Salt Diapir‐Driven Recycling of Gas Hydrate

By harnessing both hypothetical, synthetic basin and gas hydrate (GH) system models and real‐world models of well‐studied salt diapir‐associated GH sites at Green Canyon (Gulf of Mexico) and Blake Ridge (U.S. Atlantic coast), we propose and demonstrate salt movement (and in particular, diapirism) to be a new mechanism for the recycling of marine GH. At Green Canyon, for example, we show that by considering this newly proposed diapir‐driven recycling mechanism in conjunction with previously proposed lithological control on sandy‐reservoir‐hosted hydrate at the base of the GH stability zone (BGHSZ; ∼bottom‐simulating reflector, BSR), modeled GH saturations match drilling data. Overall, salt diapir movement‐induced GH recycling provides a temperature‐driven mechanism by which GH saturations at the BGHSZ may reach >90 vol. % and by which GH volumes near and free gas volumes beneath the BGHSZ may be increased significantly through time. Interestingly, comparison of salt diapir‐driven recycling and sediment burial‐driven recycling scenarios suggests notably higher rates of recycling via diapir‐driven versus burial‐driven processes. Our results suggest that GH and associated free gas accumulations above salt diapir crests represent particularly attractive targets for unconventional and conventional hydrocarbon resource exploration and for scientific and academic drilling expeditions aimed at exploiting GH systems. Salt basins containing GH systems—including passive margin basins of the Gulf of Mexico, southeastern Brazil, and southwestern Africa—are therefore compelling localities for studying salt‐driven GH recycling and for salt diapir‐associated natural gas exploration.

trapped in the sediment matrix moves downward relative to the pressure and temperature conditions to a part of the subsurface where hydrate is no longer stable (below the BGHSZ), leading to GH decomposition and release of gas and water. This buoyant gas may then migrate upward and be reincorporated as hydrate at the new BGHSZ (i.e., the BGHSZ now located in relatively shallower sediment) or may be trapped as free gas beneath the BGHSZ. Numerous processes can drive GH recycling. It has most often been described as a response to sedimentation and sediment burial (e.g., Burwicz et al., 2017;Crutchley et al., 2019;Haacke et al., 2007Haacke et al., , 2008Kvenvolden & Barnard, 1983;Minshull & White, 1989;Nole et al., 2018;Paull et al., 1994;Schmidt et al., 2022;You & Flemings, 2021), but also as a response to tectonic uplift-driven pressure reduction (Haacke et al., 2007;Minshull & White, 1989), sea-level fall-driven pressure reduction (Haacke et al., 2007), bottom-water warming (Haacke et al., 2007;A. Li et al., 2017), deglaciation (Haacke et al., 2008), and changes in geothermal gradients (Haacke et al., 2008). However, although salt bodies (including domes and diapirs) are known to modify local to regional temperature fields and heat flow, the potential impact of salt tectonics-driven temperature changes on the GH recycling process has not yet been explored.
Due to its high thermal conductivity, salt promotes enhanced heat flux in sedimentary basins, resulting in local heat flow anomalies and elevated temperatures above salt diapirs (e.g., Corrigan & Sweat, 1995;Jensen, 1983Jensen, , 1990Nagihara et al., 1992;O'Brien & Lerche, 1984, 1988Selig & Wallick, 1966;Vizgirda et al., 1985;Yu et al., 1992). Salt can therefore impact the stability of GH systems. For instance, salt diapir-modulated heat flows are speculated to influence gas hydrate stability zone (GHSZ) thickness in the Barents Sea (Chand et al., 2008), while salt diapir-associated heat flux (and flux of saline waters) is speculated to impede hydrate formation in the northern Gulf of Mexico's Garden Banks and Mississippi Canyon areas . Salt diapir-driven upwarping of the BGHSZ is observed in the Gulf of Mexico's Green Canyon and Walker Ridge areas (possibly causing hydrate dissociation; Portnov et al., 2020) and at the Blake Ridge Diapir (Hornbach et al., 2005;Taylor et al., 2000). Nonetheless, the impact of salt (via its modulation of heat flow) on GH system dynamics is relatively underexplored. Furthermore, previous studies have primarily documented the destructive effects of salt-mediated temperature changes on GH systems, whereas the potential for salt-mediated temperature changes to have a constructive effect on GH systems by causing GH recycling has, to the best of our knowledge, never been documented.
In this paper, we investigate the influence of salt diapir movement and associated changes to basin thermal conditions on the GH recycling process. Using the commercial basin simulation and petroleum system modeling software PetroMod™, we build and interrogate various two-dimensional theoretical (synthetic) basin-scale GH system models to assess the impact of salt diapirism-as well as variations in basal heat flow, salt diapir diameter, salt stock height, and sediment thermal conductivities-on hydrate formation and the potential for GH recycling. We then conduct two real-world case studies of Green Canyon (Gulf of Mexico) and Blake Ridge (U.S. Atlantic coast) salt diapirs, modeling salt movement and GH system evolution through time and deriving quantitative estimates of hydrate and free gas saturations at these locations. We compare the Green Canyon and Blake Ridge model results with previous modeling work, seismic data, and drilling data on the distribution and saturation of GH and free gas at these localities.

Synthetic Modeling Approach
We construct various purely synthetic, geologically reasonable 2-D basin and GH system models to investigate numerous scenarios for the interaction of salt diapirs, heat flow, and GH systems through time. In addition to modeling salt diapir movement and its corresponding influence on temperature and hydrate stability, we examine, via model scenarios detailed below, the impact of variations in (a) basal heat flow, (b) salt diapir diameter, (c) salt stock height, and (d) sediment thermal conductivity on GH stability and corresponding hydrate and free gas accumulations.

Synthetic Model Dimensions and Surfaces
Synthetic model dimensions are provided in Table 1, and a standard base case synthetic model is illustrated in Figure 1. Model layers, as well as their corresponding ages and properties, are listed in Table 2. Synthetic models include domed sediment above the salt diapir crest, creating positive bathymetric relief of ∼200 m, consistent with observations of domed sediment associated with various salt diapirs (e.g., Paull et al., 1996;.

Stratigraphy, Lithologies, and Rock Properties
Sediment thicknesses, sedimentary facies, sedimentation rates, and rock properties used for the base case synthetic models are listed in Table 2. Table 3 provides different sediment thermal conductivity combinations tested for the synthetic modeling scenario implemented to investigate the impact of variable sediment thermal conductivities on GH stability and distribution.
Here, we use a base case salt diapir diameter of 2 km for most synthetic modeling, but also specifically test the influence of salt diapir diameters in one set of modeling scenarios. For this scenario set, we test six diapir diameters: 1 km, 2 km, 4 km, 6 km, 8 km, and 9 km. For all modeling, we assume salt diapirs are characterized by axially symmetrical vertical stocks. We assume diapirs to be slightly conical in shape, consistent with, for instance,  (1, 2, 4, 6, 8, and 9 km), as described in the text. b Consistent with 1.5 km of sediment above the diapir crest, and a diapir height of 6 km (Portnov et al., 2020). c Based on a 2-3 km diapir width estimated from seismic data and modeling by Portnov et al. (2020); however, it must be noted that assigning an essentially constant diapir width is a highly simplified assumption, given the substantially complex diapir geometries imaged at Green Canyon (Portnov et al., 2020). d Portnov et al. (2020). e Paull et al. (1996) and Hornbach et al. (2005). f See text for explanation.  Hemipelagic sediment on continental margins (e.g., Hill et al., 2007;Müller & Suess, 1979). c Consistent with marine hemipelagic sedimentation rates (Müller & Suess, 1979) and also reflective of decreasing sedimentation rates as salt minibasins transition from basin-filling to sediment-bypass stage (e.g., Brunt et al., 2004;Prather et al., 1998). d Athy (1930) and Hantschel and Kauerauf (2009). e Except for the variable thermal conductivities modeling scenarios (Table 3). f See, for example: Horai (1971) and Revil (2000). g As in Burwicz et al. (2017), note that units were originally expressed as kcal/kg/K (whereby ∼837 J/kg/K becomes 0.20 kcal/kg/K and ∼921 J/kg/K becomes 0.22 kcal/kg/K). h See, for example: Keen and Lewis (1982) and McKenna and Sharp (1998). i Layer 41 is further subdivided into 15 sublayers, each 400 m thick. j Facies assignment is based on the observation that deposition in northern Gulf of Mexico minibasins tends to represent a fining-upward sequence whereby older sands were deposited during ponding of sediment within minibasins while younger shales were deposited during slope bypass (Prather et al., 1998), and is also based on the Green Canyon site-specific observation that above the sand-rich gas hydrate reservoir found here, reservoir quality decreases upward, grading into clay-dominated sediment Haines et al., 2017). k Sedimentation rates are compatible with observations constraining facies assignment, described above, as well as work suggesting that the Cenozoic sedimentation rate in the northern Gulf of Mexico has ranged anywhere from 30 m/Myr to 1 km/Myr (Kennett & Shackleton, 1975;Mello & Karner, 1996;Santschi & Rowe, 2008). l Horai (1971), Revil (2000), Burwicz et al. (2017), andPortnov et al. (2020). m Each layer is 30 m thick; each layer represents 100,000 years of time. n Facies assignment is based on the observation at Blake Ridge that, at least beneath ∼150 mbsf, sediment lithology is relatively uniform (Holbrook et al., 1996), and based on findings that at ODP Leg 164 Site 996, sediment TOC is relatively low (<0.5 wt. %) (Paull et al., 1996). o Sedimentation rate is unknown at this site (Paull et al., 1996). In our model, sediment thicknesses and ages define a sedimentation rate of ∼0.3 km/Myr in the upper ∼1 km of sediment, and a sedimentation rate of ∼0.35 km/Myr in the underlying ∼7 km of sediment. This is compatible with observations of sedimentation rates for moderately rapidly accumulating to rapidly accumulating hemipelagic sediments (Müller & Suess, 1979 observations that a majority of >120 salt diapirs in South Louisiana are conical (Jackson & Talbot, 1986) and that "upward-narrowing" diapirs dominate in passive margin settings (Koyi, 1998).
Various studies have successfully utilized basin modeling approaches to model salt evolution (e.g., Allwardt et al., 2009;Garcia et al., 2012;Gibson, 2012;Maystrenko et al., 2013;Mello & Henderson, 1997;Nelskamp et al., 2012). In terms of capturing salt movement through time, PetroMod™ software utilizes sequential backstripping to interpolate salt restoration timesteps in between a pre-defined present-day salt geometry (i.e., diapir geometry at 0 Ma) and a pre-salt-movement geometry (a flat-lying, undeformed source salt layer in both our synthetic and real-world models) with the assumption that salt volume is constant through time. This methodology need not discriminate between mechanisms for salt movement (i.e., differential loading-driven vs. buoyancy-driven salt ascension; e.g., Jackson & Talbot, 1986;Hudec & Jackson, 2007;Schultz-Ela et al., 1993), instead acting to capture the physical geometry of a deforming salt body at various timesteps.
Because rates of salt flow are extremely variable (e.g., Jackson & Talbot, 1986), we adopt a base case for our synthetic modeling efforts whereby salt ascends at a rate of ∼300 m/Myr, compatible with rates of 10 m to 2 km/ Myr (Jackson et al., 1994).

Heat Flow and Other Boundary Conditions
Because the major salt basins containing confirmed or inferred GH deposits occur in thermally subsiding (i.e., postrift) passive margin settings (Collett et al., 2015;Hudec & Jackson, 2007;Ruppel & Waite, 2020), we adopt heat flow values typical of such settings, which average 50 mW/m 2 with a typical range of ∼35-65 mW/m 2 (Allen & Allen, 2013). We also investigate the impact of variations in heat flow by testing the following basal heat flow scenarios: 35, 45, 50, 55, 65, and 80 mW/m 2 . Treatment of heat flow and the thermal modeling approach implemented within PetroMod™ are described in detail by previous workers (e.g., Baur et al., 2010;Hantschel & Kauerauf, 2009).
A constant sediment-water interface temperature (SWIT) of 4°C was used, as in Burwicz et al. (2017), and consistent, for instance, with estimates of Pliocene bottom-water temperatures at depths of ∼2.0-2.5 km (Dowsett et al., 2009). A constant temperature was assigned to negate the effects of changing SWIT on GH stability. Water depth is held constant at 2.2 km in all synthetic models described here, a simplification made to negate effects of changing water depth on GH stability.

Synthetic and Real-World Modeling Approach to Gas Generation, Migration, and Hydrate Formation
The synthetic and real-world models implement a similar approach to modeling biogenic gas generation, gas migration, and the GH formation and recycling process, as detailed below.

Organic Properties and Kinetics
We confine the synthetic and real-world models to the generation of biogenic gas; thus, the kinetics of Middelburg (1989)-which consider burial depth, sedimentation rate, and age of organic matter-are applied.  Horai (1971) and Revil (2000).
10.1029/2022GC010704 6 of 31 This kinetic rate law relates the reactivity of organic material to its depositional age. These kinetics have been implemented and validated in studies of GH systems, including PetroMod™-based studies (e.g., Burwicz et al., 2017;Kroeger et al., 2019;Piñero et al., 2016). In the synthetic models, a TOC (total organic carbon) of 1 wt. % and an HI (hydrogen index, used to calculate the generation potential of biogenic methane) of 100 mg HC/g TOC are used as the base case (Table 4), as in PetroMod™-based studies of GH by Kroeger et al. (2015) and Burwicz et al. (2017), and as is consistent with the average TOC of continental margin sediments (Emerson & Hedges, 1988), the general range of TOC for siliceous marine sediments (e.g., Ibach, 1982), and preliminary work on TOC of global hydrate-associated sediments (Burton, 2020). Organic properties used in the real-world models are listed and explained in Table 4.
It should be noted that while we do not consider thermogenic gas contributions in our synthetic or real-world modeling efforts, the integration of deeper thermogenic sources of gas could exert an effect on the formation of hydrate (and corresponding saturations and volumes), and in a limited number of locations has been suggested to represent an important contribution to the GH system (e.g., Brooks et al., 1984Brooks et al., , 1986Kida et al., 2006;Kroeger et al., 2015;Kvenvolden, 1995;Sun et al., 2020). However, GH at Green Canyon is predominantly biogenic (Burwicz et al., 2017;Moore et al., 2022), as is GH at Blake Ridge (Paull et al., 1996), and in general, most recovered samples of GH from global continental margins indicate biogenic rather than thermogenic sources (Chong et al., 2016;Koh et al., 2011;Kvenvolden, 1995Kvenvolden, , 1998Kvenvolden & Lorenson, 2001;Sloan, 2003).

Gas Hydrate Formation and Recycling
Several methods are available to simulate fluid flow and migration within PetroMod™. Here, free gas migration is modeled as two-phase Darcy flow. Cell-to-cell, pressure-gradient-driven Darcy flow is considered the best migration modeling approach for fine-grained and mixed (i.e., lower permeability) lithologies and for GH modeling (Baur & Katz, 2018;Hantschel & Kauerauf, 2009;Hantschel et al., 2000;Piñero et al., 2016). A relatively low critical gas saturation value (i.e., the saturation at which the gas phase becomes mobile; discussed in Hantschel and Kauerauf (2009) and Nole et al. (2018)) of 1 vol. % is assigned to enhance gas migration potential and to permit small gas bubbles to flow (as in Burwicz et al. (2017)), and methane is permitted to travel either as a free gas phase (according to Darcy's law) or in solution (according to Fick's law). The assumption of negligibly small critical gas saturation values is common practice (Hantschel & Kauerauf, 2009). It should be noted that modeled gas migration within the GHSZ is impacted by the presence of GH (e.g., Burwicz et al., 2017;Crutchley et al., 2017;Kroeger et al., 2022) because hydrate changes bulk sediment properties by reducing porosity, reducing permeability, and increasing capillary entry pressure (Liu & Flemings, 2007;Nimblett & Ruppel, 2003).
The PetroMod™ GH module integrates PetroMod™'s basin and migration modeling approach with low-temperature microbial methane generation kinetics (Middelburg, 1989) and the physical, thermodynamic, and kinetic properties of GHs to simulate hydrate formation and dissociation (Piñero et al., 2016). In the module, the formation and presence of GH is determined according to the equilibrium between methane dissolved in seawater (porewater) and methane hydrate (Piñero et al., 2016), with this equilibrium in turn governed by the dissociation pressure of methane hydrate in seawater (Tishchenko et al., 2005). GH formation from available  Middelburg (1989) a See text for explanation. b Consistent with regional geochemical work, and as modeled for Green Canyon by Burwicz et al. (2017). c Burwicz et al. (2017). d TOC contents of shallow sediments at the site are generally <0.5 wt. % but may be higher in deeper sediments, and are less than those at other Blake Ridge ODP sites (Paull et al., 1996). Nonetheless, we conservatively adopt a uniform sediment TOC content of 0.5 wt. %. e No data are available for HI values at this site, but HIs at other Blake Ridge sites are all >100 (Paull et al., 1996); thus, we adopt a uniform sediment HI of 100. methane is simulated for each cell and each time step by calculating the dissolution of methane in water and the dissociation pressure of methane hydrate depending on pressure and temperature (using the equations established by Tishchenko et al. (2005)). Within the GHSZ, GH forms whenever the saturation of methane is greater than the solubility of methane in aqueous solution and conversely dissociates when the solubility exceeds saturation (Tishchenko et al., 2005). Due to the large timesteps (i.e., >10,000 years) typically used in basin modeling, hydrate formation is assumed to be instantaneous rather than kinetically controlled (Burwicz et al., 2017). In other words, all methane that enters the GHSZ and exceeds the CH 4 solubility limit is immediately converted to hydrate (Burwicz et al., 2017). Here, the minimum gas saturation for hydrate formation is set at 1%, which allows a small portion of free and dissolved gas to be present in the GHSZ (a minimum gas saturation of 0% for hydrate formation, by contrast, would mean all methane entering the GHSZ forms hydrate, with no free gas able to occur). Relevant GH phase properties are provided in Table 5.
PetroMod™'s ability to simulate complex geological basin evolution (including burial and salt movement histories), hydrocarbon generation, multiphase flow, and the temperature-and pressure-dependent stability, formation, and dissociation of GH means that the software can be used to model GH recycling (You et al., 2019). During recycling, GH accumulations initially within the GHSZ are shifted downward, below the BGHSZ (i.e., the BGHSZ is shifted upward relative to the hydrate-bearing sediment), placing the hydrate under conditions at which it is no longer stable. The destabilized hydrate dissociates into water and free gas, and the liberated methane becomes available to migrate buoyantly upward and recrystallize GH at the new upward-shifted BGHSZ. Released gas may also be trapped as free gas beneath the BGHSZ, due to the sealing properties (i.e., capillary entry pressure effects; Liu & Flemings, 2007;Nimblett & Ruppel, 2003) of the overlying hydrate (e.g., Collett, 1993;Downey, 1984;Flemings et al., 2003;Grauls, 2001;Hornbach et al., 2004;Nimblett & Ruppel, 2003). PetroMod™ predicts temperature and pressure changes through time to model GHSZ evolution and can thereby capture hydrate recycling driven by numerous processes that might modify temperatureand pressure-dependent hydrate stability, including sediment burial-driven temperature changes (e.g., Haacke et al., 2007;Nole et al., 2018;Paull et al., 1994) as well as salt movement-driven temperature changes. This capability has been successfully employed by a limited number of studies (You & Flemings, 2021) that model sedimentation-driven (burial-driven) GH recycling and development of high-saturation hydrate deposits, including work on the Gulf of Mexico's Green Canyon area (Burwicz et al., 2017), work on two thrust ridges of New Zealand's Hikurangi Margin (Kroeger et al., 2022), and our preliminary work on the Gulf of Mexico's Terrebonne Basin (Dafov, 2021).
Porewater salinity is used by PetroMod™ in conjunction with pressure and temperature properties to calculate GH stability and solubility, in accordance with the equations set forth by Tishchenko et al. (2005). PetroMod™ assumes porewater salinity to be constant through time. Because processes are being modeled at the resolution of the model cell size (on the order of tens of meters), PetroMod™ does not account for microscale changes in porewater chemistry due to GH formation and dissociation (other than methane content) (Kroeger et al., 2015) or due to the presence of salt diapirs. For the synthetic models, a salinity of 35‰ (standard for seawater; e.g., Lyman & Fleming, 1940) was applied. Salinities applied to the real-world models are discussed below.

Real-World Modeling Approach
We construct two 2-D basin and GH system models based on known salt diapir-associated GH localities at Green Canyon (Gulf of Mexico) and Blake Ridge (offshore Carolinas).

Green Canyon
The Green Canyon area is located within the northern Gulf of Mexico salt province, and at the Green Canyon Block 955 site studied here, contains a prominent salt diapir (Portnov et al., 2020). GH occurs throughout the northern Gulf of Mexico, and has been recovered from multiple sites in the Green Canyon area (Brooks et al., 1986).
Geophysical, drilling, and previous modeling work in the area supply the inputs necessary for the construction of a 2-D model of the Green Canyon diapir and associated GH system. Green Canyon model dimensions are provided in Table 1. Green Canyon sediment characteristics and rock properties are listed in Table 2, while sediment organic properties and biogenic gas generation kinetics are listed in Table 4. Table 5 provides computational parameters for gas migration and hydrate formation.

of 31
We present model scenarios including the Green Canyon site salt diapir as well as a model scenario completely excluding the salt diapir (see below). As at the Blake Ridge site, it is noted that salt diapirism in the Green Canyon area causes doming of sediments, including expression as domes at the seafloor . The depth of the underlying salt layer at the site seems poorly constrained, as Burwicz et al. (2017) adopt a salt layer depth of ∼2,400 m below seafloor (mbsf) in their modeling effort, whereas Portnov et al. (2020) mention that the salt diapir has an apparent height of ∼6 km, suggesting an underlying source salt layer depth of ∼7,500 mbsf. Because the focus of our work is on salt diapir-related effects (and not on underlying source salt layer effects), we conservatively adopt a source salt layer depth of 7,500 mbsf. The age of this underlying salt layer is approximately Late Jurassic (∼145-164 Ma) (Salvador, 1987). Salt movement in the northern Gulf of Mexico is inferred to have initiated in the Mesozoic and early Cenozoic (Hudec et al., 2013), although movement specifically within the Green Canyon area is said to have initiated in Neogene times (Burwicz et al., 2017). We model salt movement starting at 20 Ma. Salt movement in the Green Canyon area has continued through recent times and is likely to be active today (McBride et al., 1998;Seni, 1992).
We run one salt diapir model scenario that includes a ∼30-m-thick sand-dominated layer at the approximate level of the BGHSZ and within the upper 1.5 km of shale-dominated sediment, consistent with drilling results and seismic stratigraphic analysis at this site (Boswell, Collett, et al., 2012;Collett et al., 2012;Haines et al., 2017;Meazell et al., 2020), and also run a model scenario that excludes this sand-dominated layer, so as to examine in relative isolation the potential lithological control of a sand-rich reservoir on GH saturations (a mechanism suggested for this site by Boswell, Collett, et al., 2012).
Both biogenic and thermogenic GHs have been recovered in the Green Canyon area (Brooks et al., 1986), though most hydrate is likely of biogenic origin (Burwicz et al., 2017;Flemings et al., 2020;Moore et al., 2022). Notably, BSR depth above the salt diapir is estimated at ∼450 mbsf (Boswell, Collett, et al., 2012), which is significantly (∼400 m) shallower than predicted based solely on standard temperature and pressure gradient assumptions (Portnov et al., 2020). GH saturations at the BGHSZ are as high as 90 vol. % in the domed sediments overlying the diapir, and average 50 to over 80 vol. % in the ∼30-m-thick GH reservoir identified here Flemings et al., 2020;Oti et al., 2022;Phillips et al., 2020), while saturations in the overlying sediments (183-266 mbsf) are estimated to be either negligible, based on P-wave velocity logs, or about 20 vol. %, based on a resistivity log . Burwicz et al. (2017) hypothesize that elevated GH saturations at the BGHSZ are in part due to recycling driven by high Neogene sedimentation rates, in conjunction with lithological control due to sandier reservoir sediments at this depth (e.g., Boswell, Collett, et al., 2012). Notably, it is inferred that a substantial amount of free gas is trapped beneath the BGHSZ by the overlying, relatively impermeable hydrate accumulation . Also of relevance to salt-influenced GH occurrence at this site is the postulation that continued salt diapir growth may continue to thin the GHSZ (Portnov et al., 2020). A water depth of 2,000 m is used for this site (Portnov et al., 2020), and is treated as constant through the time interval modeled here. We make this simplified assumption to negate the potential influence of changing hydrostatic pressure on the GHSZ. This approach of assuming a constant water depth follows the practice of Burwicz et al. (2017), who took a more extreme approach and modeled a relatively constant water depth for the Green Canyon area from 100 Ma to present. Nonetheless, we acknowledge that sea level has in fact been highly variable during the Cenozoic (Haq et al., 1987;Miller et al., 2005Miller et al., , 2020, including recent high-magnitude fluctuations in the northern Gulf of Mexico (e.g., Villamil et al., 1998;Wornardt & Vail, 1991). Water temperature at the seafloor is treated as 4°C throughout the modeled time interval (Burwicz et al., 2017;Portnov et al., 2020). Basal heat flow in the Green Canyon area is 42 mW/m 2 at present day (Christie & Nagihara, 2016), and heat flow is inferred to have decayed slightly from a value of ∼50 mW/m 2 at 100 Ma to the present-day value (Burwicz et al., 2017). Here, we conservatively treat basal heat flow as constant through time at 42 mW/m 2 . Geothermal gradients immediately above the diapir crest are over twice the regional geothermal gradient, and gradients in the overlying domed sediment at the level of the BSR are nearly twice the regional gradient (Portnov et al., 2020). Accordingly, heat flow in the sediments overlying the diapir crest up to the level of the BGHSZ is around twice that of the basal heat flow. Salinity at this site is conservatively assumed to be 35‰ (standard for seawater; e.g., Lyman & Fleming, 1940), consistent with values of 17-35‰ measured in the area (Portnov et al., 2020).

Blake Ridge Diapir
The Blake Ridge Diapir occurs as one of more than two dozen salt diapirs located in the Carolina Trough (Dillon et al., 1982). GH occurs in the sediments overlying the diapir, and was recovered from multiple holes drilled at Site 996 of ODP Leg 164 (Paull et al., 1996).
Drilling information and additional work at the site provide the suite of inputs necessary for the construction of a 2-D model of the Blake Ridge Diapir and associated GH system. Blake Ridge model dimensions are provided in Table 1. Blake Ridge sediment characteristics and rock properties are listed in Table 2, while sediment organic properties and biogenic gas generation kinetics are listed in Table 4. Computational parameters for gas migration and hydrate formation are provided in Table 5.
The diameter of the salt diapir itself is not fully constrained, but may be as much as ∼3.5 km based on seismic data presented in Taylor et al. (2000). We adopt a conservative diapir diameter of 2 km, as used by Hornbach et al. (2005). Similarly, the depth of the crest of the diapir is unknown but is modeled as 1 km beneath the seafloor by Hornbach et al. (2005) based on previous work suggesting that salt diapirs reach neutral buoyancy at ∼1 km depth (Talbot, 1993). We adopt this estimate, assuming a 1,000 mbsf diapir crest depth, though it bears mentioning that diapirs can rise above the level of neutral buoyancy due to differential loading of sediment (Hudec & Jackson, 2007;Jackson & Talbot, 1986;Schultz-Ela et al., 1993;Talbot, 1993). Diapirism causes doming of sediment at Blake Ridge, including expression as positive bathymetric relief on the seafloor at this site (Paull et al., 1996). The top of the diapir source (i.e., the underlying salt layer) is estimated at 8,000 mbsf (Paull et al., 1996), while the age of this salt is estimated to be Middle Jurassic (∼164-174 Ma) or older (Dillon et al., 1982). Here, we model salt movement beginning ∼23 Ma.
GH recovered at the site is of predominantly biogenic origin (Paull et al., 1996). Importantly, the BSR occurs at ∼400 mbsf away from the diapir, but shallows to ∼245 mbsf above the diapir (Hornbach et al., 2005). Taylor et al. (2000) note that salinity likely has little influence on this observed upwarping of the BGHSZ, and previous studies suggest that the upwarping is instead due to diapir-driven heat flow effects (Hornbach et al., 2005;Taylor et al., 2000). The BSR was not reached during drilling, and thus, concentrations at the BGHSZ are unknown (Paull et al., 1996). While GH and free gas saturations are not available from Site 996 of ODP Leg 164, estimations from nearby sites 994, 995, and 997 suggest hydrate saturations of ∼3%-8% of the pore space above the BSR, and free gas saturations of up to 14% of the pore space below the BSR (Lu & McMechan, 2002), although earlier work suggests that these hydrate and free gas values could be slightly higher .
A water depth of 2,160 m is adopted for the site (Hornbach et al., 2005;Paull et al., 1996). Water temperature at the seafloor is 3.5°C (Paull et al., 1996), and is treated as such throughout the modeled time interval. A heat flow of 40 mW/m 2 was assumed by Taylor et al. (2000) for sediments away from the diapir, while Hornbach et al. (2005) used a similar basal heat flow of 38 mW/m 2 . We use this basal heat flow of 38 mW/m 2 throughout the time interval modeled here. Importantly, Taylor et al. (2000) document a salt-influenced heat flow of 70 mW/m 2 in the sediments above the salt diapir and near the seafloor. A salinity of 35‰ is adopted, as in Taylor et al. (2000).

Model Limitations
Numerous assumptions and their attendant uncertainties are embedded within the basin and GH system modeling process, including certain computational limitations and simplifications (e.g., Burwicz et al., 2017;Hantschel & Kauerauf, 2009;You et al., 2019). Within the description of our synthetic and real-world modeling approaches, above, we have detailed numerous assumptions made for the purposes of this study. Basin and salt geometries are inherently simplified during the modeling process, and in the case of the real-world models described here, are subject to assumptions made due to lack of data, such as poorly constrained salt geometries and depths at Green Canyon (Burwicz et al., 2017;Portnov et al., 2020) and Blake Ridge (Hornbach et al., 2005;Taylor et al., 2000).
In the PetroMod™ backstripping-based interpolations of salt diapir geometries through time, we do not consider the possibility that the salt becomes emergent at the seafloor (e.g., Hudec & Jackson, 2007;Jackson et al., 1994;Schultz-Ela et al., 1993;Vendeville & Jackson, 1992). We do not consider the possibility that salt diapirism could be promoting faulting (e.g., Coleman et al., 2018;Jackson et al., 1994;Poliakov et al., 1996;Schultz-Ela et al., 1993;Taylor et al., 2000;Yin et al., 2009), and we do not consider the influence of faults (e.g., as permeable pathways for the enhanced migration of gas) on GH formation (e.g., Burton & Dafov, 2022;Collett, 1993;Hillman et al., 2020;Hustoft et al., 2007;Laberg & Andreassen, 1996;Tryon et al., 2002). Integration of faults could be important to future modeling efforts, as, for example, the complex network of faults at Green Canyon has been invoked as a possible migration mechanism to help explain elevated hydrate concentrations found there (e.g., Haines et al., 2017;Santra et al., 2022), and the potential importance of fault-mediated gas migration to the hydrate system has likewise been discussed for Blake Ridge (e.g., Gorman et al., 2002;Taylor et al., 2000;Wood & Ruppel, 2000). Relatedly, we do not investigate the possibility of advective heat transport by warm fluids moving along permeable faults, which can have detrimental effects on hydrate stability (e.g., Crutchley et al., 2014;Serié et al., 2012), including as a possible contributing factor to BGHSZ upwarping at Blake Ridge (Taylor et al., 2000). Importantly, we do not consider salt diapir-driven changes to porewater salinity, a limitation of PetroMod™ (You et al., 2019) that is unfortunate, as the impact of salinity on GH stability, such as inhibition of hydrate formation due to ions released into brines from dissolution of diapir-related salt, is well-established (e.g., Clennell et al., 1999;Dickens & Quinby-Hunt, 1997;Handa, 1990; Y. K. Li & Nghiem, 1986;Menten et al., 1981;Ruppel et al., 2005;You et al., 2015), and has also been invoked as a possible contributor to BGHSZ upwarping at Blake Ridge (Taylor et al., 2000), though not at Green Canyon, where we nonetheless assume a salinity at the uppermost limit of actual on-site measurements (Portnov et al., 2020). As discussed above, Petro-Mod™ does not account for microscale hydrate-or diapir-mediated changes to porewater chemistry, due to model cell resolutions (Kroeger et al., 2015). PetroMod™ does not model capillary effects on hydrate solubility and the hydrate phase boundary (You et al., 2019), which can be significant and have been invoked to help explain hydrate occurrence at Blake Ridge (e.g., Henry et al., 1999;Liu & Flemings, 2011), while capillary inhibition is unlikely in explaining reservoir-interval Green Canyon hydrate saturations (Daigle et al., 2022). GH modeling in Petro-Mod™ does not incorporate types of hydrate-forming gases other than methane (e.g., carbon dioxide, ethane, propane, and other C 2+ hydrocarbon gas molecules; e.g., Kvenvolden, 1995;Kvenvolden & Barnard, 1983;Milkov, 2005;Sloan, 1998), and the software is unable to simulate the formation of structure II or structure H GH, which are present in some parts of the Gulf of Mexico (e.g., Klapp et al., 2010;Sassen & MacDonald, 1994;Sassen et al., 2001). As discussed above, we focus on microbial gas generation, and do not model thermogenic gas generation. Anaerobic oxidation of methane is not modeled in PetroMod™, meaning hydrate saturations in the uppermost (seafloor to ∼50 mbsf) sediments are likely overestimated (Burwicz et al., 2017). As discussed above, PetroMod™ assumes that methane entering the GHSZ and exceeding the solubility limit is instantaneously converted to hydrate, and therefore does not consider the kinetically feasible coexistence of hydrate and free gas phases within the GHSZ (Burwicz et al., 2017) or, for instance, the migration of gas through the GHSZ, invoked at Blake Ridge (Gorman et al., 2002). We make various assumptions by assigning model boundary conditions, as detailed above, including the assumption of constant basal heat flows, sediment-water interface temperatures, and water depths through time, and we do not explicitly consider, for instance, impacts of sea-level fluctuations on variable delivery of organic matter or coarse-grained sediment to our real-world model areas (e.g., Blum & Hattier-Womack, 2009;Burton et al., 2023;Erbacher et al., 1996), though we do seek to replicate coringand geophysical data-based interpretations of stratigraphy and lithology, as detailed above.

Synthetic Models
We report results from synthetic modeling aimed at examining the influence of salt diapirism on GH formation by modeling variations in relevant parameters, including (a) basal heat flow, (b) salt diapir diameter, (c) salt stock heights, and (d) sediment thermal conductivities.

Variable Heat Flows
As anticipated, variations in basal heat flow exert a marked influence on the thickness of the GHSZ as well as on the nature of GH accumulations (Figures 2 and 3; Table 6; Figure S2 in Supporting Information S1).
For scenarios with modeled movement of a salt diapir, the GHSZ thins directly above the crest of the diapir from 890 m at ∼3.1 Ma to 725 m at present (∼19% reduction in thickness) for the 35 mW/m 2 heat flow scenario, from 495 to 410 m (∼17% reduction) for the 50 mW/m 2 scenario, and from 375 to 280 m (∼25% reduction) for the 65 mW/m 2 scenario ( Figure 2; Table 6; Figure S2 in Supporting Information S1). The GHSZ also thins at the edges of the model over the modeled time interval, but only by ∼6%-11% for the three aforementioned heat flow scenarios (Table 6).
By contrast, the scenarios excluding a salt diapir show only minimal reductions in GHSZ thickness over the same time interval, showing reductions in the range of ∼7%-12% at the center of the model and ∼0%-14% at the edges of the model (Figure 2; Table 6; Figure S3 in Supporting Information S1).
With-salt versus no-salt synthetic modeling scenarios (i.e., salt diapir-driven recycling vs. only sedimentation-/burial-driven hydrate recycling scenarios) reveal a significant difference in predicted GH satu-

Note.
GHSZ thickness is measured at the center of the models (i.e., directly above the salt diapir crest in with-salt scenarios) and at model edges.
rations: the with-salt 35 mW/m 2 scenario yields saturations of nearly 50 vol. % and both the 50 and 65 mW/m 2 scenarios yield saturations up to 100 vol. % of pore volume (Figures 2  and 3; Table 6), but all no-salt scenarios yield saturations no greater than 30 vol. % (Figures 2 and 4; Table 6). Saturations in the with-salt scenarios more than double from <20 vol. % at ∼3.1 Ma to >40 vol. % at present for the 35 mW/m 2 scenario, and markedly increase from a maximum of <70 vol. % and <55 vol. % at ∼3.1 Ma for the 50 and 65 mW/m 2 scenarios, respectively, to a maximum of 100 vol. % at present for both scenarios (Figures 2 and 3; Table 6). Saturations also markedly increase for 45, 55, and 80 mW/m 2 with-salt heat flow scenarios ( Figures S4-S6 in Supporting Information S1). By contrast, saturations in the no-salt scenarios are either unchanged or show only minor increases during this time (Figures 2 and 4; Table 6). Interestingly, maximum GH saturations in both the with-salt and no-salt scenarios increase by at most 6 vol. % per Myr between 3.1 Ma and 1.5 Ma, whereas the 1.5-0 Ma time period shows a marked contrast between with-salt (diapir-driven recycling) and no-salt (burial-driven recycling) scenarios: while the no-salt scenarios yield hydrate saturation increases of at most 7 vol. % per Myr (and in one case, decreases of 7 vol. % per Myr), the with-salt scenarios produce increases in GH saturation of 20-30 vol. % per Myr ( Figure 2).
Estimated GH volumes (calculated in PetroMod™ by multiplying 2-D model areas by 1 km to add a third dimension) are also markedly higher for modeling scenarios including the salt diapir versus those that exclude salt. GH volumes increase from 0.74 trillion cubic feet (TCF) at ∼3.1 Ma to 0.78 TCF at present (∼5% increase) for the 35 mW/m 2 with-salt scenario, from 0.61 to 0.90 TCF (∼32% increase) for 50 mW/m 2 , and from 0.57 to 0.70 TCF (∼19% increase) for 65 mW/m 2 . By contrast, the no-salt scenarios lack any general increasing trend, with decreasing total GH volumes in the 35 mW/m 2 scenario, constant volumes in the 65 mW/m 2 scenario, and an increase-but only from 0.06 to 0.07 TCF (<8% of the hydrate volume for the same heat flow scenario, but with salt)-in the 50 mW/m 2 scenario.
Estimated free gas volumes for the with-salt heat flow scenarios also far exceed the volumes for otherwise identical no-salt scenarios, though the increases in GH volumes observed across with-salt scenarios are not always captured in free gas volumes over the same time interval (however, free gas column heights directly above the diapir crest do in fact increase through time in all with-salt heat flow scenarios). For the 35 mW/m 2 withsalt scenario, free gas volume increases from 0 TCF to 0.05 TCF from ∼3.1 to 0 Ma and free gas column height increases from 0 to 30 m. Free gas volume increases from 1.30 to 1.37 TCF (∼5% increase) and free gas column height increases from 60 to 120 m for the 50 mW/m 2 scenario. Though free gas volume is consistently highest for the 65 mW/m 2 scenario, this volume does experience a decrease from 1.78 TCF at ∼3.1 Ma to 1.64 TCF at present (∼8% decrease), although free gas column height increases from 90 to 120 m during this same time. For all no-salt heat flow scenarios from ∼3.1 Ma to present, free gas accumulations are non-existent or negligible (0 TCF with a column height of 0 m).

Variable Salt Diapir Diameters
Interestingly, variations in salt diapir diameter seem to exert control on the efficacy with which diapirs cause recycling and concentration of GH. For the 1 and 2 km salt diapir diameter scenarios, saturations of GH exceeding 90 vol. % occur above the crest of the diapir (Figure 5). At diapir diameters of 4 km and wider, increasing diapir diameter has the effect of creating lower, more diffuse hydrate concentrations, such that for the maximum diameter tested (9 km), GH occurs at saturations restricted to <20 vol. % diffusely throughout the model ( Figure 5). This is perhaps explained by the salt-mediated heat flow being less focused or less spatially concentrated: the 2-km-wide diapir is most effective at channeling heat flow from its crest, and therefore most effective in creating the highest saturations of hydrate. At 9 km, heat flow radiates upward from the salt across the entire width of the model, preventing appreciable focusing and accumulation of GH.
As might be predicted, in the 1-8 km diapir diameter scenarios, the shape of the GHSZ corresponds to the shape of the top of the diapir, with the upwarped base of the GHSZ becoming increasingly wide from 1 to 8 km ( Figure 5). At the edges of the models in the 1-6 km diapir diameter scenarios, the depth of the BGHSZ is relatively constant at ∼3.2 km, but in the 8 and 9 km scenarios, the BGHSZ shallows to ∼3.1 and ∼2.95 km, respectively.

Variable Salt Stock Heights
For salt stock heights from 0 to 1 km, stock height exerts some influence on GHSZ thickness, whereby the GHSZ is somewhat thicker than for greater stock heights. For instance, GHSZ thickness is 300 m for the 0.25 and 0.5 km salt stock heights and 270 m for the 1 km stock height, but shrinks to 240 m for the 3, 5, and 7 km stock heights (for stock heights ≥3 km, changing stock height seems to exert no influence on GHSZ thickness).

Variable Sediment Thermal Conductivities
Unsurprisingly, increasing sediment thermal conductivity in the deeper silt to either side of the diapir (while holding shallow silt conductivity constant) leads to decreasing GHSZ thickness in the overlying sediment away from the salt diapir ( Figure S7 in Supporting Information S1). For instance, for a lower conductivity of 1.6 W/m/K, the GHSZ away from the salt diapir (i.e., toward the edges of the model) is 450 m thick, while it is 390 m thick for a conductivity of 2.0 W/m/K, and 360 m thick for conductivities of 2.4 and 2.8 W/m/K ( Figure S7 in Supporting Information S1). However, directly above the diapir crest, GHSZ thickness is actually greater (at 270 m) for the two higher deepsilt thermal conductivity scenarios of 2.4 and 2.8 W/m/K than for the 1.6 and 2.0 W/m/K scenarios (240 m). This is perhaps because the flux of heat entering the base of the model is impeded to a greater degree by lower thermal conductivities in the sediment, and is accordingly channeled through the highly conductive salt to a greater degree than if the sediment itself were more conductive.
By varying the thermal conductivity of the shallow silt (i.e., varying conductivities at depths shallower than the salt diapir crest, while holding deeper silt conductivity constant), we observe that higher shallow sediment thermal  Table 6. 10.1029/2022GC010704 13 of 31 conductivities lead to higher GHSZ thicknesses, both immediately above the salt crest and toward the edges of the model. For instance, GHSZ thickness increases from 240 m above the diapir crest and 360 m at the model edges for the 1.4 W/m/K scenario to 300 m above the crest and 630 m at the edges for the 2.6 W/m/K scenario ( Figure  S7 in Supporting Information S1). The corresponding GH stability zone simulations are displayed in Figure S2 of Supporting information S1.

Figure 4.
Model results (gas hydrate (GH) saturations) from identical scenarios as in Figure 3, but without inclusion of a salt diapir. The corresponding GH stability zone simulations are displayed in Figure S3 of Supporting information S1.

Green Canyon Model
Our Green Canyon model simulates salt movement concurrent with sedimentation from 20 Ma to present (Figure 6).
At present day, our model captures the influence of salt diapir-channeled heat on the distribution and thickness of the GHSZ (Figures 7a and 7b). We predict a BGHSZ depth of ∼470 mbsf directly above the diapir crest, compared to a BSR depth of ∼450 mbsf estimated by Boswell, Collett, et al. (2012), and we predict a BGHSZ depth of ∼860 mbsf in the sediment away from the diapir, compared to a depth of ∼850 mbsf predicted via temperature and pressure gradient assumptions by Portnov et al. (2020). By contrast, running a model scenario that excludes the diapir but is otherwise identical fails to accurately predict a GHSZ thickness compatible with the literature: the BGHSZ depth at the center of the domed sediment is overestimated by ∼470 m and the BGHSZ depth at the edges of the model is underestimated by ∼120 m (Figure 7c). This corresponds directly to temperature distributions associated with the no-salt scenario (Figure 7d), which are markedly different from temperature distributions in the with-salt scenario (Figure 7b). Our Green Canyon model predicts GH saturations ranging from ∼30 to 90 vol. %, and averaging ∼60 to >70 vol. %, at the depth of the sand-rich reservoir layer within the domed sediment (which coincides with the depth of the BGHSZ) (Figures 8 and 9), as compared to estimated average saturations of 50-80 vol. % and highs exceeding 90 vol. % for the reservoir Flemings et al., 2020;Oti et al., 2022;Phillips et al., 2020). We predict negligible hydrate saturations above this reservoir interval (Figures 8 and 9), consistent with P-wave velocity log interpretations . By contrast, performing an identical simulation of our Green Canyon model, but excluding salt, we find that hydrate saturations are completely underpredicted, with saturations reaching no greater than ∼30 vol. %, with any saturations >15 vol. % being extremely spatially limited, and with any >15 vol. % saturations occurring in the silty sediment ∼0.1-0.4 km beneath the sandy reservoir layer due to a BGHSZ that is deeper than in the with-salt scenario (Figures 7 and 9f). Notably, results from an identical simulation to the Green Canyon with-salt model, and merely excluding the ∼30-m-thick sandy reservoir interval, yield hydrate saturations up to ∼75 vol. % and averaging ∼50 vol. % at the BGHSZ ( Figure S9 in Supporting Information S1). Interestingly, our modeling of Green Canyon suggests that GH saturations in the with-salt, with-sand scenario have jumped from ∼15 to 60 vol. %, with an average of ∼40 vol. %, at ∼1 Ma to the present-day values of ∼30-90 vol. %, averaging ∼60 to >70 vol. % (Figures 8 and 9). By contrast, the hypothetical no-salt scenario actually suggests that maximum saturations have dropped from ∼40 vol. % at ∼1 Ma to ∼30 vol. % at present-day, and that saturations throughout the modeled section have dropped somewhat (Figure 9).
Estimations of GH volumes for both with-and without-salt scenarios add some nuance to the saturation results. Interestingly, present-day hydrate volumes are almost identical, and even slightly higher in the no-salt scenario, at 0.67 and 0.68 TCF for with-and without-salt, respectively. This can be attributed to the distribution of hydrate saturations predicted throughout each 2-D model transect. In the with-salt scenario, hydrate saturations are high and are essentially exclusively focused within the ∼30-m-thick sand-rich reservoir interval at the BGHSZ (Figure 9). In the no-salt scenario, hydrate saturations at the BGHSZ are much lower, but in the entire ∼500 m (at the center of the domed sediment) to ∼300 m (adjacent the domed sediment) thickness of sediment overlying the BGHSZ, hydrate saturations are ∼10-15 vol. % (whereas saturations in this sediment are ∼0 vol. % in the withsalt scenario) (Figure 9). Neither saturation scenario-whether ∼0 vol. % or ∼10-15 vol. % in sediment overlying the BGHSZ-can be excluded based on drilling data, as saturations in this overlying sediment are estimated to be either ∼0 vol. % (via P-wave velocity logs) or ∼20 vol. % (via a resistivity log) . Nonetheless, it is clear that even in the scenario whereby total hydrate volumes for with-and without-salt scenarios are equal for Green Canyon at present-day, accumulations of GH are much more concentrated for the with-salt scenario, and much more diffuse for the no-salt scenario (Figure 9). Interestingly, the with-salt scenario suggests the same trend of increasing hydrate volumes through time as is described for our synthetic modeling heat flow scenarios, whereby GH volume increases from 0.64 TCF at 0.98 Ma to 0.66 TCF at 0.16 Ma to 0.67 TCF at present-day (∼5% increase since ∼1 Ma). By contrast, the no-salt scenario shows the opposite trend, whereby GH volume decreases from 0.71 TCF at 0.98 Ma to 0.69 TCF at 0.16 Ma to 0.68 TCF at present (∼4% decrease since ∼1 Ma).
Notably, both free gas volumes and column heights increase over this time interval for the with-salt scenario (Figure 8), whereas neither >0-TCF volumes nor >0-m column heights are predicted for the no-salt scenario ( Figure S8 in Supporting Information S1). Our with-salt model therefore appears compatible with work by , suggesting that a significant amount of free gas is trapped beneath the hydrate reservoir at Green Canyon, whereas our no-salt model fails to predict free gas accumulation of any size. Free gas volumes predicted by the with-salt scenario increase from 1.44 TCF at 0.98 Ma to 1.57 TCF at present (∼9% increase), while column heights increase from 90 to 120 m during the same period.

Blake Ridge Model
We model salt movement from 23 Ma to present in our Blake Ridge Diapir model ( Figure S10 in Supporting Information S1).
Our model successfully captures the diapir-related upwarping of the BGHSZ ( Figure S11a in Supporting Information S1) previously described at Blake Ridge (Hornbach et al., 2005;Taylor et al., 2000). We predict a BGHSZ depth of ∼240 mbsf directly above the salt diapir crest (and predict the GHSZ here has thinned by ∼16% since 0.65 Ma), compared to a BSR depth of ∼245 mbsf given by Hornbach et al. (2005), and we predict a BGHSZ depth of ∼390 mbsf in the sediment toward the edges of the model (and predict the GHSZ here has thinned by ∼6% since 0.65 Ma), compared to a BSR depth of ∼400 mbsf given by Hornbach et al. (2005). As with the Green Canyon model, running a model scenario that excludes the salt diapir but is otherwise identical fails to predict a GHSZ thickness consistent with the literature: BGHSZ depth at the center of the domed sediment is predicted as ∼360 m (unchanged since 0.65 Ma) and BGHSZ depth at the edges of the model is predicted as ∼330 m (also unchanged since 0.65 Ma) ( Figure S11c in Supporting Information S1). As at Green Canyon, this is due to the influence of the diapir on temperature distributions ( Figure S11b vs. Figure S11d in Supporting Information S1).
Predicted thinning of the GHSZ (i.e., upwarping of the BGHSZ) above the salt crest as it ascends also corresponds to a shallowing of the top of the single GH reservoir (>20 vol. % saturation) layer through time in the with-salt scenario, whereby the shallowest portion of the reservoir progressively migrates upward through the sediment column from ∼2.43 km depth at 0.65 Ma to ∼2.41 km depth at 0.33 Ma to ∼2.39 km depth at present-day (Figure 10a-10c). By contrast, the no-salt scenario-where no GHSZ thickness change is observed since 0.65 Ma-shows a potentially increasing-depth trend of the shallowest portion of the primary (>20 vol. % saturation) GH reservoir layer, whereby this depth migrates downward through the sediment column from ∼2.47 km at 0.65 Ma to ∼2.50 km at 0.33 Ma to ∼2.50-2.51 km at present-day, though present-day also includes a secondary (>15 vol. % saturation) reservoir layer with the shallowest depth at ∼2.44 km (Figure 10d-10f). Furthermore, at present, the single >20 vol. % reservoir layer in the with-salt scenario has expanded laterally to greater than twice the horizontal extent of the previous two timesteps (>2-km extent at present-day vs. ∼1-km extent at 0.65 Ma and 0.33 Ma), whereas the no-salt scenario sees an evolution from two >10 vol. % saturation reservoirs at 0.65 Ma to one such reservoir at 0.33 Ma to two such reservoirs at present-day, and also sees no change in horizontal extent of any of these reservoir layers, which stay constant at ∼1-km widths ( Figure 10).
The greater areal extent of the with-salt, single reservoir layer at present-day corresponds to a slightly higher (albeit quite low in comparison to the Green Canyon model) total GH volume of 0.07 TCF versus 0.06 TCF for the no-salt, multiple-reservoir layers. Both the with-salt and no-salt hydrate volumes have increased by 0.01 TCF since 0.65 Ma, representing a ∼17% increase for the with-salt scenario and a 20% increase for the no-salt scenario.
Also of potential relevance is the observation that only in the with-salt scenario are modeled free gas volumes and column heights greater than zero at present-day. Nonetheless, this with-salt free gas volume (at 0.03 TCF with a maximum saturation <30 vol. %) and column height (at 30 m, or the minimum possible layer thickness) are both relatively minimal ( Figure S12 in Supporting Information S1). Free gas volumes and column heights are 0 TCF and 0 m, respectively, for with-salt timesteps at 0.65 and 0.33 Ma (indicating an increase to present-day values) and for all no-salt timesteps (Figures S12 and S13 in Supporting Information S1). Figure 11. Conceptual illustrations of sedimentation-/burial-driven gas hydrate (GH) recycling versus salt diapir-driven hydrate recycling. During sedimentation-driven recycling, hydrate-bearing sediment is buried below the base of the hydrate stability zone (BGHSZ), destabilizing and dissociating the hydrate into free gas and water. This free gas may then migrate upward and be reincorporated as higher-concentration hydrate at the new BGHSZ. During diapir-driven recycling, the ascent of the salt diapir causes progressive upwarping of the BGHSZ. As the BGHSZ is pushed above hydrate-bearing sediment, the hydrate decomposes, releasing free gas that may be recycled into the new shallower BGHSZ as hydrate. Free gas may preferentially migrate along and toward the apex of sediment domed/folded by diapir ascent, promoting both formation of high-saturation hydrate and accumulation of a free gas column beneath the recycled high-saturation GH seal. Note that both sediment burial-driven and salt diapir-driven recycling, as illustrated here, involve temperature-driven hydrate destabilization. Abbreviations and symbols: GH, gas hydrate; sat., saturation; sed., sediment; T, temperature; t 2 , time at present (on left, present-day seafloor/sediment surface; on right, present-day salt diapir crest depth); t 1 and t 0 , time intervals in the past, with t 0 being the oldest (indicating older, buried sediment surfaces on the left, and previous, deeper diapir crest depths on the right).
Though yielding quite low GH and free gas saturations for the with-salt scenario (especially compared to Green Canyon) and low hydrate saturations for the no-salt scenario, our model estimates (at 10-20 vol. % or more maximum hydrate saturation for with-and without-salt scenarios) are at the upper end of previous estimates for nearby sites at Blake Ridge, which postulated ∼3-8 vol. % (Lu & McMechan, 2002) or possibly slightly higher  hydrate saturations above the BSR. However, because the BSR was not encountered during drilling at the Blake Ridge Diapir site, concentrations at the BGHSZ (such as those predicted in the with-salt scenario modeled here) are unknown, as are potential underlying free gas saturations (Paull et al., 1996). Estimates at nearby sites, also performed by Lu and McMechan (2002), approximate free gas saturations up to 14 vol. % below the BSR (about half the value of the maximum saturations predicted in our with-salt model, but higher than the 0 vol. % saturations predicted in our no-salt model), although, as with hydrate saturations, earlier work by  suggested slightly higher free gas saturations might be present at these adjacent sites. In any case, it is of note that only the with-salt scenario produces any free gas accumulation, which is more consistent with the inference that, however minimal, free gas accumulations are present at at least some of the Blake Ridge sites.

Salt Diapir-Driven Gas Hydrate Recycling Demonstrated via Synthetic and Real-World Models
Through numerous synthetic basin and GH system models as well as real-world models from Green Canyon and Blake Ridge, we confirm that salt diapir movement causes an upwarping of the BGHSZ and an overall thinning of the GHSZ through time, as demonstrated in numerous previous studies (e.g., Chand et al., 2008;Hornbach et al., 2005;Portnov et al., 2020;Taylor et al., 2000). However, while previous work has described the influence of salt diapirism as a cumulatively destructive mechanism within the context of GH systems (e.g., Portnov et al., 2020;Ruppel et al., 2005), we show through our combined synthetic and real-world modeling that salt diapirs can serve as a constructive mechanism by promoting GH recycling (Figures 2  and 11).
Previous work demonstrates recycling as a mechanism for concentrating (i.e., increasing through time) GH saturations at the BGHSZ (e.g., Burwicz et al., 2017;Haacke et al., 2008;Paull et al., 1994;Schmidt et al., 2022;You & Flemings, 2021). Here, we present a newly described mechanism for GH recycling: salt diapirism. We show that the movement of salt diapirs and resultant changes to local heat flows lead to GH recycling and produces elevated saturations and volumes of GH at the BGHSZ and elevated volumes of free gas beneath the BGHSZ ( Figure 11). We demonstrate that this salt diapir-driven GH recycling is a mechanism by which saturations of hydrate accumulations at the BGHSZ may reach >90 vol. % (vs. saturations limited to <30 vol. % in synthetic scenarios that exclude a salt diapir) and by which volumes of both GH at and free gas beneath the BGHSZ may be significantly elevated through time. Notably, a comparison of with-salt model scenarios (i.e., salt diapir-driven hydrate recycling scenarios) against no-salt model scenarios (i.e., burial-driven hydrate recycling scenarios, in which recycling occurs at a background rate paced by sedimentation) indicates a significant contrast between the rates of hydrate recycling in each scenario (Figure 2). In our synthetic models, salt diapir-driven hydrate recycling increases maximum GH saturations by 20-30 vol. % per Myr, whereas solely sedimentation-driven recycling increases maximum saturations by no more than 7 vol. % per Myr since 1.5 Ma (Figure 2).
By examining the influence of salt diapir-mediated temperature changes on the GH recycling process, this work represents an investigation into the poorly understood (Schmidt et al., 2022) influence of local GHSZ perturbations on the recycling process. Furthermore, the demonstration of salt diapir-driven hydrate recycling provides an effective recycling mechanism that may lead to locally elevated hydrate saturations on old, rifted continental margins (i.e., where many major global salt basins are located; Hudec & Jackson, 2007) where rates of recycling due to non-diapir-related processes are otherwise thought to be low (Haacke et al., 2007(Haacke et al., , 2008.

Salt Diapir-Driven Gas Hydrate Recycling Can Explain Observations at Green Canyon
A close agreement between our Green Canyon model results (GHSZ thicknesses) and geophysical observations (BSR depths from seismic data; Boswell, Collett, et al., 2012) and between our model results (GH saturations, and presence of free gas accumulations) and logging-while-drilling data (estimates of GH saturations from P-wave velocities and a resistivity log, and the inference of a substantial sub-hydrate free gas accumulation; Collett et al., 2012;, as well as a close adherence to model input data available from previous work in the region and at the Green Canyon diapir modeled here, lend confidence to the results and interpretations described herein. We show that lithological control on observed GH saturations (Boswell, Collett, et al., 2012;Santra et al., 2022) is important-here, the with-salt, with-sand scenario (Figure 9) is estimated to account for hydrate saturations elevated by ∼10-15 vol. % beyond the with-salt, no-sand scenario ( Figure S9 in Supporting Information S1)but that salt diapir-driven recycling may be a more important control in the case of the Green Canyon diapir: we show that the diapir may account for saturations elevated by ∼35-45 vol. % (derived by comparing the with-salt, no-sand scenario with the no-salt, with-sand scenario; Figure 9 and Figure S9 in Supporting Information S1). In our modeling work presented here, it is only by combining the effects of lithological control (Boswell, Collett, et al., 2012) and salt-driven recycling control (this study)-and not by considering either control in isolationthat our Green Canyon model accurately predicts the average and maximum GH saturations estimated from logging data during previous studies of the site. This said, there are various processes in addition to lithological control and salt diapirism that could help explain observed present-day hydrate saturations at Green Canyon, and while our modeling does account for the process of sedimentation-driven recycling invoked for the site (Burwicz et al., 2017), we do not, for example, investigate the possible role of faults (e.g., Haines et al., 2017;Santra et al., 2022). Overall, however, recycling is acknowledged as a plausible explanation for observed elevated hydrate saturations at Green Canyon Wei et al., 2022), and may be more compatible with findings of predominantly biogenic gas at the site (Burwicz et al., 2017;Flemings et al., 2020;Moore et al., 2022) than fault-mediated gas migration (as noted by Flemings et al., 2022).
We also find that sedimentation-driven recycling (e.g., Figure 11) need not fully explain the elevated hydrate saturations observed within the sandy reservoir interval here (e.g., Burwicz et al., 2017). We arrive at this conclusion through the consideration of a Green Canyon model that excluded salt but would be able to capture sedimentation-driven recycling (the "no-salt, with-sand" scenario; Figure 9), in which average predicted reservoir-interval saturations were at least ∼35% lower and maximum reservoir-interval saturations were at least ∼60% lower than saturations estimated from logging data by previous workers. Although the influence of sedimentation-driven recycling could potentially be enhanced by very high sedimentation rates (i.e., greater than the ∼0.3-0.4 km/Myr modeled here for Green Canyon), the influence of heat flow-driven (i.e., salt diapir-driven) recycling seems inextricable from the case of Green Canyon GH saturations. Furthermore, the influence that diapir movement has on the observed doming of overlying sediment at this locality, which-as demonstrated by our models-controls focused gas migration into the GHSZ (this focusing is also illustrated by Flemings et al., 2022), and without which hydrate saturations would be much more diffuse across the ∼10-km-wide modeled transects, must be considered. This diapir-driven sediment folding and focused gas migration may provide a mechanism for overcoming the initial hydrate saturation needed to initiate an effective recycling process at the BGHSZ (Burwicz et al., 2017). Finally, failure to consider salt influence (i.e., the "no-salt, with-sand" model scenario; Figure 9) also means that the significant sub-hydrate free gas accumulation interpreted by  for this site is completely missed, with neither >0 TCF free gas accumulations nor >0-m column heights produced by this no-salt scenario.

Salt Diapir-Driven Gas Hydrate Recycling May Explain Some Observations at Blake Ridge
For the with-salt scenario, progressive upward migration of the >20 vol. % saturation GH reservoir layer from 0.65 Ma to present (Figures 10a-10c) is consistent with GHSZ thinning due to upward migration of the salt diapir crest and is consistent with repeated BGHSZ upwarping-driven disassociation of GH into free gas and re-accumulation as hydrate at stratigraphically shallower depths (i.e., GH recycling). Additionally, the consistent localization of any appreciable (>10 vol. % saturation) amount of hydrate into this single hydrate reservoir layer through the three timesteps examined for the with-salt scenario (as reflected by the doubling in horizontal extent of this single reservoir from ∼1 km at 0.65 Ma to >2 km at present-day) versus the somewhat more diffuse, dispersed distribution of hydrate-bearing reservoir (>10 vol. % saturation) layers in the no-salt scenario (and as reflected, by contrast, in the unchanging ∼1-km extent of the various reservoirs) at various timesteps is consistent with recycling in the with-salt scenario driving progressive focusing of and concentration of GH accumulations specifically at the BGHSZ, rather than a lack of recycling leading to a less focused distribution of hydrate saturations throughout the GHSZ as in the no-salt scenario (Figure 10a). The observation of predicted present-day free gas accumulation (albeit small in volume) in the with-salt scenario ( Figure S12 in Supporting Information S1) but the lack of any free gas accumulation in the no-salt scenario ( Figure S13 in Supporting Information S1) may provide additional evidence for some amount of salt diapir-driven recycling and locally elevated hydrocarbon saturations at and below the BGHSZ at the Blake Ridge Diapir site.
This said, for the no-salt scenario, the observed secondary (>15 vol. % saturation) GH reservoir layer with apex at a shallower, ∼2.44 km depth may indicate some degree of sedimentation-driven (i.e., burial-driven) recycling (Figure 10f) (e.g., as discussed by Burwicz and Rüpke (2019)), which potentially explains the downward migration (burial?) of the primary (>20 vol. % saturation) hydrate reservoir since 0.65 Ma (Figure 10d-10f), and may partially explain the observed increase in no-salt GH volume from 0.65 Ma, though it should be noted that these sorts of relatively small magnitude increases in GH saturation and volume through time can also be products simply of continued biogenic gas generation, migration, and accumulation in the GHSZ.
In general, previous estimates of <10 vol. % hydrate saturations and <20 vol. % free gas saturations at Blake Ridge GH sites Lu & McMechan, 2002) and our estimates of maximum hydrate saturations of ∼10-20 vol. % (for both with-and without-salt scenarios) and free gas saturations no greater than ∼30 vol. % (for the with-salt scenario) as well as our estimates of present-day GH volumes of 0.07 TCF or less and free gas volumes of 0.03 or 0 TCF for both scenarios suggest that Blake Ridge Diapir may not be of particularly high resource prospectivity, with neither compelling unconventional nor conventional hydrate-associated natural gas potential.

Implications of Salt Diapir-Driven Gas Hydrate Recycling for Resource Exploration
Given particular relevance to both conventional and unconventional resource exploration, it bears mentioning again that, through numerous synthetic and real-world basin and GH system models, we demonstrate that salt diapir-driven GH recycling is a mechanism by which saturations and volumes of hydrate accumulations near the BGHSZ (as an unconventional exploration target) can be significantly elevated beyond a no-salt baseline (with saturations attaining >90 vol. %) and by which volumes of free gas trapped beneath the BGHSZ (as a conventional exploration target) may likewise be significantly elevated. With regard to resource accumulation, we demonstrate through synthetic models and Green Canyon and Blake Ridge models that GH can act as a feedstock in producing appreciable free gas accumulations via salt-driven decomposition and recycling, and may concurrently act as a seal (in the classical petroleum system sense; Magoon & Dow, 1994;Max & Johnson, 2014) to allow build-up of an underlying free gas column ( Figure 11). Due to the particularly elevated saturations that may be found within salt diapir-associated recycled GH (as demonstrated here), salt-associated recycled hydrate near the BGHSZ may represent a sealing element of particularly high integrity (at least within the context of considering GH as a potential seal). This inference is supported by the finding that throughout our synthetic and real-world modeling efforts, predicted free gas columns were consistently and exclusively observed in connection with modeling scenarios that included a salt diapir and that showed some amount of inferred GH recycling near the BGHSZ.
Because of elevated hydrate saturations and volumes and elevated free gas volumes and column heights associated with salt diapir-driven recycled hydrates, we suggest that salt diapir-associated hydrate deposits and free gas accumulations represent an attractive exploration target for future marine drilling efforts. This is relevant both to industry interests from a resource exploitation perspective, and to scientific and academic interests, particularly for real-world, targeted testing of the salt diapir-driven GH recycling hypothesis presented here. Interestingly, our scenario-testing of variable salt diapir diameters seems to suggest there is an ideal diapir diameter (<4 km, and perhaps most ideally ∼2 km) for promoting recycling, which has implications both in helping guide hydrate-related exploration and for the Green Canyon and Blake Ridge diapirs discussed here, both of which fall within this ideal diapir diameter range (with Green Canyon's diapir width estimated at 2-3 km by Portnov et al. (2020), and Blake Ridge Diapir's width estimated at 2 km by Hornbach et al. (2005), or a maximum of ∼3.5 km by Taylor et al. (2000)).
Our study has particular implications for major salt basins with known or inferred hydrate occurrences, such as basins of the Gulf of Mexico, off southeastern Brazil, and off the west coast of Africa (Collett et al., 2015;Hudec & Jackson, 2007;Ruppel & Waite, 2020). These linked salt-GH systems represent particularly attractive localities for examining potential salt-driven hydrate recycling, and for associated future resource exploration.

Conclusions
We use synthetic basin and GH system models and real-world models to demonstrate a salt diapir-driven mechanism for the recycling of GH (Figure 11). This salt diapir-driven GH recycling serves as a mechanism that can elevate hydrate saturations above 90 vol. % and can significantly elevate free gas saturations and volumes. Comparison of salt diapir-driven recycling and sediment burial-driven recycling suggests markedly higher rates of hydrate recycling in diapir-driven versus burial-driven scenarios (e.g., respective hydrate saturation increases of 20-30 vol. % per Myr vs. less than 7 vol. % per Myr in our synthetic models since 1.5 Ma; Figure 2). Using case studies of Green Canyon and Blake Bridge, we demonstrate how GH recycling can help explain concentrations of GH observed during drilling. Overall, GH recycling due to salt diapir activity implies that hydrate and free gas above salt diapir crests may represent attractive targets for scientific and academic drilling as well as for unconventional and conventional hydrocarbon resource exploration.