The Scripps Guyot has a nearly flat top at a depth of ~1200–1500 m and rises approximately 4000 m above the surrounding abyssal plain ( Figures 1A, B ). The summit extends 40 km in the north-south direction and 25 km in the east-west direction. The Japan Organization for Metals and Energy Security (JOGMEC) was issued an exploration contract for the area by the ISA (International Seabed Authority, 2024). In this study, the seamount summit was defined as the seafloor at a depth of ≤1500 m with a slope of less than 2°, as calculated using QGIS ( Supplementary Figure S1 ).

Modeling around the seamount was carried out using the numerical model Delft3D (Deltares, 2022). The mathematical principle of Delft3D is to solve the Reynolds-averaged Navier-Stokes equations for incompressible fluids using finite difference methods. The model was set to be three-dimensional and hydrostatic. Turbulent processes were modeled based on eddy viscosity. The vertical eddy viscosity and eddy diffusion coefficients were calculated using the k-ϵ model, a method that derives turbulent kinetic energy k and turbulent kinetic energy dissipation rate ϵ obtained from the transport equations (Deltares, 2022).

The model area was 4.2° × 4.4° (157.3–161.5° E, 21.6–26.0° N) and the area near the seamount (158.6–160.0° E, 23.15–24.4° N) was increased in resolution by two-way nesting ( Supplementary Figure S2 ). The horizontal resolution was set to Δ x = Δ y = 1/50° (≈ 2.2 km) in the outer region and Δ x = Δ y = 1/250° (≈ 0.44 km) in the inner region. The vertical resolution was set at 36 layers in the Z-coordinate system, extending from the surface to the abyssal plain of 5500 m depth (see Supplementary Figure S3 for details). The finest vertical resolution was Δ z = 27.5 m directly above the summit seafloor and the coarsest resolution was Δ z = 440 m directly above the abyssal plain. The calculation target period spanned 15 days, from December 1 to December 15, 2019. The time step was set at 0.6 minutes. The target period was preceded by a 10-day spin-up period, resulting in a total calculation period from November 21 to December 15, 2019.

The model focused on tidal currents, with the main forcing provided by the tides at the open boundaries. Note that this model does not include the effects of background baroclinic currents and mesoscale eddies. To avoid model complexity, tidal components were calculated only for M2 (semi-diurnal) and K1 (diurnal), which are more dominant than inertial oscillation components in field flow measurements at the summit ( Supplementary Figure S4 , see below for observation details). The tidal levels for the open boundary conditions were taken from the tidal model TPXO9-atlas (1/30° resolution) (Egbert and Erofeeva, 2002). In addition, Delft3D was configured to include tide generating forces at open boundaries, incorporating the contribution of gravitational forces to the motion of large water masses in the deep sea (Deltares, 2022). The direction of tidal waves propagating over the model area was predominantly from west-northwest to east-southeast ( Supplementary Figure S5 ). To weaken the reflection of waves propagating from the inside of the model to the boundary, a reflection parameter α = T H / g was set at the western and southern lateral boundaries, where T is the time it takes for a free surface wave to travel from one open boundary to the boundary in the opposite direction, H is the water depth, and g is the gravitational acceleration. Multiplying α by the time derivative of the Riemann invariant and adding it to the water level resulted in the water surface boundary condition F ζ   ( t ) . That is, F ζ   ( t ) = ζ + α ∂ ∂ t { U ± 2 g H } , where ζ is the free surface elevation above the still water level, and U is the horizontal velocity (Deltares, 2022).

Water temperature and salinity for initial and open boundary conditions were taken from in situ CTD data at the base observation site ( Figure 1B ). CTD profiles, from the surface to a depth of ~5200 m, were obtained from 7:00 to 12:00 on October 24, 2019, by vertically deploying an SBE 911 plus (Sea-Bird Scientific) from the research vessel, with a measurement interval of 1 m. Meteorological conditions, i.e., air temperature, relative humidity, fraction cloud coverage, and net solar radiation, were obtained from the meteorological reanalysis data NCEP/NCAR Reanalysis 1 (2.5° resolution) (Kalnay et al., 1996) and used to calculate heat fluxes between the atmosphere and sea surface. Topographic data were obtained from the GEBCO 2022 Grid (1/240° resolution) (GEBCO Compilation Group, 2022). In order to focus specifically on the interaction between the topography of Scripps Guyot and flow, convex topography was removed from the model domain outside Scripps Guyot and two small adjacent seamounts ( Figure 1B ). Topography was smoothed by a moving average of 10 neighboring cells.

Modeled tidal currents were validated by comparison of modeled results with in situ flow observations from mooring systems at three sites: the summit, slope, and base ( Figure 1B ). A total of 13 single point current meters (DW-Aquadopp, Nortek AS) and two 75 kHz acoustic Doppler current profilers (ADCPs) (WorkHorse, Teledyne RD Instruments) were installed. The instruments and observed depths for each site are illustrated in Figure 1C . The observation period was from October 21, 2019 to August 11, 2020. The measurement interval was one hour for both instruments. There were 30 observation layers for ADCPs, each with a thickness of 20 m. The accuracies of both instruments were ± 0.5 cm/s. For more details on field flow observations, see Saito et al. (2024). In the observations and model results, the tidal ellipses, which represent the amplitudes and directions of the tidal currents by their magnitude and orientation, were determined by the analysis package UTide for Python (Codiga, 2011).

The physical properties of the plume from modeled mining operations were set based on a small-scale crust excavation test at Takuyo-Daigo Seamount in 2020 (Saito et al., 2023; Washburn et al., 2023). In the test, the excavator was equipped with a drum cutter head and a cyclone tank for material collection. Crusts, with accompanying rocks and sediments, were collected from an area of ~100 m². During excavation, ore and seawater were separated inside the cyclone tank, and seawater was discharged through the outlet 3.75 m above the seafloor. Fine particles, primarily composed of crustal material, were released along with seawater, forming sediment plumes near the seafloor. The discharge rate of the crust particles was estimated to be 0.022 m³/h. Seawater containing the plume was collected using Niskin bottles attached to the ROV, and the size distribution and settling velocity of the particles in the plume was determined (Suzuki et al., in preparation). For more details on observations of the excavation test, see Saito et al. (2023) and Washburn et al. (2023). The discharge height, discharge speed, and particle size distribution of the plume in the excavation test were used as settings in the simulation.

The simulation focused on the dispersion and deposition of plume particles in the vicinity of the mining site. The modeled plume consisted of particles larger than 30 μm in diameter that accounted for 81% of the plume volume in samples collected during the crust excavation test ( Table 1 ). Finer particles (i.e., less than 30 μm), which were excluded from this study, are expected to remain suspended for longer periods and to disperse over a wider area at very low concentrations (Rolinski et al., 2001; Gillard et al., 2019; Spearman et al., 2020), making them difficult to observe and model. This study focused on relatively high concentrations of particles, which are of particular concern for the ecological impacts (Glover and Smith, 2003; Washburn et al., 2019; Amon et al., 2022). Four particle sizes were used for simulated discharges: 30 μm, 50 μm, 90 μm, and 230 μm ( Table 1 ), which were representative sizes of the sediment plume measured in the excavation test. Settling velocities ranged from 0.954 mm/s for 30 μm particles to 56.1 mm/s for 230 μm particles ( Table 1 ). Particles were discharged from 3.75 m above the seafloor, the location of the cyclone tank outlet. Particles were assumed to have been deposited when they settled 3.75 m below the discharge height. The effects of resuspension and of terrain features smaller than the model’s spatial resolution were not considered. With the above setup, particles of all four sizes were always deposited within 1.09 hours of discharge. This simulation was targeted for the flat-top with slopes of 2° or less, and the process of sliding downslope was not included. Note that these simplifying assumptions may lead to underestimation of the spatial extent of deposit thickness.

The simulations were performed by Lagrangian particle tracking analysis and used the Python package Parcels (Delandmeter and van Sebille, 2019). The trajectories of the particles were calculated by the following equation: X ( t + Δ t ) = X ( t ) + ∫ t t + Δ t v ( x ,   τ ) d τ + Δ X b ( t ) , where X is the position of the particle, v ( x ,   τ ) is the three-dimensional velocity field at that location, and Δ X b ( t ) is a change in vertical position due to settlement. The velocity field was obtained from a hydrodynamic model, with linear interpolation applied for values finer than the model’s spatial and temporal resolution. The focus was on particle advection, so random diffusion due to turbulence or thermal motion was not considered. The calculation time step was 1 minute. At the seamount summit, particles were discharged from a total of 604 locations spaced at 880 m intervals. Particles representing the four particle sizes were emitted from all emission points, one per hour. The simulation period was set to one week, from December 1 to December 7, 2019, covering a tidal cycle from spring tide to neap tide ( Supplementary Figure S6 ). Thus, a total of 672 particles were calculated for each site and 405,888 particles across all sites. As the hydrodynamic model was primarily driven by tidal forcing, the simulation results were expected to exhibit periodicity and to be free from statistical noise (see also the Results section).

The “deposition distance” was defined as the distance from the discharge point to the farthest deposited particle. The deposition thickness T for each 50 m × 50 m area was calculated using the following equation: T = ∑ s ( V s / A ) , where s is the particle size, V s is the volume of deposited particles of size s, and A is the area of 50 m × 50 m. The volume of particles of each size released per hour, v s , was determined by: v s = v × ( P s / 100 ) , where v is the plume emission rate of 0.022 m³/h from the excavation test (Saito et al., 2023; Washburn et al., 2023), and P s is the percentage of the plume volume for each particle size. For example, particles of 30 μm, which accounted for 33.1% of the plume volume, had a calculated emission rate of v s = 0.0073 m³/h.

Modeled tidal currents were stronger over the seamount summit compared to the surrounding area ( Figure 2 ), indicating the occurrence of internal tides (Baines, 2007; Lavelle and Mohn, 2010) here. At 3.75 m above the seafloor, which corresponds to the plume discharge depth, the tidal currents at the summit were ≤7 times stronger than those outside the seamount.

Tidal current strengths were uneven across the summit. At 3.75 m above the seafloor, the amplitude of M2 tidal currents varied by ≤9.6 times (ranging from 0.5 cm/s to 4.8 cm/s), while the amplitude of the K1 tidal currents varied by ≤8.0 times (ranging from 0.4 cm/s to 3.6 cm/s) within the summit. The tidal currents were strongest at the northern tip of the summit for both M2 and K1. The distribution patterns of the tidal current strength differed between M2 and K1. The M2 tidal currents were relatively stronger (>3 cm/s) at the western edge of the summit, whereas K1 tidal currents were relatively stronger (>2 cm/s) at the center of the summit.

The modeled tidal currents were validated by comparing them with field observations. For both M2 and K1 tidal components, field observations showed that the tidal currents were ≤4.1 times stronger at ≤200 m above the seafloor at the summit than in other areas, and the model was able to reproduce this trend ( Figure 3 ). At 1100 m depth, the difference in tidal current amplitude between the model and observations was less than 1.0 cm/s for both the M2 and K1 ( Supplementary Tables S1, S2 ). The modeled M2 amplitude (2.2 ± 0.06 cm/s, ± indicates 95% confidence interval) was 0.1 cm/s larger than the field observation value (2.1 ± 0.1 cm/s), while the modeled K1 amplitude (2.2 ± 0.1 cm/s) exceeded the field observation value (1.4 ± 0.1 cm/s) by 0.8 cm/s. The difference between the model and the field observations for the direction of the maximum tidal current was within ~30° for both M2 and K1. The difference was 5.7°for M2 (model: 169.9 ± 2.6°, observation: 164.2 ± 7.1°) and 31.1°for K1 (model: 47.8 ± 9.1°, observation: 16.7 ± 65.2°).

Deposition distances varied by a factor of ≤6.5 depending on the particle discharge points, ranging from 122 m to 798 m ( Figures 4 , 5 ). The average deposition distance among all the discharge points was 346 m, with a standard deviation (SD) of 121 m. The median was nearly identical at 347 m. The deposition distances corresponded to the tidal current strength at the discharge points. Longer deposition distances occurred at the northern tip, where both M2 and K1 were strong ( Figure 2 ). Short deposition distances occurred predominantly from the eastern margin, where both M2 and K1 were weak ( Figure 2 ). The deposition distances were highly correlated with M2 (R = 0.77; p < 0.001) and K1 (R = 0.66; p < 0.001) amplitudes. Due to the periodicity of the tidal currents, particles were deposited in an elliptical region centered on each discharge point ( Figure 5 ), with no particles deposited at exceptionally distant locations that could introduce statistical noise.

Plumes dispersed farther and resulted in thinner depositional layers at discharge points with stronger tidal currents, while being closer and thicker at discharge points with weaker tidal currents ( Figure 6 ). At the site with the maximum tidal currents ( Figure 6A ), the depositional thickness was ≤0.31 mm within 50 m from the plume discharge points, ≤0.045 mm at 50–150 m, and ≤0.0087 mm beyond 150 m. In contrast, at the site with the weakest tidal currents ( Figure 6D ), the depositional thickness nearly doubled, with ≤0.57 mm within 50 m from the discharge points, ≤0.074 mm at 50–150 m, and no deposition occurred beyond 119 m.

Relatively coarse particles (≥90 μm) were deposited within ~7 minutes ( Table 1 ), forming thicker sediment layers (≥0.3 mm) within ~50 m of all discharge points. The finer the particles, the longer they remained in suspension, leading to greater variation in deposition distances among the discharge points ( Supplementary Figure S7 ). For 230 μm particles, the difference between maximum and minimum deposition was 30 m, whereas for 30 μm particles, it was 676 m ( Supplementary Table S3 ). The SD of deposition distance was only 5 m for 230 μm particles but 121 m for 30 μm particles. The average deposition distances per particle size were 15 m for the 230 μm particles, 47 m (SD = 16 m) for the 90 μm particles, 131 m (SD = 44 m) for the 50 μm particles, and 346 m for the 30 μm particles.