SW Pacific (Fig. 3b) (Chadwick Jr et al. 2019).Towards the NE of the volcano base, bathymetric depths decreased by up to 64 m across a 0.73 x 10 6 m 2 area over some period between 2012 and 2016 (Chadwick Jr et al. 2019).Part of this depth change can be attributed to emplacement of lava from a NE-SW trending fissure, but most relates to uplift of seafloor sediments and creation of a dome (forced fold) bisected by numerous fractures (Fig. 3b) (Chadwick Jr et al. 2019).

'Fold B'
Seismic reflection data reveal 'Fold B' is a dome-shaped forced fold, 3.5 km in diameter and up to ~250 m high, developed ~1 km above a 3 x 2.5 km, up to ~300 m thick, saucer-shaped sill (Hansen and Cartwright 2006;their Fig. 4).This sill-fold pair is situated in the NE Rockall Basin and formed in the Late Paleocene-to-Early Eocene within a siliciclastic sedimentary succession (Hansen and Cartwright 2006).A series of normal faults cross-cut the forced fold (Hansen and Cartwright 2006;their Fig. 10).3a).For West Mata, I use high-resolution (~1 m) bathymetry data collected using a multibeam sonar system on the AUV Sentry during part of a two-leg expedition by the R/V Falkor crew in 2017 (Fig. 3b) (Chadwick Jr et al. 2019).The 3D seismic reflection survey (T38) used to map 'Fold B' has a vertical and horizontal resolution of up to ~68 m, if we consider each is equivalent to a quarter of the seismic wavelength (λ/4) (Hansen and Cartwright 2006;Brown 2011).I also generate fold and fracture maps for physical experiments using select published images (see Supplementary Material) (Montanari et al. 2017;Henriquet et al. 2019;Poppe et al. 2019;Montanari et al. 2020;Warsitzka et al. 2022).

Methodology
Where fold outlines could be confidently identified in the remote sensing data and model images, I measure fold length, width, and map-view area (A).I also measure fold amplitudes where elevation data, or cross-sections through the folds, are available; for these measurements, I assume that the pre-fold datum follows the regional trend of the current free surface outboard of the fold outline (e.g.Figs 1a and c).However, there are some uncertainties in the measurement of forced fold length and amplitude: (1) we can rarely establish the original surface topography prior to emplacement and folding, so often cannot accurately constrain true amplitudes; (2) syn-or post-emplacement deposition of sediments or resurfacing by lavas may alter apparent forced fold heights or regional base levels (e.g.Dobb et al. 2022;Warsitzka et al. 2022) ; (3) fold crests may have been eroded (e.g.Hansen and Cartwright 2006); and/or (4) measurements from 2D seismic reflection data, which are rarely depth-converted and decompacted (Magee et al. 2019), or physical model crosssections may not intersect forced fold maximum amplitude or length (e.g.Jackson et al.

2013).
For the EAVS forced folds, available Shuttle Radar Topography Mission (SRTM) 1 Arc-second global data allow me to measure their current surface area (Ai).Comparing the map-view and current surface area measurements of the EAVS forced folds provides an estimate of the extensional strain across the fold tops.The SRTM data also allow me to calculate the Gaussian curvature (K; Fig. 1b) of the EAVS folds by extracting a point cloud grid, with spacings of 30 m, for import into the PyVvista module for Python (Sullivan and Kaszynski 2019); forced folds elsewhere were not analysed with this method as their respective data were not in suitable formats.PyVista takes the X, Y, Z co-ordinate data of the point cloud to create a mesh (Sullivan and Kaszynski 2019), and then calculates the Gaussian curvature of each node from their principal curvatures (k1, k2; Fig. 1b) (Lisle 1999).
I interpret linear features recognised across the studied forced folds as fractures and faults, but acknowledge some may be related to fluvial incision (e.g.Henriquet et al. 2019), gravitational collapse, and/or processing artefacts; without ground-truthing, the fracture maps cannot be validated.Although some linear features mapped may thus not relate to extension during folding, I note that: (1) fractures can focus fluvial incision, meaning mapped channels may be a proxy for fracture locations (Henriquet et al. 2019); and (2) gravitational processes could affect the long-term distribution of fractures in folds.From the mapped fractures and faults, I use FracPaQ software to analyse their network properties, such as trace and segment line length and strike, fracture intensity and density, and connectivity (Fig. 4) (Healy et al. 2017).Because the entire traces of all resolved fractures are mapped across the forced folds, no adjustments are required to account for fractures extending beyond study limits.Deriving fracture trace and segment length distributions is critical predicting fracture network attributes at smaller or larger scales (e.g.Rizzo et al. 2017).In FracPaQ, these distributions are statistically analysed using Maximum Likelihood Estimators (MLE), which establishes the probability of whether the data is best-fit by power-law, log-normal, or exponential distributions (Healy et al. 2017;Rizzo et al. 2017).Fracture intensity (P21) describes the total fracture length in set area, whereas fracture density (P20) measures the number of fractures in the same area (Healy et al. 2017); these parameters were calculated using a circular scan window method (Healy et al. 2017), but only assessed for natural forced folds because the number of fractures created in modelled forced folds is generally too low to be statistically meaningful.As Gaussian curvature is a measure of 3D strain, fracture intensity and density should increase where K is greatest (Lisle 1999).To assess network connectivity, FracPaQ identifies I-, Y-, and X-nodes of fractures, whereby I-nodes correspond to isolated fracture tips, Y-nodes occur where one fracture abuts another, and X-nodes where fractures cross-cut each other (Fig. 4) (Sanderson and Nixon 2015;Healy et al. 2017).All node maps obtained from FracPaQ were manually verified and adjusted where needed.From the number of these nodes (NI,Y,X) per forced fold, I calculate the number of lines bound by I-and Ynodes (NL), the number of branches (NB) defining portions of fractures bound by any two nodes, the average number of connections per line (CL), and the average number of branches per line (CB) (Sanderson and Nixon 2015): Both CL and CB are useful indicators of connectivity (Sanderson and Nixon 2015;Healy et al. 2017).For example, simulated percolation of randomly oriented lines of a fixed length occurs at CL = 3.57 (Sanderson and Nixon 2015).

Results
The studied natural forced folds are circular to elliptical, with length to width aspect ratios of ~1.00-2.05,and they have amplitudes, lengths, and map areas that range from ~40-368 m, ~0.9-4.1 km, and ~0.27-16.01km 2 , respectively (Figs 2-3 and 5-6; Table 1).For those in the EAVS, comparison of map-view area and current surface area measurements suggests the top of the forced folds have increased in size by 0.1-4.97%during uplift (Table 1).There is a moderate, positive power-law relationship between fold length and maximum amplitude of most these natural forced folds (R 2 = 0.51), but this fit decreases if the Dallol forced fold is included (R 2 = 0.13) (Fig. 7).Other forced folds show similar length to amplitude relationships, akin to published lengths and thicknesses of sub-horizontal tabular intrusions (Fig. 7).For forced folds produced in physical models (Fig. 8; Table 2), there also appears to be a moderate, positive power-law relationship between the fold length and amplitude (R 2 = 0.66) where this data is available (Fig. 7).The power-law fit between all natural and modelled fold lengths and amplitudes is strong (R 2 = 0.89) and positive, regardless of whether Dallol is included or not (Fig. 7).
The fractures and their constituent segments mapped along the top of the natural forced folds vary in number and length (Figs 5-6 and 9-10; Table 1).Segment numbers and total trace length are particularly low for the seismically imaged 'Fold B' (253 segments totalling 2.376 km long), compared to those mapped in satellite or bathymetry data, for which 758-4160 segments are mapped that total trace lengths are ~13.4-114.6 km (Figs 5 and 10; Table 1).Within each natural forced fold, there is typically a relatively reduced amount of fracture traces or segments at small length fractions, particularly for 'Fold B' (Figs 9-10).The probability that the length distributions of these fracture populations describe log-normal, power-law, or exponential relationships are often similar, but: (1) the probability that trace or segment lengths define power-law distributions are always >95%; and (2) for some forced folds, the probability that trace and segment lengths define a log-normal (e.g.Alu South) or exponential (e.g.Gada'Ale West) distribution are relatively low (<80%) (Figs 9-10; Table 3).
There is a moderate, positive power-law relationship between fold area and total trace length of the natural forced folds excluding 'Fold B' (R 2 = 0.55), and those physical models where these geometry parameters are reported (R 2 = 0.60); the power-law fit between the natural (excluding 'Fold B') and model forced folds is strong (R 2 = 0.99) (Fig. 11; Tables 1-2).
The fracture networks, including those observed across modelled forced folds, typically show a broad range of strike orientations, although many contain fracture populations preferentially oriented sub-parallel to the fold long axes (Figs 5-6 and 8).
Fracture distributions are also variable across individual forced folds (Figs 5-6 and 8); e.g.fracture intensity and density typically appear greatest where major normal faults are developed (e.g. in Borale'Ale) and/or Gaussian curvature is highest (Figs 5-6).The connectivity of the studied fracture systems in natural forced folds is low (CL <1.31 and CB <1.17), being dominated by I-nodes and containing <10% X-nodes (Figs 5-6 and 12a; Table 2).Connectivity of fracture networks produced within modelled forced folds is also typically low but does increase up to CL values of 3.01 and CB values of 1.66 (Fig. 12A; Table 3).
Where physical models provide constraints on how fracture networks developed through time, it is clear that connectivity generally increases via the proportional formation of more Y-and X-nodes following power-law (e.g.SPCTIN06; R 2 = 0.78) or exponential (e.g.Exp1B; R 2 = 0.80) relationships (Fig. 12).However, decreases in connectivity can occur when eruptions resurface the folds (e.g.Fig. 13).

Discussion
As the four-way dip closure form of intrusion-induced forced folds can trap fluids (e.g.For example, their association with magmatism means intrusion-induced forced folds may trap hydrothermal fluids, creating: (1) suitable geothermal energy exploration targets in active volcanic settings (e.g.Scott et al. 2015;Montanari et al. 2017); or (2) important mineral/metal accumulations, such as porphyry copper deposits (e.g.Weis 2012).Similarly, ancient intrusion-induced forced folds may host aquifers (e.g.Wilson et al. 2021), or could potentially provide suitable CO2 storage sites (cf.Tueckmantel et al. 2012).Critically, forced folding involves bending of a rock volume, which locally induces internal fracturing and faulting, thereby modifying permeability (e.g.Jackson and Pollard 1990;Cosgrove and Ameen 1999;Wilson et al. 2021).These changes in permeability can enhance the fluid storage potential of these traps, but can leading to breaching and fluid leakage (see Cosgrove 2015 and references therein).To appraise whether intrusion-induced forced folds may provide suitable fluid storage sites, we need to establish how their evolution affects fracture connectivity, which controls host rock permeability (e.g.Sanderson and Nixon 2015;Wilson et al. 2021).Furthermore, because most intrusion-induced forced fold exploration targets will be in the subsurface, we will lack direct information on the geometry or growth of their fracture network.Stochastic models are thus required to simulate potential fracture patterns and their impact on fluid flow, which themselves need to be underpinned by statistical characterisation of natural fracture networks (e.g.Riley 2005).

Intrusion-induced forced folding and fracturing
Emplacement of tabular magma bodies at shallow-levels commonly drives roof uplift, producing dome-like forced folds bisected by extensional fracture and normal fault networks (e.g.Pollard and Johnson 1973;Magee et al. 2013;Segall 2013; van Wyk de Vries et al.

2014; Magee et al. 2017).
There is a strong, positive relationship (R 2 = 0.89) between forced fold length and maximum amplitude, broadly consistent with reported lengths and thicknesses of tabular intrusions (Fig. 7) (see also Magee et al. 2017 and references therein).
In addition to this geometrical similarity, the fracture networks the studied forced folds all contain fractures and faults of variable orientation and length, which increase in intensity and density with fold curvature (Figs 5-6 and 8; Table 1).Previous studies examining fracturing across periclinal folds associated with other forced folding mechanisms also describe similar fracture networks characteristics (e.g.Stearns 1978;Cosgrove and Ameen 1999;Lisle 1999;Cosgrove 2015).These findings support other works showing that it is the behaviour of the deforming rock volume during bending, itself a function of lithology, size of the forcing feature, and strain rate, which primarily controls forced folding and fracturing (e.g.Pollard and Johnson 1973;Stearns 1978;Gholipour et al. 2016).The similarity between intrusioninduced forced folds (e.g.Alu, Alu South, Cordón Caulle, and West Mata) (Pagli et al. 2012;Castro et al. 2016;Magee et al. 2017;Chadwick Jr et al. 2019) and the Gada'Ale East and West domes, which have been attributed to underlying salt movement (Barberi and Varet 1970), could thus be interpreted as evidence that: (1) the Gada'Ale folds formed in response to magma (and salt?) movement, consistent with recognition of recent dyke-related uplift near Gada'Ale East (Amelung et al. 2000); or (2) the geometry and fracturing of dome-like folds is not diagnostic of their driving mechanism.Overall, it thus seems that the geometry and growth of fracture networks during bending is largely independent from the mechanism driving folding, implying we could use exposed intrusion-induced forced folds to benchmark fracture network prediction of those in the subsurface (e.g.Gholipour et al. 2016).

Fracture length distribution
The fracture length distribution of a sample set is often used to predict fracture network characteristics at smaller and/or larger scales (e.g.Bonnet et al. 2001).By using a robust statistical approach to assess probability of fit to different distributions (Rizzo et al. 2017), I show that most intrusion-induced forced folds contain fracture networks with trace and segment lengths compatible with a power-law relationship (Figs 9-10; Table 3).Yet it should be noted that for some datasets, log-normal (e.g.Dallol segment lengths) or exponential (e.g.Alu trace lengths) distributions appear more probable fits (Table 3).A limitation with this analysis is that the resolution of the data may mean short fracture traces or segments are undersampled (Figs 9-10), which can cause power-law distributions to appear log-normal (Bonnet et al. 2001).

Fracture connectivity
My data reveal that the connectivity of fracture networks across the tops of natural and modelled forced folds tends to remain relatively low (CL <3.01 and CB <1.66; Fig. 12; Tables 1-2) (see Sanderson and Nixon 2015 and references therein).These results contrast with field-based analyses of a forced monocline above the Trachyte Mesa intrusion, Utah, which show fractures and deformation bands are locally well-connected (Wilson et al. 2021).There are several possible reasons for these disparities in fracture connectivity.Firstly, the remote sensing and seismic reflection data I use to analyse natural forced folds have limited resolutions of metres to tens of metres (e.g.Hansen and Cartwright 2006).It is thus plausible that unidentified fractures may be present, or identified fractures extend further, at scales below these data resolutions, which could lead to an increase in connectivity (e.g.Nixon et al. 2012).Secondly, my study distils a connectivity value for the entire top surface of each forced fold, obscuring zones where connectivity may locally be enhanced due to relatively higher fracture intensity, density, and/or fold curvature (e.g.Lisle 1999;Wilson et al. 2021); future work should focus on partitioning intrusion-induced forced folds, likely based on variations in curvature, to further assess connectivity patterns.Finally, I analysed fracture patterns on the top surface of forced folds, whereas the analyses of Trachyte Mesa examined fractures and deformation bands expressed on rock walls that form a cross-section through the forced fold (Wilson et al. 2021); i.e. my work provides some insight into the lateral connectivity of intrusion-induced forced fold fracture networks, but Wilson et al. (2021) provide a robust assessment of vertical connectivity.Overall, comparing our work demonstrates that ground-truthing is ideally required to test remotely determined connectivity of fracture networks (Wilson et al. 2021).Furthermore, it will be crucial to establish how connectivity varies in 3D across entire forced folds (Wilson et al. 2021), which will require integrating analyses of well-exposed forced folds and those imaged in 3D seismic reflection data.

An opportunity for CO2 storage?
Successful sequestration of CO2 through in situ mineral carbonation in basaltic rocks has opened up lava fields and volcanoes as potential exploration targets for CO2 storage (e.g.Matter et al. 2016;Holford et al. 2021;Raza et al. 2022;Fedorik et al. 2023).This method typically relies on injection of either water and dissolved CO2, or supercritical CO2, into basalts that the fluids react with to permanently fix CO2 in carbonate minerals (e.g.McGrail et al. 2014;Snaebjörnsdóttir et al. 2020).Permeability of the basalts is thus key as it enables fluid flow away from injection sites, and increases the surface area of the rock that fluids can react with (e.g.Fedorik et al. 2023).Basalt lavas often contain a variety of fracture sets (e.g.cooling joints) and porosity (e.g.Snaebjörnsdóttir et al. 2018;Holford et al. 2021), but I those within intrusion-induced forced folds (e.g. in the EAVS) will contain additional fracture sets due to bending-related stresses.Given forced folds can also form suitable fluid traps, if reservoirs and seals are in place, it seems reasonable that those containing basaltic lava flows may form suitable CO2 storage targets.

Conclusions
Intrusion-induced forced folds commonly contain an array of fractures generated by bendingrelated stresses during uplift.Coupled with the dome-like geometry of these forced folds, the presence of complex fracture networks suggests they may form suitable pathways and traps for fluid flow.By comparing natural forced folds developed recently and those produced in physical experiments, I show that there is: (1) a positive relationships between forced fold length and amplitude; (2) fracture networks comprise traces and segments with variable lengths, predominantly conforming to power-law distributions, and orientations; (3) fracture intensity and density generally increases with fold curvature and/or the presence of major normal faults; and (4) connectivity across the top of forced folds appears relatively low, but this may be due to limitations in the resolution of the data used.Fold and fracture development appear to be largely independent of the mechanism driving uplift, instead being related to the behaviour of the deforming rock during bending; we can thus use forced folds at the surface to inform predictions regarding fracturing of subsurface forced folds if they share similar host rock geology.Critically, intrusion-induced forced folds should be considered as exploration targets in the search for water aquifers, geothermal potential, and minerals/metals.Fracturing induced by bending may also increase the permeability of lavas within forced folds, potentially enhancing their suitability for CO2 storage. (https://www.marine-geo.org/tools/search/DataSets.php?data_set_uids=24446,24447).
Seismic reflection and physical model data are published, with some of the images used taken from associated supplementary files.The vertices of fractures mapped in this work are Any other data is provided in the Figures and Tables.showing how buckling creates folds that can typically be described as a developable surface, whereby one of the principal curvatures (k1, k2) is zero, meaning the Gaussian curvature (K) is also zero.In contrast, uplift driven by forced folding creates a non-developable surface with non-zero Gaussian curvature (modified from Lisle 1999).(c) Seismic reflection image from the Glencoe 3D survey offshore NW Australia depicting a forced fold above a thick sill (same sill as studied by Dobb et al. 2022).Field photograph showing forced folding above the dioritic Trachyte Mesa intrusion in the Henry Mountains, Utah.The sandstone beds thin across the fold due to bending-related porosity reduction (Morgan et al. 2008).The probability charts test the fit of the data to a log-normal, power-lar, or exponential distribution (see Table 4).n is the number of fracture traces mapped.Estimator probability (Pr) charts of fracture segment length for the natural forced folds studied.The probability charts test the fit of the data to a log-normal, power-lar, or exponential distribution (see Table 3).n is the number of fracture traces mapped.Exp1B and Exp2B (Warsitzka et al. 2022) and SPCTIN06 (Poppe et al. 2019), showing its evolution can be described by power-law or exponential relationships.Reductions in connectivity may be attributed to resurfacing of the forced fold by erupted products (see Fig. 13).

I
use Google Earth and ArcGIS Pro World imagery of different vintages to map potential forced folds and linear features across them in the EAVS and at Cordón Caulle, at a resolution of ~30 m (Figs 2 and

Schutter 2003 ;
Rodriguez Monreal et al. 2009), and these structures are present in many sedimentary basins and active volcanic settings worldwide (e.g.Pollard and Johnson 1973;Holford et al. 2012;van Wyk de Vries et al. 2014;Magee et al. 2016;Magee et al. 2017;Tian et al. 2021;Kumar et al. 2022), we should consider them as potential fluid storage sites.

Figure captions Figure 1 :
Figure captions

Figure 3 :
Figure 3: (a) Google Earth imagery of the Cordón Caulle forced fold, from before and after

Figure 4 :
Figure 4: Schematic showing fracture trace and segment definition, as well as I-, Y-, and X-

Figure 5 :
Figure 5: Compilation of maps for the EAVS forced folds highlighting their elevation (SRTM

Figure 6 :
Figure 6: Compilation of maps for Cordón Caulle, West Mata, and 'Fold B' showing their

Figure 7 :
Figure 7: Plot of fold length and amplitude for the studied forced folds, including those

Figure 8 :
Figure 8: Segment strike sand fracture connectivity maps of the physical model forced folds

Figure 11 :
Figure 11: Plot of forced fold area and total fracture trace length (see Tables 1-2 for data).

Figure 12 :
Figure 12: (a) Ternary diagrams showing the proportion of I-, Y-, and X-nodes for each

Figure 13 :
Figure 13: Model photograph and fracture map of Exp2B showing fractures developed 4662 Fig. 2a

Table 1 :
Natural forced fold geometry and fracture networks Forced fold geometry Fracture connectivity † indicates measurements could not be acquired from available data

Table 2 :
Modelled forced fold geometry and fracture networks

Table 3 :
Maximum Likelihood Estimation of fracture length distribution