**1. Introduction**

The Gulf of Mexico continental margin generates >1.7 million barrels of oil per day, through >3500 oil platforms. The northern Gulf of Mexico houses >45,000 km of underwater pipes that may be exposed to structural damage from extreme oceanic events. During the passage of a hurricane, storm waves can exceed 10 m in height, resuspending seafloor sediment and potentially liquefying the seafloor. Both of these mechanisms may induce sediment turbidity currents, and in fact, ~5% of the underwater petroleum pipes appear to be broken or damaged by sudden powerful turbidity currents (BOEM pers. comm. 2015). For example, in 2004, a large sediment failure in the wake of Hurricane Ivan toppled an oil platform offshore of the Gulf of Mexico and moved it ~0.17 km downslope, initiating oil and gas leaks at a water depth of 140 m [1]. Leakage from such offshore oil and gas infrastructure puts at risk about 40% of the USA's coastal and estuarine wetlands, which are vital to recreation, agriculture, and a \$1B/y seafood industry [2].

Turbidity currents are important transport mechanisms in submarine canyons [3,4], such as the Mississippi and the De Soto Canyons, which incise the continental slope offshore of the Mississippi Delta. Several processes have been shown to have the potential to generate turbidity currents, including physical oceanographic mechanisms. Internal wave breaking on the upper slope may mobilize seafloor sediment [5]. Wave-current interactions on continental shelves during large oceanic storms can initiate wave-supported gravity flows [6]. Continental slope deposits may experience sediment failure triggered by sediment loading and over-steepening, and aided by excess pore pressure brought on by ground accelerations [7,8]. Localized events may grow into sustained currents via a self-amplifying 'ignition' process with accelerating erosion and entrainment of sediment from the seafloor [9,10].

While the relative importance of these mechanisms in the northern Gulf of Mexico remains to be seen, evidence points to the potential for oceanic storms to mobilize sediment there, either during the passage of moderate storms [11] or more extreme events such as hurricanes [12]. Analysis of sediment deposits indicated that most (~75%) of the sediment budget of the Mississippi Canyon could be attributed to delivery during major hurricanes, likely through gravity-driven transport [13]. Several processes affect the seafloor during short-lived hurricane passages, including sediment mass failures, erosion, and suspension. For example, mudflows in the Mississippi Delta area, triggered by the 1969 Category 5 Hurricane Camille, destroyed the offshore platform SB-70B. The seafloor at a depth of about 90 m moved more than 1000 m downslope with soil flows up to 30 m in thickness [14]. Seafloor shear stresses from waves and currents of up to 1 <sup>N</sup>/m<sup>2</sup> were monitored at a depth of 90 m during the 2004 Category 5 Hurricane Ivan, reaching the critical shear stress for fine gravel [15]. The Ivan event lifted suspended sediment as high as 25 m in the water column and eroded the seafloor up to 0.30 m vertically over more than 500 km2, thus removing hundreds of millions of tons of sediment with deposits at the shelf edge and upper slope [16], and additionally causing apparent damage to oil infrastructure [1]. Evidence of the effects of large storms at grea<sup>t</sup> depth in the Gulf of Mexico has been seen in conjunction with other hurricanes, such as Hurricane Georges in the Mississippi Canyon [12]; Hurricane Frederic in the De Soto Canyon [17]; and Hurricane Allen [18]. Rapid loading of sections of the seafloor locally enhances the prospects for gravitational slope failures, given the associated rapid increase in pore pressures and reduction in effective sediment strengths [7]. Process-based numerical modeling offers a way to study such ephemeral high-energy processes.

Studies of sediment dispersal on continental margins, including the northern Gulf of Mexico, have typically focused on an individual component of the transport path such as gravity-driven transfer via canyons, shelf resuspension, or flood plume dispersal. For example, numerical models for suspended sediment transport have been developed and applied to the northern Gulf of Mexico [19–22], but these types of suspended transport models have not been directly linked to turbidity current models. This paper describes a numerical capability to simulate the transport of sediment, from fluvial sources, to the continental shelf, the deeper continental slope, and ultimate depocenters. Accounting for these sediment transport pathways, and the hazards that they present, is a problem of multi-scale physics, ranging from continental-scale drainage basins that deliver sediment to the sea, to shelf-wide storm systems that mobilize and redistribute sediment, to small-scale turbulent motions that a ffect turbidity current generation and structure.

This paper describes a loosely coupled numerical workflow that has been developed to address land-sea pathways for sediment routing of terrestrial and coastal sources, across the continental shelf, and ultimately down the continental slope and canyons of the northern Gulf of Mexico. Few studies have attempted to integrate the various transport mechanisms into a single comprehensive framework, accounting for the multi-scale physics that are relevant to the full sediment transport pathway. The workflow was used to explore conditions that may trigger episodes of sediment transport onto the continental slope and to evaluate two hypotheses: (1) episodic sediment transport down a submarine canyon is fed by sediment input at the canyon head from wave and current resuspension, and (2) turbidity currents are triggered by failures near the shelf-slope break and are likely to pass into the canyons of the continental slope. Simulation results were based on oceanographic and meteorological conditions that could impact the generation of turbidity currents. The workflow (Figure 1) includes modules that:

(1) Simulate the fluvial delivery of water and sediment into the Gulf of Mexico with the Water Balance Model-Sediment (*WBMsed*) and as augmented by USGS (US Geological Survey) and USACE (US Army Corps of Engineers) gauged river data;

(2) Develop domain grids and bathymetry for ocean circulation and sediment transport models;

(3) Compute spatial griddings of seabed sediment texture from *dbSEABED,* and of topographic channelization from the bathymetry, for use in sediment transport and seabed failure models;

(4) Employ a high resolution (10 km) spectral wave action model (*WaveWatch III*®*)* driven by *GFDL-GFS* (Geophysical Fluid Dynamics Laboratory–Global Forecast System) winds for use in the ocean and sediment transport models;

(5) Calculate hourly-timescale ocean circulation at a spatial resolution of a few kilometers via the Regional Ocean Modeling System (*ROMS*) forced with *ECMWF* (European Centre for Medium-Range Weather Forecasts) *ERA* (ECMWF Re-Analysis) winds;

(6) Represent seafloor resuspension and transport at the same resolution as *ROMS*' hydrodynamics using the Community Sediment Transport Modeling System (*CSTMS*);

(7) Apply seabed mass-failure and a sediment suspension model (*HurriSlip*) to determine failure and ignition locations, and the conditions to be used as input to the turbidity current model;

(8) Develop and deploy a Reynolds-averaged Navier–Stokes (*RANS*) model (*TURBINS*) to route sediment flows down the Gulf of Mexico slopes and canyons, providing estimates of bottom shear stress needed for ascertaining possible damage to o ffshore infrastructure.

**Figure 1.** Workflow showing models employed and boundary data usage. Models and data systems discussed in detail in the text. The white text identifies whether the models ran were Point models, 2D horizontal plan-view; 2D vertical transect or 3D.

#### **2. Materials and Methods**

Section 2.1 describes the northern Gulf of Mexico, where the model workflow was applied. Section 2.2 provides descriptions and methods for each component of the workflow, noting how the various components can interact with one another. Section 2.3 outlines the implementation of the suite of models used to evaluate sediment routing in the northern Gulf of Mexico.

### *2.1. Environmental Setting*

The Mississippi River drains 41% of the continental United States before entering the northern Gulf of Mexico (Figure 2). The discharge of the Mississippi River is regulated so that approximately 70% of it enters the Gulf through its main Mississippi River channel, while the remaining 30% enters through the Atchafalaya River channel [23]. Average modern-day sediment loads of the Mississippi and Atchafalaya Rivers are 115 and 57 Mt/yr, respectively [23]. Sand is deposited near the river mouths while most of the remaining suspended silts and muds are dispersed more widely [19,24,25]. Rapid delta progradation during the Holocene has narrowed and steepened the continental shelf (~20 km wide, ~0.4◦ gradient). The Mississippi Canyon, which cuts into the continental slope to the west of the bird-foot delta has been implicated as a conduit for shelf sediment during large storms [12,26].

A fair amount is known about suspended sediment dispersal on the Gulf of Mexico continental shelf. Frontal systems that occur frequently during winter months can create energetic waves and currents that cause significant sediment transport [27,28]. Wave contributions dominate the bed stresses on the continental shelf offshore of the Mississippi Delta, but fairweather waves are typically capable of mobilizing the seabed only in the surf and nearshore zones [19]. During extreme oceanic storms, however, deep-water wave heights exceed 10 m, with nearshore waves east of the bird-foot delta reaching 9 m in 15 m of water during Hurricane Ivan [29]. Storm waves, either from moderate storms or intense but infrequent hurricanes, have been shown to mobilize sediment mass failures on the Mississippi River Delta Front at water depths of ~75 m [11]. Sediment trap data and allied mooring and camera data from deep-water locations (~1000 m) have indicated that frequent, small magnitude resuspension events driven by inertial currents contribute to sediment transport there [30]. Less is known, however, about the mechanisms that drive shelf–slope sediment exchange or transport down the continental slope or canyons.

**Figure 2.** Study area identifying locations of bird-foot delta and Southwest (SW) Pass of Mississippi River; Atchafalaya River and Bay; Mobile Bay; Tarbert Landing (site of commonly used river gauge). Satellite image of Mary 17, 2011 from MODIS on NASA's Aqua satellite.
