By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 Before being disrupted by a magmatic event in 1999, the vent temperatures and salinities along the axis of the Main Endeavour Field on the Juan de Fuca Ridge exhibited a quasi-steady spatial gradient in which the southern vent fluids were hotter and less saline than the northern vent fluids. We present 2-D numerical models of two phase flow in a NaCl-H2O system to understand these gradients. We consider homogenous permeability models with a range of bottom boundary temperature distributions and heterogeneous permeability models by imposing layer 2A extrusives with a constant bottom boundary temperature distribution. The aim is to understand the impact of both bottom boundary temperature and layer 2A permeability on hydrothermal fluids and to determine what combination of these controlling factors could cause the observed trend. We find that variations in bottom boundary temperature alone cannot explain the span of surface temperatures and salinities measured at the Main Endeavour Field. Heterogeneous permeability within layer 2A that has higher overall permeability in the northern part of the vent field than the southern part can reproduce the observed north to south temperature gradient, but such a permeability distribution cannot reproduce the observed salinity gradient. We conclude that both deep-seated heterogeneous permeability, perhaps localized by a fault zone, and a heterogeneous layer 2A are required to produce the trend of temperatures and salinities in vent fluids at the Main Endeavour Field prior to the 1999 event.
 Hydrothermal systems associated with oceanic spreading centers account for a quarter of Earth's total heat flux and one third of the heat flux through the ocean floor [Stein and Stein, 1994; Elderfield and Schultz, 1996]. Circulation of seawater through such systems alters both the crust and the circulating fluid, impacting global geochemical cycles [Wolery and Sleep, 1976]. The warm vent fluids rich in nutrients support a wide variety of unique biological communities [Kelley et al., 2002]. Thus, understanding hydrothermal processes at oceanic spreading centers is important to provide insight into thermal and biogeochemical processes.
 As seawater circulates through oceanic crust at spreading centers, heat transfer from the subaxial magma chamber results in fluids exiting the seafloor as “black smokers” with temperatures ranging between 300 and more than 400°C [e.g., Spiess et al., 1980; Kochinsky et al., 2008]. Moreover, the salinity of fluids at these high-temperature vents is often different from seawater, which is generally thought to indicate subsurface boiling and phase separation [Bischoff and Pitzer, 1989; Von Damm, 1990]. For pure water, boiling occurs along a pressure-temperature curve on which liquid and vapor coexist. In seawater, however, the presence of dissolved salts means that boiling occurs in a region of pressure-temperature-salinity space. Because seawater is well represented by a NaCl-H2O solution of concentration 3.2 wt % [Bischoff and Pitzer, 1989], a basic understanding of boiling and phase separation in seawater hydrothermal systems can be gleaned by investigating two-phase flow and phase separation in a NaCl-H2O fluid. When phase separation occurs in a seafloor hydrothermal system, the more saline, denser brine tends to sink to the bottom of the system and the less saline vapor phase tends to rise. If the vapor phase vents on the surface, with or without mixing with seawater, the salinity of the venting fluid is less than seawater. In some cases vent fluids with salinity higher than seawater have also been observed [e.g., Butterfield et al., 1990; Butterfield and Massoth, 1994], which indicates that phase separated brine may also ascend and mix with seawater before venting. We call these fluid mixtures vapor-derived (VDF) or brine-derived (BDF) fluids, respectively.
 Numerical modeling of two phase flow in a NaCl-H2O fluid is a challenging task. The physical properties of brine and vapor change rapidly as functions of temperature and pressure. The partitioning of salt and heat between the two phases complicates the equations of state relating the temperature, pressure, and salinity (T-P-X) to fluid properties such as the density and enthalpy. Earlier attempts at numerical modeling of seafloor hydrothermal systems were, therefore, simplified, most commonly by either using single phase model [e.g., Rosenberg et al., 1993; Wilcock, 1997; Cherkaoui et al., 1997; Cherkaoui and Wilcock, 1999; Lowell et al., 2007] or by considering two-phase flow using pure H2O as the hydrothermal fluid [e.g., Lowell and Xu, 2000; Coumou et al., 2006]. Lowell and Germanovich  and Fontaine and Wilcock  discuss brine formation and storage in seafloor hydrothermal environments, whereas Fontaine et al.  attempt to infer implications for a two phase hydrothermal field while using a single phase model with variable density.
 The clear evidence of phase separation at mid-ocean ridges and its impact on fluid properties on the surface has recently led to hydrothermal models assuming a NaCl-H2O fluid with realistic thermodynamic properties in the two phase region [Coumou et al., 2009; Lewis and Lowell, 2009a, 2009b; Driesner, 2010; Han et al., 2013]. Han et al.  use the numerical modeling code FISHES (Fully Implicit Seafloor Hydrothermal Event Simulator [Lewis and Lowell, 2009a, 2009b] to investigate the relationship between surface temperatures and salinities and bottom boundary temperatures for different homogenous permeability models. These previous models of phase separation in a NaCl-H2O fluid have not been applied to a specific seafloor system or set of observations, however. Therefore, a natural extension of previous work is to apply such modeling to a particular vent field.
 In this paper, we have used numerical code FISHES to construct models of phase separation within the Main Endeavour Field (MEF) on the Juan de Fuca Ridge (JDFR). In particular, we attempt to explain the observed quasi-steady north to south gradients in salinity and temperature of the hydrothermal vent fluids [Butterfield et al., 1994] that existed between 1984 and the noneruptive magmatic event in 1999 [Johnson, 2000; Lilley et al., 2003, Figure 1].
 Our models, while not addressing all aspects of the hydrothermal circulation system, provide significant insight into the effects of bottom boundary temperature and crustal permeability structure on temperature and salinity of the vent fluid as well as on fluid mixing between deep seated, high-temperature circulation and low temperature circulation in layer 2A. In the next section, we describe the study area and the principal field data used to constrain the models. Then we present the numerical modeling code FISHES, the model setup and results, and finally, the implications of the model results. We end by offering some concluding remarks.
2. Main Endeavour Field
 The MEF located at 57°47′N on the JDFR is one of five high-temperature hydrothermal fields on the Endeavour Segment. Each of these vent fields is several hundred meters in length and separated by a few kilometers; they are characterized by a number of large sulfide edifices and areas of low-temperature diffuse flow [Kelley et al., 2002]. The MEF has had a history of study that spans nearly two decades [Kelley et al., 2012 and references therein]. Before it got significantly perturbed by 1999–2000 magmatic events [Johnson et al., 2000; Davis et al., 2001; Lilley et al., 2003], the MEF had been relatively stable since its discovery in 1982 [Delaney et al., 1992; Kelley et al., 2002]. The consistency of high temperature, low salinity fluids at MEF and prolonged scientific interest in the area make it a type-setting for mid-ocean ridge hydrothermal studies.
 In order to develop a numerical model of two-phase flow and phase separation at MEF, we incorporate a number of observational constraints. These data include seafloor bathymetry data, seismic data revealing important features of the subaxial magma chamber (AMC) and structure of the extrusive layer 2A [Van Ark et al., 2007], time series measurements of vent fluid temperature (T), salinity (X) and chemistry [Butterfield et al., 1994], measurements of hydrothermal heat output [Bemis et al., 1993; Ginster et al., 1994; Veirs et al., 2006; Johnson et al., 2010], and integrated thermal blanket heat flow data [Johnson et al., 2010].
2.1. Bathymetry and Structure
 Figure 2 [Kelley et al., 2012] shows that the MEF, which is located near the western margin of the median valley, is situated at seafloor depths between approximately 2000 and 2200 m. The MEF, along with other hydrothermal systems on the Endeavour Segment, are localized by major faults [Wilcock and Delaney, 1996], the existence of which may be partially related to episodic dike emplacement [Carbotte et al., 2006].
 Seismic data [Van Ark et al., 2007] indicate that layer 2A (identified as extrusive basalts) is variable in the neighborhood of the MEF with a mean thickness of 460 ±160 m. The data also show a subaxial magma chamber beneath the MEF at a depth ranging between 2 and 2.5 km with a width of 0.8 km, dipping to the east (Figure 3). As discussed below, Johnson et al.  suggest that this dip controls hydrothermal flow paths and gives rise to an east-west gradient in conductive heat flow.
2.2. Thermal Data
 Data collected in 1984, 1987, and 1988 from the MEF [Butterfield et al., 1994] show a significant gradient in salinity of the vent fluid, with the lowest salinity fluids in the southwest approximately half of that of the fluids in the northeast (Figure 1). Although there is more variability as a function of time, the temperatures of the vent fluids show a complementary trend, in which vent fluid temperatures were highest in the southwest and lowest in the northeast (Figure 1). The low salinity of the vent fluid across the MEF indicates the occurrence of phase separation and loss of brine phase below the seafloor [Butterfield et al., 1994]. This characteristic fluid salinity varied very little from year to year but changed dramatically after a diking event in 1999 [Lilley et al., 2003]. Following this event, the vent temperatures in the southern part of the field (∼ 380°C prior to the event) have declined and salinity of fluids (< 2.5 wt% seawater) has risen back toward seawater values (Figure 1).
 Heat flux at the MEF was first measured by Bemis et al.  and Ginster et al. . According to their estimates, heat output from focused venting on MEF was ∼ 200 to 350 MW. Veirset al.  conducted a hydrographic study on the MEF to show that the total heat output from the field is partitioned nearly equally between focused and diffused flow and estimated the total heat output to be ∼ 500 MW. Using a mean heat output of 450 MW derived from a number of studies [Baker, 2007], a mean vent temperature of 365°C, and a vent field area of 106 m2, Lowell et al. , estimate the permeability of the discharge zone to be 2 × 10−13 m2. Consequently, we have used a value of 10−13 m2 in our homogenous permeability models.
Johnson et al.  use thermal blankets to measure conductive heat flow on the Endeavour Segment. They interpret the low heat flow sites on the valley boundary wall directly east of the MEF to indicate a pervasive east-west two-dimensional cross-valley flow, with seawater recharge on the eastern valley wall and fluid discharge near the western wall of the axial valley (Figure 4).
 Although hydrothermal circulation is likely to be three-dimensional, we use the thermal blanket data as a first-order justification to consider a 2-D across-axis model for the MEF hydrothermal system. As described in section 4, we assign the western bottom boundary of the model a temperature so that the basal region is in the two-phase liquid-vapor region of P-T-X space. We then assume the temperature declines sharply along the dip of the AMC and then assign a shallower gradient for the remaining width of the model. To explain the differences in vent salinity and temperature across the MEF, we consider several of 2-D across-axis simulations in which the bottom boundary temperatures and/or crustal permeability structures are different.
3. Numerical Modeling Code FISHES
 Fully Implicit Seafloor Hydrothermal Event Simulator or FISHES is a numerical scheme capable of simulating two-phase flow in a permeable medium containing a NaCl-H2O fluid [Lewis and Lowell, 2009a, 2009b]. It provides solutions to the equations governing fluid flow together with an equation of state that calculates thermodynamic properties of the fluid as functions of temperature, pressure and salinity at each time step. These equations stem from Faust and Mercer  with the addition of an equation for conservation of salt. Conservation of mass is given by
where φ is the porosity, ρ is the bulk density, v is the Darcian velocity and the subscripts v and l refer to the vapor phases and the liquid phases, respectively. Conservation of momentum is assumed to be governed by Darcy's Law for each fluid phase, which states
 Where k is the permeability, P is the pressure, z is the depth, µ is the dynamic viscosity, and g is the gravitational acceleration. The relative permeability kr is assumed to be linearly proportional to the respective volume saturation of each phase, and the residual saturation is assumed to be zero. The energy conservation equation is
where h is the specific enthalpy, T is the temperature, λm is the effective medium thermal conductivity, and the subscript r refers to the rock; ρ and h without subscript refer to bulk quantities. The equation describing the conservation of salt from Bai et al.  is given by:
 The solutions to equations (1) through (5) are obtained using Patankar's  scheme for a finite control volume with upstream weighting, implying that each of the internal nodes are considered as the centroid of a control volume through which the fluid flows. The densities, salinities, and enthalpies as functions of P-T-X are evaluated via linear interpolation between values in thermodynamic lookup tables. These tables are valid for pressures between 85 and 1000 bars, temperatures between 0 and 800°C, and salinities between 0 and 100 wt% NaCl.
 The bulk density from 300 to 800°C and the salinity and density on the upper boundary of the two-phase region are based on the work of Anderko and Pitzer , and the bulk density from 0 to 300°C is based upon the work of Archer . The specific enthalpy is compiled using the framework of Tanger and Pitzer  and Archer . The salinity on the halite-saturated vapor surface is obtained using the correlation equation from Palliser and Mckibbin . Lewis and Lowell [2009a] provide additional details on the FISHES code. FISHES and the user's guide may be downloaded using the link: http://www.geophys.geos.vt.edu/rlowell/kaylal/ from the “downloads” section.
4. Model Setup
 The across-axis models (Figure 5) are assigned a depth of 2 km and a width of 1.5 km. Side walls of the system are impermeable and insulated. The top boundary of the system is permeable and maintained at constant seafloor pressure of 220 bars. For fluid leaving the surface of the model, the numerical scheme is upwinded for temperature and salinity, which means that the value at the surface is calculated based on the value of the parameter one node below. The bottom boundary of the system is impermeable and maintained at a constant temperature. The western (left) bottom boundary is made hotter than the eastern (right) bottom boundary to simulate the effect of the dip in the magma chamber.
 In the “basic” model, the highest bottom boundary temperature (BBT) is 500°C (Figure 5a). Simulations are also run for highest BBT of 475, 450, and 400°C. We consider a maximum BBT of 500°C to be a reasonable assumption based on Fontaine et al. . They have used geothermobarometry to infer that, for several mid-ocean ridge hydrothermal systems, the temperature on the top of the “reaction zone” or zone where chloride and silicate equilibrate lies between 415°C and 445°C. Additionally, Jupp and Schultz  argue that that water in hydrothermal convection cells upwells at ∼ 400°C when driven by a heat source above ∼ 500°C.
 In each case, the temperatures are decreased from the western to eastern edge with similar temperature gradients. Temperature at the seafloor is set to 10°C and initially increases linearly from top to bottom of the system. The initial pressure distribution is hydrostatic and salinity is set to 3.2 wt% everywhere. Three permeability models were used: homogenous, heterogeneous with layer 2A, and heterogeneous with layer 2A and an anhydrite shell. Each is described below.
 The cell size in our models is 25 m × 25 m. Although this spacing is too coarse to model individual vent chimneys, this cell size corresponds roughly to the spatial scale of the large sulfide structures at the MEF. Vent fluids from these structures tend to have relatively uniform temperature [Butterfield et al., 1994]. Because we are not trying to model individual vents, but only the general N-S gradients, we believe that this resolution is sufficient.
4.1. Homogenous Permeability Model
Han et al.  show that permeability plays an important role in plume structure and heat output of hydrothermal systems, but that it has relatively little effect on vent fluid temperature and salinity, given the same bottom temperature distribution. Therefore, the purpose of the homogeneous permeability model was to understand the effect of different maximum BBTs (500, 475, 450, and 400°C) while keeping other model parameters the same. The goal was to determine whether changes in bottom boundary temperature could explain the observed gradients in vent temperature and salinity in the MEF.
 Although the Rayleigh number does not fully describe convection in a phase separating system, it is still a useful parameter for approximating the vigor of convection. Here we use the classical expression
where α, μ, and ρl represent the thermal expansion coefficient, dynamic viscosity and density of the fluid, respectively; g is the gravitational acceleration, H is the vertical thickness of the system, k is the permeability, ΔT is the temperature difference between the top and the bottom boundaries, and κ is the effective thermal diffusivity. Assuming approximate high-temperature values of α = 10−3 °C−1, μ = 10−4 Pa-s, ρl= 103 kg/m3, g = 9.8 m/s2, ΔT = 400°C, κ = 8 × 10−7 m2/s, H = 2 × 103 m, and k = 10−13 m2, we find Ra ≈ 10000 for these simulations. This value greatly exceeds the critical Rayleigh number Rac =27.1 for a fluid-saturated open-top porous medium [Nield, 1968]; consequently, we expect temporal and spatial variations in temperature and salinity of vent fluids. However, by specifying the basal temperature, P-T conditions for phase separation at the base of the system for a given temperature and permeability remain essentially constant [Han et al., 2013]. This enables us to potentially determine the effect of changing BBT for our homogenous permeability model and also understand how permeability affects seafloor temperature and salinity for the same temperature boundary condition at depth.
4.2. Heterogeneous Permeability Model
 The permeability structure of the crust may be important because it controls the extent of seawater mixing with rising hydrothermal fluid and thus affects both the temperature and salinity of venting fluid. Lowell et al.  show that the hydrothermal discharge temperature decreases as the permeability in the extrusive layer 2A increases relative to that in the deep discharge zone. They also show that the presence of a low-permeability barrier, perhaps resulting from anhydrite precipitation, lessens the impact of the higher permeability in layer 2A. Fontaine et al.  explore a similar idea using single-phase, two-dimensional numerical models of hydrothermal circulation to investigate the physical controls on maximum venting salinities. They argue that the venting salinity is dependent on the permeability structure and that the simplest models with uniform or layered permeability are unable to account for both the maximum observed salinity and temperature. When they use a more complex model in which the upflow is surrounded by a low-permeability shell, they infer that such models are more consistent with the observational data. Their models fit the observations best when the permeabilities of the upper layer differ from the permeability of the lower layer by factors of ∼10 and ∼100 in upflow and downflow regions, respectively, and when the permeability of the shell is one tenth that of the lower layer.
 Seismic data collected at the MEF [Van Ark et al., 2007] indicate that layer 2A is continuous for the region with a mean thickness of 460 ±160 m. To study the impact of layer 2A on surface fluid properties, we have assumed that it is 500 m thick and has higher permeability than the underlying sheeted dikes. Furthermore, to account for heterogeneities in layer 2A due to local mineral precipitation, we added a low permeability “shell” surrounding the upflow region to some models. The assumed values of permeability in layer 2A and in the anhydrite shell, for various combinations of magnitude and spatial location, are given in Table 1. Figure 5b shows a general schematic.
Table 1. Model Specifications for Generalized Model Shown in Figure 5ba
Permeability of Layer 2A (m2)
Thickness of Layer 2A (m)
Anhydrite Layer k = 10−15 present
Location of Anhydrite Layer From Starting Node (m)
Names of each simulation have been used as legend in temperature and salinity versus time plots.
5.1. Homogeneous Permeability Model
 Figure 6a shows the results of a “basic” across-axis simulation with a maximum BBT of 500°C after a simulation time of 60 years. For the specified geometry and P-T conditions of our model, phase separation occurs in a small region close to western end of the bottom boundary (Figures 7b and 7c). The rest of the system remains in a single phase regime (Figure 7a). We interpret the single plume to represent a MEF vent structure associated with a cross-axis convection cell. The surface temperature varies from ∼370°C at the center of the plume and drops of sharply to the seafloor temperature of 10°C. Although the plume structure is stable, the maximum temperature of the fluid fluctuates by a few degrees on a time scale of years. The average salinity of the plume also varies with time and VDF and BDF fluids vent at different times, again with episodic fluctuation on the time scale of a year.
 We ran the homogeneous model for BBTs 475°C, 450°C and 400°C to investigate the effect of different BBTs on the vent fluid temperature and salinity. These simulations are referred to as T500, T475, T450, and T400, respectively. Figure 6b is a plot of average of maximum temperatures values (AMT) with their respective standard deviations for the different simulations. These values were calculated by averaging the maximum values of temperature in the plume for each year over the duration of the simulation. Similarly, average minimum salinity (AMS) values were obtained at the center of the plume (where the temperature is the greatest) for each of the homogenous permeability models (Figure 6c).
 Figure 6b shows that there is not much difference in the AMTs of models T500, T475, and T450, whereas for T400, AMT is noticeably lower. The high AMTs for models T500, T475, and T450 are close to what is observed for southern MEF vent structures such as Puffer and Sully. However, for the range of BBT temperatures between 450°C and 500°C, it is not possible to detect differences in surface temperature. Low AMT obtained for T400 ∼ 305 ± 20°C is similar to the vent temperatures at northern vent structures such as Hulk, which recorded values ∼ 330°C before 1999. The simulation T400 is in the single phase regime, however, resulting in vent salinity fixed at seawater value of 3.2 wt %. Figure 6c shows that although there is some variation in the AMS values for all the models, the range of temporal variability suggests that AMS values are also essentially independent of BBT in the range 450–500°C.
 Although the AMT values correspond to values recorded at vent structures on the southern MEF, salinity values for the same simulations are higher than those observed. As mentioned before, the zone of phase separation in these models is confined to the bottom-most region of the system. As the hot, vapor rich, low salinity fluid rises, it mixes with relatively hot, but single-phase fluid that is circulating deep in the system. As a result, although both the temperature and salinity of the upwelling fluid are changed as a result of mixing, vent salinity may be changed to a greater extent than the temperature. To investigate the effects of mixing, particularly in the shallow part of the system, we consider the effects of heterogeneous permeability by imposing a layer 2A on the model.
5.2. Heterogeneous Permeability Model With Layer 2A
 For these simulations, BBT distribution is fixed as shown in Figure 5b and simulations were run for scenarios 1–6 in Table 1. The results of scenarios 1–4 are shown in Figure 8 and scenarios 5 and 6 are shown in Figure 9.
 Introduction of layer 2A in our models increases overall permeability of the system thereby increasing the effective Rayleigh number. This leads to more vigorous convection and increased mixing of cold seawater into the plume before the fluids come to the surface (Figure 8) thereby decreasing the temperature and increasing the salinity. By introducing layer 2A, the vent fluid temperature which initially was similar to that at southern vent structures (∼370°C) is now more similar to temperatures at northern vent structures (∼320°C). AMS values are still below but quite close to seawater value, similar to what is observed at northern MEF vent (Hulk, Dante, Figures 1 and 7c). The thickness of layer 2A does not seem to cause major variations in surface temperature and salinity (Figure 8c). Compared to the homogenous models, lower temperature VDF are obtained on the surface while the system is undergoing phase separation at depth.
5.3. Heterogeneous Permeability With Anhydrite Shell
 We ran simulations 5 and 6 as shown in Table 1 with a region of low permeability (10−15 m2) in the extrusive layer near the main discharge zone. This zone represents the precipitation of anhydrite that might occur as a result of mixing between the deeply circulating sulfate-poor hydrothermal fluid and sulfate-rich seawater circulating through the extrusive layer 2A [e.g., Lowell et al., 2003, 2007]. High-resolution magnetic data from the Endeavour vent field [Tivey and Johnson, 2002] also supports the idea of a vertical chemical barrier surrounding high-temperature vent fields. As in Lowell et al. , we simulate the precipitation of anhydrite by introducing the “shell” directly without actually considering the process of precipitation. We assume the shell is 50 m thick, extending the entire depth of layer 2A, which for each case is 500 m. We assume the permeability of the low-k barrier has a constant value of 10−15 m2.
 Introduction of the low permeability shell partially prevents the rising high-temperature plume from mixing with seawater within layer 2A. As a results the plume temperature increases somewhat, but not to values as high as those obtained without layer 2A. The temperature of the discharging fluid then lies nearer to those observed in the central part of the MEF (Figure 1). Some part of the hot hydrothermal fluid travels around the low permeability zone (Figure 9a), causing moderately warm temperature flow outside of the high temperature plume (10–20°C). Such flow is analogous to diffuse flow common around vent fields. Such diffuse flow was not observed in previous simulations and seems to result from a more complex permeability structure where parts of the rising plume may find alternate paths and express themselves on the surface as a variety of mixtures with sea water. The diffuse-flow fluid is persistent in both cases with the anhydrite layer. When this layer is farther from the center of the plume the maximum temperature of the main plume on the surface is less than when the anhydrite shell is closer to the rising plume (Figures 9b and 9c). There is no significant difference in temperatures of diffuse fluids in either case.
 As the anhydrite shell limits the mixing between seawater and hot hydrothermal fluids, AMS values in these models are lower than those obtained in models with layer 2A alone. These values correspond to the northern vents at the MEF (Figure 1).
6.1. Role of Bottom Temperature
Han et al.  use similar modeling techniques to investigate how permeability and maximum bottom temperature affect vent fluid temperature and salinity in sea-floor hydrothermal systems. While their simulations are for a different set of model parameters (depth = 1 km, maximum BBT = 450°C), they arrive at some general conclusions for high Rayleigh number convection models. Their results suggest that while hydrothermal heat output increases linearly with permeability, the BBT plays a more important role in determining surface temperature and salinity. This argument led to the reasonable assumption that for MEF, the gradients in surface temperature and salinity could be a reflection of a gradient in the bottom boundary temperatures. With the homogenous permeability models, our aim was to explore this possibility. However, surface temperatures and salinities obtained for our homogenous permeability models do not correlate as well to the BBTs in the system as those in Han et al. .
 This discrepancy between our results and those obtained by Han et al.  may be due to the difference in the model parameters used for MEF and those used by Han et al. . We use a depth of 2 km as opposed to the 1 km depth used in their models. Han  shows a few simulations for models that are 2 km deep (BBT = 450–500°C). In those models, the correlation between the BBT and surface temperature and salinity is not as strong as for systems with a depth of 1 km. This reiterates the importance of modeling individual vent fields rather than using generic models to explain vent-specific observations. The homogenous permeability models that were phase separating give rise to temperatures that were similar to the southernmost vent structures at the MEF, but could not explain the salinity of these structures nor the observed gradients in temperature and salinity. For MEF, one cannot use BBT alone to explain the observed gradients. This result calls for some other mechanism to control temperature and salinity of hydrothermal vent fluids.
6.2. Heterogeneous Permeability
 By assuming layer 2A has higher permeability than the sheeted dikes, we found that enhanced mixing and cooling of hydrothermal fluid before discharging at the surface can produce a significant difference in the surface temperature and salinity of the fluids. In particular, the results of the models with heterogeneous permeability suggest that higher permeability in layer 2A lowers the temperature of the venting fluid and brings the salinity closer to that of seawater due to greater mixing between seawater and hydrothermal fluids. By contrast low permeability barriers around the upflow zone can shield the rising fluid from mixing. For different thicknesses (250, 500, and 750 m), permeabilities (5 × 10−13 and 10−12 m2) and configurations of layer 2A (e.g., with a low permeability shell resulting from anhydrite precipitation), focused and diffuse flow with a wide range of temperature and salinity values can be obtained. The lowest temperature values obtained for our model with a layer 2A (see results for P5e-13 and I0Pe-15) matches the temperature values at the northern vent fields. By using an appropriate combination of magnitude of the permeability in layer 2A together with the location and thicknesses of a locally lower permeability barrier such as might result from mineral precipitation, a variety of temperature and salinities can, in principle, be obtained for the vent fluids. Without evidence to support a certain distribution of permeability, we chose not to extend this discussion further.
 A drawback of our modeling is that although the surface temperatures in our model corresponded, or could be made to correspond, with vent structures along the MEF, the surface salinity values were all higher than the observed values. At the deepest and hottest part of our model (initial temperature = 500°C; initial pressure ≈ 375 Bar), the salinity of the vapor phase is 0.175 wt% NaCl [also see Bischoff and Pitzer, 1989]. Assuming that this value is close to the salinity of vapor formed deep in the system at MEF, and that surface fluids at MEF are a mixture of this vapor and seawater, the extent of mixing between the two fluids can be estimated by correlating the end member composition of vapor and seawater mixture with the surface salinity of vent fluids obtained at MEF from Butterfield et al. . Such a correlation indicates that the amount of vapor in the surface vent fluids varies from ∼ 7 wt % (surface salinity = 505 mmol/kg; at Hulk) to ∼ 56 wt % (surface salinity = 253 mmol/kg; at Peanut). Lower salinity values, especially at southern MEF, indicate that a greater percent of the venting fluid is vapor than that obtained in our simulations. This indicates minimal or little mixing of the rising hydrothermal fluid with seawater for the southern vents. Implications of low salinity fluids obtained at the MEF are discussed below.
6.3. Low Surface Salinity at the MEF
Johnson et al.  suggest that high-temperature venting at MEF is controlled by high permeability local faults and that recharge and discharge occur along narrow cylinders that penetrate deep to the heat source. Given the complex nature of faulting at mid-ocean ridges, the relative depths of recharge and discharge limbs may determine how much hot fluid mixes with seawater before venting. As seems with the southern MEF, venting occurs with comparatively little mixing between the rising plume and sweater. This indicates that in addition to shallow heterogeneities in the oceanic crust, permeability barriers also exist deep in the system that inhibit mixing leading to low salinity values on the surface. Similar results are obtained by Crone et al.  for 9°50′ N on East Pacific Rise (EPR), where they argue that lateral variations in permeability could contribute to the pattern of seismicity at EPR, and influence the location of upflow and downflow zones of hydrothermal flow. Therefore, the range of surface temperature and salinity values at MEF are a likely result of both shallow heterogeneities in layer 2A and lateral permeability contrasts deep in the crust.
 Heterogeneity in the oceanic crust may be present as a result of various factors such as magma supply, lava flow, faulting, mineral precipitation, diking, etc. and may occur at all depths and scales. Rising fluids driven by buoyancy will find the least resistive path to the surface via higher permeability channels. This will determine the amount of mixing and cooling it undergoes before exiting on the seafloor. Overall, higher permeability will lead to more mixing and therefore more cooling of hydrothermal fluids and vice versa. Therefore, we find our study in agreement with previous studies [e.g., Fontaine et al., 2007; Driesner, 2010] that in order to accurately understand the temperature and salinity of vent fluids, and to accommodate the heat flow data from mid-ocean hydrothermal systems, it is important to recognize the role played by permeability on the system output.
 We have used a two-phase numerical modeling code and data from studies at the Main Endeavour Field to investigate the possible causes of the observed trend in vent temperature and salinity between 1984 and 1999. We ran several models to study different influences on hydrothermal fluid circulation in the region. We find that although bottom boundary temperature in a hydrothermal system influences the surface temperature of a hydrothermal plume in general, the latter is not directly proportional to the former. Hydrothermal fluids may mix locally while rising to the surface generating a variety of temperatures and salinities. Moreover, low surface salinity values at the Main Endeavour Field are likely a result of lateral permeability contrasts deep in the system which inhibits mixing of rising vapor rich fluids with seawater. Finally, we conclude that the trend of temperatures and salinities in vent fluids at the Main Endeavour Field results from a heterogeneous oceanic crust which has higher overall permeability in the northern part of the vent field than the southern part.
 We thank the Associate Editor Ed Baker and two anonymous reviewers for the helpful comments on the original draft of this paper. This research was supported in part by NSF grants OCE-0819084 and OCE-0926418 to R.P.L and OCE-0818783 to K.C.L.