Simulation of hydrothermal vents in the Izena Cauldron, Mid Okinawa trough, Japan and other Pacific locations

Available online at www.sciencedirect.com

ScienceDirect Journal of Hydro-environment Research 8 (2014) 343e357 www.elsevier.com/locate/jher

Research paper

Simulation of hydrothermal vents in the Izena Cauldron, Mid Okinawa trough, Japan and other Pacific locations Anusha L. Dissanayake a, Poojitha D. Yapa a,*, Kisaburo Nakata b a

Department of Civil and Environmental Engineering, Clarkson University, Potsdam, NY 13699, USA b Faculty of Marine Science and Technology, Meijo University, Tenpaku, Japan Received 2 November 2013; revised 23 May 2014; accepted 30 May 2014 Available online 12 July 2014

Abstract Simulation results of Jade hydrothermal vent in Okinawa trough, Japan and Dante hydrothermal vent in Endeavour ridge in the north east Pacific are presented using an improved version of MOHTV model. The results are presented along with some limited field comparisons. A limited study on the sensitivity of model results to key model parameters is also presented. The model used is capable of simulating the hydrodynamics, thermodynamics, and mineral formations in underwater hydrothermal vents. The transport and spread of the plume fluid and minerals formed are simulated in three stages: plume dynamics stage that is momentum and buoyancy driven; transition to far field conditions as a gravity current; and far field conditions where the mineral particles move according to advection diffusion governed by the ambient currents and mineral particle settling velocities that eventually lead to their bed deposition. Thermodynamics include the changes in plume temperature and its related properties such as the vent fluid density. Physico-chemical processes are the chemical reactions that occur in the plume due to the hot vent fluid mixing with cold ambient water. Chemical reactions form new compounds (mostly minerals), which change the plume properties and therefore its behavior. Model simulates the formation of minerals namely FeS, FeS2, ZnS, CuFeS2, CuS, PbS, CaSO4, BaSO4, and Particulate Manganese e PMn. This improved version of MOHTV considers a user defined particles size distribution, the bathymetry of the ocean bottom, and can be used to run long-term simulations. The model results compare reasonably well with the field data. The parametric analyses show, which input and/or ambient conditions most affect the distribution of particles. © 2014 International Association for Hydro-environment Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights reserved.

Keywords: Dante; Endeavour ridge; Hydrothermal vents; Izena; Jade; Mineral distribution

1. Introduction Hydrothermal vent plumes are releases of geo-thermally modified seawater (vent fluid) into the ocean through fissures close to the mid ocean ridges. Vent fluids are rich in dissolved minerals. They are released at temperatures and * Corresponding author. E-mail addresses: [email protected] (A.L. Dissanayake), pyapa@ clarkson.edu, [email protected] (P.D. Yapa), [email protected] (K. Nakata).

velocities ranging from approximately 10 to 400
C and 0.01e6.2 m/s respectively (Ginster et al., 1994). When vent fluid is released at the seabed, it mixes with cold seawater triggering many chemical reactions forming precipitates (mineral sulfides, anhydrite, sulfates) due to the sudden changes in temperatures. Behavior of the vent fluid varies depending on the fluid composition, temperature and velocity at release, and the types of precipitates formed. Slow releases with low temperatures are known as diffused vents and the amount of precipitates formed in these vents is minimal. High temperature vents with high velocities form plumes are

http://dx.doi.org/10.1016/j.jher.2014.05.003 1570-6443/© 2014 International Association for Hydro-environment Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights reserved.

344

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

clouded with many mineral precipitates (black, gray, or white in color) and can even rise a few hundred meters in the ambient water. The ocean currents carry these mineral precipitates along. They get distributed and deposited in different locations, sometimes very far away from their source. Examples of mineral precipitates found in hydrothermal vent fields are pyrite (FeS2), iron sulphide (FeS), chalcopyrite (CuFeS2), sphalerite (ZnS), covellite (CuS), anhydrite (CaSO4), barite (BaSO4), galena (PbS), Mn, Au, and Ag (Tivey, 2007). Some of these minerals are found in abundance in vent fields while others are rare and found only in specific sites. Many mineral deposits are formed in and around the vent fields over millions of years, which are the origins of deposits found on land (Tivey, 2007; Glasby and Notsu, 2003). Many countries are interested in harvesting these deposits, which are of high economic value. Economic grade minerals such as CuS, ZnS, PbS, Ag, and Au based minerals are reported to be common in deposits found in the Okinawa trough near Japan, and the Manus Basin and Conical Seamount near Papua New Guinea (Glasby and Notsu, 2003). Tao et al. (2013) developed a model to simulate hydrothermal plumes and compared their simulations with analytical results. They present short-term simulation results of the plume in high resolution. Their model however, did not include the formation of chemicals, their transport and distribution at the seabed. No comparisons with experimental data were presented. Dissanayake et al. (2014) developed a new model named MOHTV (MOdel for HydroThemal Vents) to simulate hydrothermal vent plumes using integral lagrangian control volume method, which included near field, far field, and re-circulation of particles. The model included a limited number of chemical formations within the plume. In this paper MOHTV is improved by adding more chemicals, and tested by comparing simulation results with experimental data. In addition, this paper simulates the mineral transport and deposition in the Okinawa trough, and the Dante hydrothermal vent in Endeavour ridge in the Pacific. The model algorithms are significantly improved to run much longer term (3 months) simulations compared to the earlier version of the model taking into account the bathymetry of the ocean bottom as well. While the previous paper (Dissanayake et al., 2014) presented the model formulation and some preliminary results, this paper details the model testing with available experimental data, application to real field conditions, and a parametric study of key model parameters.


! d mV dt

Rate of change of total momentum of the CV

±

dmsep ! V dt

¼

!

2. Model for hydrothermal vent plumes (MOHTV) MOHTV model's detailed formulation can be found in Dissanayake et al. (2014). Here, we briefly outline the model along with the improvements made. 2.1. Plume dynamic stage (PDS) During the PDS the plume is driven by the momentum of the fluid released at the vent and buoyancy force due to the density difference between the ambient seawater and the plume fluid which consists of a mixture of vent fluid, seawater, and particles. The effect of the particles on the composite density is negligible for most common thermal vent conditions because the particles are so small and the concentrations are low. The currents force the ambient water into the plume and change its momentum, which is a dominant factor that determines its path. The shear due to the difference in plume velocity and ambient velocity also causes the plume to entrain ambient water. Details of how entrainment is modeled are the same as given by Lee and Cheung (1990) and Yapa and Zheng (1997). As the plume rises, the dominating force changes from momentum to buoyancy mainly as a result of entrainment. Termination of the PDS is assumed to occur when the plume reaches a neutral buoyancy level (NBL). NBL is the level at which the density of the plume is equal to the ambient water density (Dasanayaka and Yapa, 2009). The plume dynamics are modeled using integral Lagrangian control volumes (CV) similar to that of Zheng et al. (2003). In a CV, properties such as velocity, salinity, and temperature are averaged across the cross section. CVs travel along the three dimensional path of the plume. The plume temperature changes due to the entrained ambient water. The effects of the Coriolis force, and mineral precipitation calculations are accounted for during the PDS. Concentrations of dissolved metals are high closer to the release point. Therefore, most of the mineral precipitation reactions occur near the vent orifice. MOHTV assumes that the vent fluid is released into the seawater through a circular orifice in the vertical direction. Mass and momentum conservation of a CV are the key equations of the plume model and are given in Eqs. (1) and (2) respectively dm dmsep ¼ ra Q e ± dt dt

V a ra Qe þ Entrained momentum into the CV due to ambient water

Momentum of the separated mass from the CV

ð1Þ

ra  r ! mg k r

Buoyant upward thrust on the CV



! !' 2 ! V 2prcv hra CD  V   V a ! V 

Drag force acting on The CV due to velocity difference between plume and ambient liquids



! ! 2m U  V

Coriolis force acting on the CV

ð2Þ

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

where m ¼ mvf þ mp þ mew ¼ total mass of the CV (kg); dm/dt is the rate of change of total mass of the CV (kg/s); ra ¼ density of the ambient water (kg/m3); Qe ¼ entrainment rate of the ambient water (m3/s); msep ¼ mass of separated (or re entrained) particles from the CV (kg); mvf ¼ mass of the vent fluid (kg); mp ¼ mass of particles (kg); mew ¼ mass of entrained ambient water (kg); ! ! V ¼ velocity of the CV (m/s); V a ¼ ambient velocity (m/s); r ¼ density of' the CV (kg/m3); g ¼ the gravitational accelera! ! ! tion (m/s2); V a ¼ projection of V a in the direction of V (m/s); rcv ¼ radius of the CV (m); CD ¼ drag k ¼ unit !coefficient;  vector in the vertical direction; hcv ¼  V Dt ¼ height of the ! initial CV (m); U ¼ angular velocity of the earth (s1); Dt ¼ time step !(s); Lee and Cheung (1990) suggested thatDt ¼ 0:1bo = V ; b0 ¼ radius at the point of release (m). The drag term is included for completeness and also to be consistent with some earlier models. For thermal vents, we found that its effect on the overall plume is negligible. Hence, the drag coefficient is assumed to be zero for the simulations in this paper. 2.2. Advection diffusion stage (ADS) ADS is where the plume movement is dominated by the ambient currents and is modeled using the advection diffusion equation for each specie given in Eq. (3). The mineral particle distribution in the far field of the plume is simulated here and the particles are carried to the final deposition areas or out into the large ocean. vCi vðuCi Þ vðvCi Þ vðw þ wbi ÞCi þ þ þ vx vy vt vz       v vCi v vCi v vCi ¼ Dx Dy Dz þ þ þ Si vx vy vz vx vy vz

ð3Þ

where i indicates the specie (i.e. type of chemical or particle size); Ci ¼ mass concentration (kg/m3) of specie i; u, v, and w corresponds to ambient velocity (m/s) in the horizontal ( x, y) and vertical (z) directions respectively; Dx, Dy, and Dz are diffusion coefficients (m2/s) in water in x, y, and, z directions respectively; Si ¼ source or sink term (kg/m3/s) (due to dissolved or deposited precipitates); and wbi ¼ vertical velocity (m/s) of specie i. 2.3. Transition between PDS and ADS During the transition from PDS to ADS, a gravity current is formed at the NBL. The driving force of the gravity current is the pressure build-up at the plume terminal level. In many situations when the plume reaches the NBL, the vertical component of plume velocity is yet to reach zero. This velocity causes the plume water mass to continue to move in the vertical direction under negative buoyancy, until the velocity reaches zero. The water mass that moved above the NBL has a density higher than the ambient water at the same level of their existence and continues to move back to the NBL forming a gravity-current. The pressure build-up mentioned above is caused by this water mass above the NBL. This plume fluid

345

movement at the NBL is modeled as a gravity current in the model by Dissanayake et al. (2014). 2.4. Chemical reactions and mineral precipitation Formations of chemical reactions in the vent plume are modeled using their rate equations. This improved version of MOHTV includes Anhydrite (CaSO4), barium sulfate (BaSO4), and particulate Mn along with previously modeled minerals: FeS, pyrite (FeS2), chalcopyrite (CuFeS2), sphalerite (ZnS), Galena (PbS), and Covellite (CuS). CaSO4, and BaSO4 precipitation reactions are modeled based on the rate equation presented by Granbakken et al. (1991). They assumed the reaction rate to be surface controlled and the precipitation rate is given by Eq. (4).
2þ  2 1=2  0 . 2 1=2 2 d½MeAðsÞ ¼ kprec S Me  Ksp g ð4Þ A dt Where MeA, kprec, S, t, [Me2þ], [A2], K0sp, g are the amount of the solid mineral precipitated; precipitation rate constant; specific surface area of the crystals per unit volume of the solution; time, concentration cation at time t; concentration anion at time t; thermodynamic solubility product; and the mean activity coefficient of MeA in solution respectively. The rate constant kprec is given by Eq. (5). kprec ¼ A expð  Ea =RTÞ

ð5Þ

where A, Ea, R, and T are the pre-exponential factor; activation energy; gas constant; and the temperature in Kelvin respectively. The parameters A, and Ea for CaSO4 are 9.8  103 and 58 ± 9 kJmol for temperature range 120e150
C and 2.39  107 and 63 ± 2 kJmol for temperature range 15e70
C respectively. A and Ea for BaSO4 are 3.05  104 and 33.5 ± 4 kJmol for temperature range 25e125
C. Other mineral formations in the model are done in a similar way and the details are given in Dissanayake et al. (2014). Majority of the chemicals are formed closer to the release point and taken up to the NBL by the plume and released into the far field where they travel with the resultant of their settling velocities and ambient currents. Settling velocities of the mineral particles are calculated according to an integrated formulation given in Zheng and Yapa (2000). Their work was based on Clift et al. (1978). Buoyant/settling velocity depends on the shape of the bubbles: sphere, ellipsoid, or a sphericalcap. The ambient velocity, salinity, and temperature are given as input to the model in a defined grid in x (West to East), y (South to North), and z (Depth) directions in the region. The maximum depth of each grid is defined in the model based on the ocean floor geometry in the region. This is important when simulating the deposition of particles. 2.5. Mineral particles settling on the ocean bottom and the local geography in the region The particles released into the far field of the plume move with their settling velocities and the ambient currents. They

346

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

settle on the seabed and accumulate over time. The settling patterns are dependent on the particle properties, regional hydrographic patterns and ocean bottom bathymetry as well. According to a defined grid, the ocean bottom elevation is given as an input to the model so the particles are settled according to the regional variation in the ocean bottom. The settled particles can be re-suspended, but these effects are not taken into account. 2.6. Model components and their interconnections Fig. 1 shows a block diagram of how the different modules (components) developed are integrated together to form the model. A single headed arrow shows how one component affects another. A double-headed arrow shows that two components may mutually affect each other. 3. Mineral particles in hydrothermal vent fields 3.1. Particle distribution Reactions in hydrothermal vents release a large number of particles continuously into the ocean waters. They are distributed and deposited in and around vent environments. Several studies have been conducted related to mineral particle distribution in the Izena cauldron in Mid Okinawa trough, Japan and in Endeavour segment in the Juan de Fuca ridge. Halbach et al. (1997) studied the minerals in the Jade field, Izena cauldron. They found that Zn and Pb are common in the forms of sphalerite and galena respectively, along with Cu, Sb, As, Ag, and Au. Marumo and Hattori (1999) observed that fallout sediments are distributed and settled on the sea floor at Jade site in the Izena as sulfide and sulfate rich minerals. However, the total amounts of deposits are still publicly unknown. MOHTV simulates the distribution and deposition patterns of these minerals over time.

Black smoker particles in Endeavour segment thermal vents are abundant with Fe, S, Ca, Cu, and Zn and their particle sizes range from 10 mm to 500 mm (Feely et al., 1987). Baker and Massoth (1987) described two hydrothermal vents in the Juan de Fuca ridge. They explained that the vent fluid in the Endeavour segment is dominated with sulfur, and a high proportion of coarse-grained Fe sulfides are deposited. They estimated that the particle flux in the Endeavour segment varied from 546 ± 312 to 204 ± 116 g/s at different distances from the plume. Dymond and Roth (1988) collected and analyzed hydrothermal vent originated particles in the Endeavour ridge at different distances, and heights from their source at different times. A near field sediment trap in a black smoker, positioned 21 m above the discharge point located at a depth of around 2200 m, collected 10 to 20 times more materials than a far field trap at a depth of 2100 m and two kilometers north east of the hydrothermal field. Fe and Cu enriched particles tend to settle closer to the plume while Mn based particles are settled away from the plume. The reason for this Mn precipitate distribution was thought to be slow oxidation rates of Mn, which is also bacterially controlled. Variability of the data with time due to the variations in ambient velocity and vent fluid discharge are also detected. Hydrothermal particle distribution is considered to be of two types: the ones produced soon after the vent fluid is mixed with sea water (non-aggregated) and secondary particles precipitated under the influence of bacterial or inorganic processes. Fine particles in the far field get collected together with each other and biogenic debris to form large particles. Dymond and Roth (1988) suggested considering the above difference in the particle size distributions in hydrothermal vent models. Their observed non-aggregated maximum particle sizes are: pyrite at 1950 m depth e 22e31 mm, and at 2100 m depth e 140e180 mm; sphalerite at 2100 m depth e 220 mm; and anhydrite at 2100 m depth e 340 mm. The mean particle size, for the above minerals ranges around 2 ± 1.5 mm at both depths.

Fig. 1. Block diagram of the model components and their interdependency.

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

347

3.2. Organic/inorganic processes affecting the particle sizes

4.1. Simulation of Jade hydrothermal Vent in Izena Cauldron, Mid Okinawa trough

Organic processes also affect the reactivity, transport, and the residence time of particles in the water column (Breier et al., 2012). For example, Mn particles oxidation rates are slow and their concentrations are high in the far field of the plume (Dymond and Roth, 1988). Toner et al. (2009) explain that the Fe-based minerals in hydrothermal vents undergo both inorganic and organic processes, which we believe will change the particle sizes. Toner et al. (2012) further pointed out the effects of microbial activity on the fate of hydrothermally originated iron (II) particles in the oceans. Anantharaman et al. (2013) also explained the importance of microbial activities on hydrothermal originated elements. But, the consideration of changes in particle properties is not included in the present model. CaSO4 is a mineral that dissolves fast compared to other hydrothermal minerals such as chalcopyrite, pyrite, sphalerite and barite. Anhydrite can be precipitated by mixing Ca rich hydrothermal fluid with seawater at temperatures above 140
C in the Eastern Pacific Rise and Mariana trough and above 120
C in Okinawa trough (Nakashima et al., 1995). But they start to dissolve below these temperatures. Feely et al. (1987) studied the dissolution rates of mineral particles based on in situ dissolution studies in the Southern Juan de Fuca ridge. According to them the time taken to dissolve 0.2 mm anhydrite particles is 0.9 h and for sphalerite and pyrite particles are 48 and 218 days respectively. Dissolution changes particle sizes, so their settling velocities change while the particles travel from the point of release from the plume to the point of deposition. Particle dissolution is not considered in the model at present. Following sections discuss two hydrothermal plume simulations in the Izena cauldron and Endeavour segment. The plume properties and mineral particle distribution patterns are quantitatively and qualitatively compared with field data and discussed in detail.

Hydrothermal vents were found in the Izena cauldron of the mid Okinawa trough in 1988 (Halbach et al., 1989; Watanabe et al., 1995). Two high temperature hydrothermal plume sites were observed in the cauldron namely the Jade and Hakurai (Kawagucci et al., 2010). Jade field lies at water depths of 1350e1400 m and the coordinates are 270 150 N and 1270 04.50 E (Glasby and Notsu, 2003). It has a black smoker plume located inside the Northeastern wall of the cauldron at a depth of 1350 m and several white or gray color hot fluid discharges with temperatures of 130
C (Glasby and Notsu, 2003). We simulated a Jade black smoker plume for twelve weeks for the velocity conditions in the Izena cauldron starting from January 2010.

4. Simulations Hydrothermal plume simulations need the ambient conditions as input to the model. Mainly they are obtained from available measurements in respective areas. Details of ambient velocity, temperature, and salinity data, and vent fluid composition, density, and temperature at the release point are given under respective sections of the simulations. For all the simulations it is assumed that plume fluid is released from a circular orifice in the vertical direction. Input particle size distributions are decided based on the observed size distributions in the respective regions. Particles are assumed to be spherical and respective densities of them are used in calculating their settling velocities in the water column. Bottom bathymetry is given as depths of relatively small grids so that the bottom variation is well captured when particles are settled on the seabed.

4.1.1. Model input 4.1.1.1. Bathymetry of the region. The dimensions of Izena cauldron are 6  3 km, with a maximum depth of 1665 m (Glasby and Notsu, 2003). Contour map and a three dimensional (3-D) view of the Izena cauldron are shown in Figs. 2 and 3 respectively. The bathymetry of the cauldron is input into the model as depths on a horizontal grid of size 25 m  25 m. 4.1.1.2. Variation of ambient velocity, temperature, and salinity. Velocities in Izena Cauldron are available for Sep. 2009 to Aug. 2011 at 1-h time intervals (ERI, 2012). These data are for depths of 3.3e1000 m above the bottom (mab) at intervals varying from 15 to 40 m. Samples of the velocity variation with depth in x (East) and y (North) directions (U and V components respectively) are given in Fig. 4. The average velocity is approximately 8.5 and 5.2 cm/s at depths varying from 540 to 1000 mab and from 150 to 500 mab respectively. The salinity and temperature profiles in the Izena cauldron observed by Watanabe et al. (1995) are shown in Fig. 5. They calculated the vertical eddy diffusivity of the Izena cauldron based on the density gradient. The water inside the cauldron is said to be vertically highly unstable with a very small density gradient. Therefore, the vertical eddy diffusivity values are large. Vertical variation of vertical eddy diffusivity with depth is shown in Table 1. 4.1.1.3. Composition of the vent fluid and particle size distribution. Jade field black smoker composition given in Glasby and Notsu (2003), and Ishibashi (1991) is used as the initial composition of the vent fluid (Table 2). Particle size distribution (PSD) based on observations in hydrothermal environments is given as an input to the model. Bemis et al. (2006) used PSDs ranging from 5 to 500 mm in their simulations. In German et al.’s (1993) observations in the TAG hydrothermal vents, Fe oxy-hydroxide phases are finely divided in submicron range and colloidal. They mentioned that the colloids are formed within the plume creating coarse particles of

348

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

Fig. 2. Contour map of Izena cauldron and the location of Jade site (black star).

sizes ranging up to 10 mm. Walker and Baker's (1988) observations in the Juan de Fuca ridge show that a high volume of thermal vent particles are less than 2 mm and found in higher levels of the plumes while particles of sizes 6e10 mm are found closer to the source and settled quickly. Based on the

above observations, PSD input to the model is 3 mme30%, 7 mm-35%, 10 mm-20%, 20 mm-10%, and 50 mm-5%. 4.1.1.4. Release conditions of the fluid. At Jade black smoker the depth of release, discharge velocity, density, and

Fig. 3. Three-dimensional view of the Izena cauldron (exaggerated vertical scale).

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

349

Table 2 Concentration of selected chemicals in Jade Black smoker fluid (based on the data by Glasby and Notsu, 2003; Ishibashi, 1991). Chemical species

Molar weight (g)

Composition of Jade fluid (mol/l)

H2S Fe2þ Zn2þ Pb2þ Cu2þ Ca2þ Ba2þ Mn2þ

34.081 55.845 65.409 07.210 63.546 40.078 137.327 54.938

0.010880000 0.000017280 0.000006080 0.0000000288 0.0000000024 0.018320000 0.000048000 0.000299200

temperature of the hydrothermal fluid released are 1340 m, 1 m/s, 638 kg/m3, and 320
C respectively (Glasby and Notsu, 2003). The discharge rate is 0.039 m3/s from a vent orifice of 0.12 m. Fig. 4. Velocity profiles in Izena Cauldron at (a) 720 h, (b) 1440, and (c) 2160 h.

Fig. 5. Temperature and salinity profiles in the Izena cauldron.

Table 1 Variation of vertical eddy diffusivity with depth (based on the data by Watanabe et al., 1995). Depth (m)

Vertical eddy diffusivity (cm2/s)

1100 1150 1200 1250 1300 1350 1400 1450 1500 1550

0.9 1.2 1.6 2.4 4.0 4.9 6.2 12.8 20.1 25.3

4.1.2. Results and discussion A three-month simulation is carried out for the conditions in Izena cauldron. Measured velocities are given as input in 6h intervals. The plume rises to the terminal level until its upward momentum is diminished and spread horizontally under ambient velocities, releasing mineral particles at higher levels. The simulated plume height varies between 30 and 180 m with time within the water depths from 1150 to 1300 m. Marumo and Hattori (1999) explained that the vent plumes in the Izena cauldron rise and spread laterally at 1110e1300 m depth. Further, the simulated plume heights are comparable to the observed plumes at EPR 80 400 to 110 500 N and 130 330 to 180 400 S where the height varies between 50 and 200 m. The model considers the mineral formations due to mixing of hot vent fluid with cold seawater. FeS, PbS, ZnS, and CuS are examples of chemicals in the model. Extensive Sulphide mineralization was observed at Jade site, which are composed mainly of Pyrite, Calcopyrite, Sphalerite, Wurzite, Galena, Anhydrite, and Barite with high concentrations of Pb, Ag, and Au (Glasby and Notsu, 2003; Marumo and Hattori, 1999; Nakashima et al., 1995). The model results of Jade black smoker are shown in Figs. 6 and 7 as concentration plots. Plots for water column shown in Fig. 6 are the maximum projected concentrations in the direction perpendicular to the viewing direction. Bed settled mineral concentrations (Fig. 7) are based on the total amount of particles settled during the simulation period. Fig. 6 shows a snap shot of concentrations of FeS in the water column after three months. High concentration red patches are distributed in the depths between 1300 m and 1150 m. Particle distribution patterns are the result of the prevailing velocities and the mineral type and the PSD. Particle settling velocity is a function of the density and particle size. Therefore, different particles have different settling velocities that cause them to travel in different paths. Fig. 6 shows a high concentration value of FeS is 1  105 mol/ m3 (880 ng/l) at depths ranging between 1150 and 1350 m. For CuS (not shown here) the highest concentration simulated in the water column is at a similar depth range as for FeS and has

350

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

Fig. 6. FeS concentration variations in the water column (a) xz section (b) xy section (plan).

a value of 1  109 mol/m3 (0.096 ng/l). High concentrations of Fe (1100 ng/l) and Cu (300 ng/l) were observed in the water column 100e300 m directly above the Jade vent field at depths between 1100 and 1300 m (Marumo and Hattori, 1999). These observations are summarized in Table 3. The model results are comparable with the experimental data except for CuS concentration, which is off by orders of magnitude. Most likely the reason for the discrepancy is that the formation rate could not be measured well because of fast reactions. Hence the reported coefficient (used in the simulations) is most likely much less than the actual coefficient (Luther and Rickard, 2005). Another factor contributing to the discrepancy could be because the model does not consider the reaction rate dependency on temperature. Fig. 7 shows the settled concentration patterns of ZnS, and FeS in the Izena cauldron. Coordinates (0, 0) show the Jade black smoker vent location. A high volume of particles is settled on the northeastern walls of the cauldron. They are settled at different depths according to the bathymetry of the region and their transport patterns. Therefore, it is important to have the bathymetry and the correct mineral PSD to predict

their distribution patterns. After three months the highest concentration of chemicals settled at the seabed are 1  106, 1  107, 1  104, 1  103, and 1  108 mol/m2 for PbS, CuS, ZnS, FeS, and FeS2 respectively. Almost all of the minerals tend to stay within the cauldron and therefore get settled in the same confined area. The distribution patterns however vary from each other depending on particle size and type. 4.2. Simulation of Dante hydrothermal Vent at the Endeavour field, Juan de Fuca ridge The Main Endeavour hydrothermal field is located in the axial valley of the Endeavour segment of the Juan de Fuca ridge in the northeastern Pacific Ocean (Butterfield et al., 1994). The depth of the field is approximately 2200 m. Main Endeavour field has many hydrothermal fluid releases and Dante vent is identified to be the largest and the most active with approximately ten high temperature black smokers on top of it (Xu and Di Iorio, 2012). Grotto, Crypto, Dudley and Hulk are examples of other high temperature vents located

Fig. 7. Mineral concentration at the bottom of the Izena cauldron after three months of simulation (a) ZnS, and (b) FeS.

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357 Table 3 Summary of comparison of simulated and observed concentrations and plume heights in Okinawa Trough. Parameter

High concentration patches variation Maximum FeS concentration Maximum CuS concentration

Simulated

Observed

Values Depth (m)

Values

e 880 ng/l 0.096 ng/l

References Depth (m)

1150e1300 e

1100e1300 Marumo and Hattori 1150e1350 1100 ng/l 1100e1300 (1999) 1150e1350 300 ng/l

1100e1300

near Dante (Butterfield et al., 1994). Marine Geoscience Data system (http://www.marine-geo.org/portals/ridge2000/) gives the location of Dante to be approx. 129.10
W and 47.95
N. Fig. 8 shows the location of Dante in Endeavour field. We simulated the formation and distribution of mineral particles in Dante black smokers.

351

4.2.1. Model input 4.2.1.1. Bathymetry of the region. In the Endeavour segment the hydrothermal venting is within the axial valley with approximately 1 km wide 10 km long area along the crest of the ridge (Thomson et al., 2003). The bathymetry of the axial valley area is shown in Fig. 9 and is used as input for depth in the model on a horizontal grid of size 50 m  50 m. 4.2.1.2. Variation of the ambient velocity, temperature, and salinity. Thomson et al. (2003) describes that the current meter data at the Endeavour ridge has two flow structures. One is immediately above the ridge crest and the other at elevations 75e100 m above the floor of the valley. They explained that the background flow is approximately along the axis and frequently surpass 5 cm/s Baker and Massoth (1987) indicate that the mean velocity measured during a 27 day survey in the Endeavour segment was 1.5 cm/s at 170
C. The mean flow

Fig. 8. Location of the Dante black smoker plume (black star) in the Endeavour Field and the bathymetry of the axial valley region.

352

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

Fig. 9. Three-dimensional view of the bathymetry of the axial valley region.

within the axial valley in the Endeavour segment is 1e5 cm/s directing north (Thomson et al., 2003, 2005; Veirs et al., 2006). Veirs et al. (2006) measured the regional mean flow above the ridge crest to be 5 cm/s towards southwest. Three typical velocity profiles were given as input to the model in the north, central, and southern parts of the axial valley based on Thomson et al. (2003). They are shown in Fig. 10. The vertical and horizontal eddy diffusivities are taken as 0.0025 and 0.1 m2/s respectively (Feely et al., 1987). Fig. 11 shows the temperature and salinity profiles used for the simulations.

observations an equally divided volume PSD with sizes 10, 30, 50, 70, and 90 mm is used for all the chemical species in the simulation. 4.2.1.4. Release conditions of the fluid. Based on the data in Marine Geoscience Data system, the release depth of Dante

4.2.1.3. Fluid composition and particle size distribution (PSD). The initial composition of the vent fluid released from Dante is shown in Table 4. Feely et al.'s (1987) black smoker particles studies in the Juan de Fuca ridge show their sizes vary from 0.1 to 500 mm. The particle sizes for specific minerals are: calcopyrite 10 to 60 mm; anhydrite e 100e500 mm; sphalerite e 50e100 mm; fine grained phyrrotite e 0.1e5 mm; and barite - 5e40 mm. A high proportion of particles in the vent fluids from Endeavour segment are coarse grained (Baker and Massoth, 1987). Based on the above

Fig. 11. Salinity and temperature profiles of the axial valley in Endeavour segment (based on the data by Thomson et al., 2005).

Table 4 Concentration of selected chemicals in Dante Black smoker fluid (based on the data by Butterfield et al., 1994).

Fig. 10. Typical velocity profiles in the (a) Northern, (b) Southern, and (c) Central parts of the axial valley in Endeavour segment.

Chemical species

Molar weight

Composition of Dante fluid (mol/l)

H2S Fe2þ Zn2þ Pb2þ Cu2þ Ca2þ Ba2þ Mn2þ

34.081 55.845 65.409 07.210 63.546 40.078 137.327 54.938

0.003552500 0.000706875 0.000018125 0.000000000 0.000010875 0.026172500 0.000000000 0.000189225

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

black smoker is at 2175 m. The discharge velocity; release orifice diameter; release fluid density and temperature that represent the Dante hydrothermal vents are 0.3 m/s, 0.4 m, 667 kg/m3, and 325.7
C respectively (Xu and Di Iorio, 2012). Baker and Massoth (1987) give the regional hydrothermal fluid release temperature and density as 350
C and 690 kg/m3 respectively. 4.2.2. Results and discussion Fig. 12(a and b) show the maximum projected concentration plots in the water column (similar to the plots for Izena) of FeS, after one day of simulation. The simulation terminated at 5 days because the particles are transported outside of the study domain. Sectional concentration plots clearly show the varying distribution patterns of different sized mineral particles. The simulated plume height in Dante is approximately 150 m. The results show that most of the mineral particles are taken up to this height before they are released into the ambient ocean to travel along with the currents. Simulated plume height is within the comparable height ranges of the plumes in the region given in previous studies (Thomson et al., 2005; Baker and Massoth, 1987). Veirs et al. (2006) stated that all the focused plumes in the axial valley were higher than the axial valley depth, which is around 100 m. They explain that during quiescent conditions where the cross flow is about 1 cm/s, plume rise can be around 200 m and it can reduce down to 100 m under a typical peak cross flow of 10 cm/s. They also stated that during extreme velocity condition of 20 cm/s plume height would even go down to 50 m. Simulated plume heights from this study for 1 cm/s and 20 cm/s ambient velocities are 152 m and 51 m respectively. Fig. 12 shows that after 24 h of simulation the mineral particles have spread about 3 km horizontally in the water column spreading out of the axial valley. Baker and Massoth (1987) reported that the physical evolution of the plume is extending downstream as far as 7.5 km under a current of 2.1 cm/s at 205 0 from the north, within about 4.1 days. But in

353

the simulations here the ambient currents at heights about 150 mab are considered to be 5 cm/s at 290 0 (Veirs et al., 2006). Therefore, the plume extent after the same time duration is almost double the size of that observed by Baker and Massoth (1987). In Fig. 12(a) streaky pattern of concentrations show the distribution patterns of different sized mineral particles with non-linearly correlated settling velocities. The use of limited number of particles in a discrete PSD causes this noncontinuity of the particle cloud. Large particles in the water column settled closer to the release point due to high settling velocities while smaller particles were transported away by the ambient currents mainly towards the southwestern direction. The currents in the central and southern parts of the axial valley dominate towards the north along the valley within 15e50 mab but at 100 to 250 mab they dominate towards the south (Thomson et al., 2003). Therefore, particles at higher elevations move towards the southeastern parts of the axial valley. Baker and Massoth (1987) also described plumes in the Endeavour segment being skewed to the south by the ambient currents about 100e200 mab as clearly seen in Fig. 12 (a). When particles reach lower levels while settling, they deviate from their initial path and tend to move towards the north in the valley (Fig. 12b). Some particles (~30 mm) closer to the source experience these northward currents early because they reach these lower levels while settling. Therefore, simulating the correct plume height and PSD is important to predict their correct distribution patterns. The height of Dante plume plays an important role in determining whether the mineral particles stay in the valley or released into the large ocean. Light attenuation plots in Kadko et al. (1990) in the Dante plume region show a similar particle distribution pattern closer to the source, as shown in Fig. 12(b). They explain that the particles smaller than 30 mm mostly spread away from the source while larger particles are separated fast from the plume settling close to it similar to the simulation results presented. Particle clouds of other minerals formed in Dante plume follow a distribution pattern more or less similar to Fig. 12. But the maximum concentration values have an order of

Fig. 12. Concentration of FeS in the water column (a) yz plane; (b) xy plane at 24 h.

354

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

magnitude differences and they are FeS-1  104; FeS21  109; ZnS-6  106; CuS-6  106; CaSO4-1  1010; and PMn-1  1015 mol/m3. The variation of concentration is dependent on their formation rates and initial concentrations, and ambient velocity variations. Very low water column concentrations of PMn are due to its slow formation rate (9.513  109 s1 Cowen et al., 1990) and low release concentration (2  1015 mol/m3). Concentration of ZnS and FeS particles settled at the bed within the axial valley after 5 days, are shown in Fig. 13 (a) and (b) respectively. Large particles of both ZnS and FeS settled closer to the release point while most of the finer particles are taken out of the axial valley and away from the ridge crest. High concentration patches show the settled particles of different sizes at different bed elevations. 5. Parametric analysis Parametric and sensitivity analyses are conducted to better understand the variability in the model results due to internal and external uncertainties of the model. Internal uncertainties include the numerical errors in model calculations and structure. External uncertainties are due to the variability in the input data and initial conditions. The ambient velocity, PSD, and release temperature are varied to study their impact on the model results. The purpose is mainly to identify the impact on mineral particle distribution in the water column, their settling patterns, and plume height variations. Simulations of Jade hydrothermal vent plume in the Okinawa trough are done with a 10% increase in the ambient velocity (described in Section 4.1.1.2). The simulated plume heights under the existing and increased ambient velocity conditions are compared in Table 5. It shows that the plume height decreased for all the increased velocity profiles and the percentage decrease in plume heights varies between 8.7 and 9.3%. The release temperature is varied from 200 to 400
C and simulations of the Jade plume are done to compare the plume

Table 5 Comparison of plume heights for the increased ambient velocity case. Ambient velocity, Vamb (cm/s)

Plume height (m) V ¼ Vamb

V ¼ 1.1 Vamb

4 5 6 7 8 9

59.4 48.04 40.21 34.54 30.23 26.83

54.21 43.79 36.73 31.4 27.5 24.34

Decrease in plume height (%) 8.7 8.8 8.7 9.1 9.0 9.3

heights for different cases. The results are shown in Table 6. The plume height is increased when the release temperature is increased. This is due to the decrease in plume density at high temperature, which in turn increase the buoyancy force on the plume. Plume temperature decreases fast due to ambient coldwater entrainment. Therefore, the initial low plume density won't last for a significant period to cause a drastic plume height change. For the cases tested the percentage plume height increase, varies between 14.7 and 21.5%. It can be observed from the parametric analyses that the variation in the plume heights of two plumes with two different higher release temperatures, is lower than two plumes that are released at lower temperatures with same difference in release temperature (Table 6). Low temperature plumes are denser than the high temperature plumes. Higher the initial plume density lesser the initial buoyancy force on the plume. A lower buoyancy force causes slower plume rise, resulting in lesser shear entrainment. Therefore, the rate at which the plume reaches its neutral buoyancy level is slower for a low buoyancy plume. Also to be noted here is that the shear entrainment is a function of the square of the plume velocity. As a result, at higher velocities the entrainment rate is much higher, resulting in higher rate of plume density increase. Therefore, two plumes with close but high release temperatures will not have much difference in their heights. However, due to low entrainment rate at lower velocities the plume density increase rate is lower. Therefore, two plumes with close low release temperatures will have bigger

Fig. 13. Concentration of bed settled particles after five days (a) ZnS, (b) FeS.

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357 Table 6 Comparison of plume heights for the increased release temperature case. Release temperature (
C)

Plume height (m)

Change in the plume height (%)

200 250 300 320 350 375 400

156.5 183.4 204.0 202.2 212.9 210.2 199.4

Base case 14.7 23.3 22.6 26.5 25.5 21.5

Table 7 Particle size distributions (PSD) used in the simulations. Particle size (mm)

100 90

70

50

30

20

10

7

3

Distribution No A 5% 10% 20% 35% 30% B 5% 10% 20% 35% 30% C 100%

difference in their heights to reach their respective neutral buoyancy levels. Another set of simulations of Jade plume is done considering three PSDs A, B, and C as shown in Table 7. In PSD A majority of particles vary in sizes between 30 and 70 mm while

355

in PSD B most particles vary in sizes between 3 and 10 mm. PSD C has only 20 mm particles. The bed settled FeS mineral concentration patterns for PSDs A, B, and C after three months are shown in Figs.14(a), (b), and (c) respectively. Settled particle concentration plots have three distinct patterns. PSD A had large particles compared to the other two distributions and they are settled close to the release point at (0, 0). Conversely the PSD B, which had majority of small particles, has a distributed particle-settling pattern across the Izena cauldron. The PDS C has a particle-settling pattern which lies in between the above two patterns as expected. The different settling patterns are formed mainly due to the differences in particle settling velocities. Large particles with high settling velocities settle closer to the plume. Smaller particles, which have comparatively smaller settling velocities, are taken away by the ambient currents and settled away from their release points. It is evident that model results change significantly according to the PSD used. 6. Summary and conclusions MOHTV model (Dissanayake et al., 2014) has been improved and a series of new tests and simulations in the Asia Pacific region are conducted. These include new comparisons of simulations

Fig. 14. Concentration of bed settled FeS after three months of simulations for the particle size distributions (a) A, (b) B, and (c) C.

356

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357

with field data, simulating some field scenarios, and a parametric analyses to look at how the changes in input data affects the results. Simulations of hydrothermal vent plumes and their mineral particle distribution in two black smokers are carried out using the improved version of the model. Mineral formation, their deposition rates, and patterns are important in estimating the locations and extents of the mineral accumulations in hydrothermal vent fields. These estimates can be used as a guide for decision making related to excavation of economic grade minerals such as PbS, ZnS, CuS, Au, and Ag based minerals, some of which are found in abundance. The model simulates the formations of FeS, PbS, ZnS, CuS, FeS2, CuFeS2, and several other chemicals and their distribution in the ambient water along with bed settling patterns. MOHTV model considers the bathymetry, and spatial and temporal velocity variation in the region to calculate the mineral distribution patterns. Mineral deposition patterns and water column mineral concentrations in the Izena cauldron due to Jade black smoker, and in the axial valley of Endeavour segment due to Dante black smoker are presented here. Simulation results match qualitatively well with the observed conditions in the field. Parametric analyses of the model shows that the mineral particle settling patterns significantly vary with the PSD used. Large particles tend to settle closer to the plume while smaller particles are distributed to far away distances from their source. These distances are affected by how much the plume rises before the particles are swept away by advection diffusion. Ambient current patterns vary from one level to another in the ambient water, affecting the distribution of particles. Therefore, the inclusion of the plume dynamics stage is an important aspect. Plume height variation due to increase in temperature and ambient velocity are also presented to show how the model results vary with the ambient conditions. A vent fluid with high release temperatures has low densities so their plume heights are higher than low temperature releases as a result of higher buoyancy forces. Large ambient velocities increase the ambient water entrainment into the plume, therefore, decreasing its height. Plume height is an important parameter that decides the particle distribution patterns in the water column. Bathymetry in the region plays an important role in the deposition pattern of the particles as well. This was clearly seen in the simulations of hydrothermal vents in the Izena cauldron and Main Endeavour ridge. Acknowledgments This work is supported by Marine Biological Research Institute of Japan, Tokyo and Meijo University, Tenpaku, Japan. References Anantharaman, K., Breier, J.A., Sheik, C.S., Dick, G.J., 2013. Evidence for hydrogen oxidation and metabolic plasticity in widespread deep-sea sulfuroxidizing bacteria. PNAS 110, 330e335. Baker, E.T., Massoth, G.J., 1987. Characteristics of hydrothermal plumes from two vent fields on the Juan de Fuca ridge, Northeast Pacific Ocean. Earth Planet. Sci. Lett. 85, 59e73.

Bemis, K., Silver, D., Rona, P., Cowen, J., 2006. A particle sedimentation model of buoyant jets: observations of hydrothermal plumes. Cah. Biol. Mar. 47, 379e384. Breier, J.A., Toner, B.M., Fakra, S.C., Marcus, M.A., White, S.N., Thurnherr, A.M., German, C.R., 2012. Sulfur, sulfides, oxides and organic matter aggregated in submarine hydrothermal plumes at 9
5000 N East Pacific Rise. Geochim. Cosmochim. Acta 88, 216e236. Butterfiel, D.A., McDuff, R.E., Mottl, M.J., Lilley, M.D., Lupton, J.E., Massoth, G.J., 1994. Gradients in the composition of hydrothermal fluids from the Endeavour segment vent field: phase separation and brine loss. J. Geoph. Res. 99, 9561e9583. Clift, R., Grace, J.R., Weber, M.E., 1978. Bubbles, Drops, and Particles. Academic Press, NY. Cowen, J.P., Massoth, G.J., Feely, R.A., 1990. Scavenging rates of dissolved manganese in a hydrothermal vent plume. Deep Sea Res. Part A Oceanograp. Res. Papers 37 (10), 1619e1637. Dasanayaka, L.K., Yapa, P.D., 2009. Role of plume dynamics phase in a deepwater oil and gas release model. J. Hydro-environ. Res. 2, 243e253. Dissanayake, A.L., Yapa, P.D., Nakata, K., 2014. Modeling of hydrothermal vent plumes to assess the mineral particle distribution. J. Hydraul. Res. 1, 49e66. Dymond, J., Roth, S., 1988. Plume dispersed hydrothermal particles: a timeseries record of settling flux from the endeavour ridge using moored sensors. Geochim. Cosmochim. Acta 52 (10), 2525e2536. ERI, 2012. Preliminary Environmental Impact Assessment on Harvesting Minerals in Hydrothermal Vents. Report in Japanese by Environmental Research Institute-Kankyo Sogo Technos. March. Feely, R.A., Lewison, M., Massoth, G.J., Robert-Baldo, G., Lavelle, J.W., Byrne, R.H., Von Damm, K.L., Curl Jr., H.C., 1987. Composition and dissolution of black smoker particulates from active vents on the Juan de Fuca ridge. J. Geophys. Res. 92, 11,347e11,363. German, C.R., Sparks, R.S.J., 1993. Particle recycling in the TAG hydrothermal plume. Earth Planet. Sci. Lett. 116, 129e134. Ginster, U., Mottl, M.J., Herzon, R.P., 1994. Heat flux from black smokers from on the endeavour and cleft segments, Juan de Fuca ridge. J. Geophys. Res. 99, 4937e4950. Glasby, G.P., Notsu, K., 2003. Submarine hydrothermal mineralization in the Okinawa Trough, SW of Japan: an overview. Ore Geol. Rev. 23, 299e339. Granbakken, D., Haarberg, T., Rollheim, M., Ostvold, T., Read, P., Schmidt, T., 1991. Scale formation un reservoir and production equipment during oil recovery. III. A kinetic model for the precipitaion/dissolution reactions. Acta Chem. Scand. 45, 892e901. Halbach, P., Nakamura, K., Wahsner, M., Lange, J., Sakai, H., K€aselitz, L., Hansen, R.D., Yamano, M., Post, J., Prause, B., Seifert, R., Michaelis, W., Teichmann, F., Kinoshita, M., M€arten, A., Ishibashi, J., Czerwinski, S., Blum, N., 1989. Probable modern analogue of Kuroko-type massive sulfide deposits in the Okinawa Trough back-arc basin. Nature 338 (6215), 496e499. Halbach, P., Hansmann, W., Koppel, V., Pracejus, B., 1997. Whole-rock and sulfide lead-isotope data from the hydrothermal JADE field in the Okinawa back-arc trough. Miner. Deposita 32, 70e78. Ishibashi, J.I., 1991. Geochemical Studies of Hydrothermal Fluids in the Middle Okinawa Trough Back Arc Basin. University of Tokyo. PhD thesis. Kadko, D.C., Rosenberg, N.D., Lupton, J.E., Collier, R.W., Lilley, M.D., 1990. Chemical reaction rates and entrainment within the Endeavour Ridge hydrothermal plume. Earth Planet. Sci. Lett. 99, 315e335. Kawagucci, S., Shirai, K., Lan, T.F., Takahata, N., Tsunogai, U., Sano, Y., Gamo, T., 2010. Gas geochemical characteristics of hydrothermal plumes at the Hakurei and Jade vent sites, the Izena Cauldron, Okinawa Trough. Geochem. J. 44, 507e518. Lee, J.H.W., Cheung, V., 1990. Generalized Lagrangian Model for buoyant jets and in current. J. Environ. Eng. 116 (6), 1085e1105. Luther III, G.W., Rickard, D., 2005. Metal sulfide cluster complexes and their biogeochemical importance in the environment. J. Nanoparticle Res. 7, 389e407.

A.L. Dissanayake et al. / Journal of Hydro-environment Research 8 (2014) 343e357 Marumo, K., Hattori, K.H., 1999. Seafloor hydrothermal clay alteration at Jade in the back-arc Okinawa Trough: Mineralogy, geochemistry and isotope characteristics. Geochim. Cosmochim. Acta 63 (18), 2785e2804. Nakashima, K., Sakai, H., Yoshida, H., Chiba, H., Tanaka, Y., Gamo, T., Ishibashi, J.I., Tsunogai, U., 1995. Hydrothermal mineralization at the Okinawa Trough. In: Sakai, H., Nozaki, Y. (Eds.), Biogeochemical Processes and Ocean Flux in the Western Pacific. Terra Scientific Publ, Tokyo, pp. 487e508. Tao, T., Rosswog, S., Bru¨ggen, M., 2013. A simulation modeling approach to hydrothermal plumes and its comparison to analytical models. Ocean. Model. 61, 68e80. Thomson, R.E., Mihaly, S.F., Rabinovich, A.B., McDuff, R.E., Veirs, S.R., Stahr, F.R., 2003. Constrained circulation at Endeavour ridge facilitates colonization by vent larvae. Nature 424 (6948), 545e549. Thomson, R.E., Subbotina, M.M., Anisimov, M.V., 2005. Numerical simulation of hydrothermal vent-induced circulation at Endeavour ridge. J. Geophys. Res. 110, C01004. Tivey, M.K., 2007. Generation of seafloor hydrothermal vent fluids and associated mineral deposits. Oceanography 20, 50e65. Toner, B.M., Fakra, S.C., Manganini, S.J., Santelli, C.M., Mar- cus, M.A., Moffett, J.W., Rouxel, O., German, C.R., Edwards, K.J., 2009.

357

Preservation of iron (II) by carbon-rich matrices in a hydrothermal plume. Nat. Geosci. 2, 197e201. Toner, B.M., Marcus, M.A., Edwards, K.J., Rouxel, O., Ger- man, C.R., 2012. Measuring the form of iron in hydrothermal plume particles. Oceanography 25, 209e212. Veirs, S.R., McDuff, R.E., Stahr, F.R., 2006. Magnitude and variance of nearbottom horizontal heat flux at the Main Endeavour hydrothermal field. Geochem. Geosyst. 7, Q02004. Walker, S.L., Baker, E.T., 1988. Particle size distributions within hydrothermal plumes over the Juan de Fuca ridge. Mar. Geol. 78, 217e226. Watanabe, S., Tsurushima, N., Kusakabe, M., Tsunogai, S., 1995. Methane in izena cauldron, okinawa trough. J. Oceanogr. 51, 239e255. Xu, G., Di Iorio, D., 2012. Deep sea hydrothermal plumes and their interaction with oscillatory flows. Geochem. Geophys. Geosyst. 13, Q0AJ01. http:// dx.doi.org/10.1029/2012GC004188. Yapa, P.D., Zheng, L., 1997. Simulation of oil spills from underwater accidents I: model development. J. Hydraul. Res. 35 (5), 673e688. Zheng, L., Yapa, P.D., 2000. Buoyant velocity of spherical and nonspherical bubbles/droplets. J. Hydraul. Eng. 126 (11), 852e854. Zheng, L., Yapa, P.D., Chen, F., 2003. A model for simulating deepwater oil and gas blowoutsepart I: theory and model formulation. J. Hydraul. Res. 41 (4), 339e351.