disperse. Although spreading and horizontal dispersion start to work immediately after a spill occurs, spreading is nearly complete within a day while dispersion continues to increase. For most offshore spills, dispersion will move more oil around than will spreading.
Dispersion originates from ocean eddies of various scales, Langmuir circulation, boundary-layer shear (e.g., wind gusts blowing on the sea surface), and other seemingly random turbulence. Dispersion is typically modeled using a Fickian law that assumes a neutrally buoyant, noncohesive substance. Clearly oil is different, so at the very least the dispersion coefficients used in a Fickian model will likely be different from those determined for miscible substances. Some composite oil slick models simply ignore horizontal dispersion and focus on the “center of mass” of the slicks. The National Oceanic and Atmospheric Administration’s GNOME model uses a Fickian law. Others have developed heuristic methods with coefficients tuned to observed slick data. Examples include Morales et al. (1997) who have developed a random-walk method and Howlett et al. (1993) who break the spill into parcels called “spillets” and disperse them numerically.
Vertical dispersion and entrainment are the movements of oil droplets of sizes less than about 100 μm into the water column. Typically droplets that display a residence time of minutes to hours have droplet sizes less than about 20 μm (Reed, 1992). Larger droplets will rise quickly to the surface. MacKay developed an early model of entrainment based on the square of wind speed, the viscosity of oil, slick thickness, and surface tension (Reed, 1992; ASCE, 1996). Tests of this model showed that it provided reasonable results at moderate wind speeds, but otherwise deviated from experimental values.
Delvigne et al. (1987) and Delvigne (1993) developed a series of models based on a number of different flume tests, tank tests, and at-sea measurements. These commonly used models are empirical and are based on breaking wave energy, film thickness, oil type, and temperature. Energy is included as turbulent energy dissipation by the waves per unit area. Later models were developed to account for energy applied by other than breaking waves and included movement around obstacles and hydraulic jumps. The models have been applied successfully under a variety of circumstances. They do not, however, account for the stability of the droplets in the water column, a factor that largely depends on droplet size and has been modeled based on empirical data (Delvigne et al., 1987). Few tests of models have been done at sea because of the analytical difficulties of measuring the many factors involved. The tests conducted thus far have been mainly an extrapolation of the fate of oil to the Delvigne model (ASCE, 1996). The depth of mixing was found to conform largely to the rule of thumb that states that the depth of mixing is 1.5 times the wave height (Delvigne et al., 1987).
Sinking is the mechanism by which oil masses that are denser than the receiving water are transported to the bottom. The oil itself may be denser than water, or it may have incorporated enough sediment to become denser than water. Sedimentation is the sorption of oil to suspended sediments that eventually settle out of the water column and accumulate on the seafloor. There is a significant difference in the relative amount of oil incorporated by the two processes; sinking oil may contain a few percent sediment, whereas contaminated sediments accumulating on the seafloor will contain at most a few percent oil (McCourt and Shier, 2001). Sedimentation requires a mechanism for oil to become attached to sediments. One mechanism is ingestion of small oil droplets dispersed in the water column by zooplankton and excretion of oil in fecal pellets that then sink to the seafloor. This process has been documented only during the Arrow spill in Chedabucto Bay (Conover, 1971).
The National Research Council (NRC, 1999) developed conceptual behavior models for nonfloating oils that described the factors determining whether spilled oil will sink. Figure 4-5 shows the interaction of these factors. Because most nonfloating oils are only slightly denser than water, the presence of currents can keep the oil in suspension and prevent its accumulation as a coherent mass on the bottom. For example, little or no oil accumulation on the bottom was reported after heavy-oil spills in the Mississippi River (Weems et al., 1997) and Puget Sound (Yaroch and Reiter, 1989). In very few spills of oil that was heavier than water, the oil sank directly to the bottom, and these kinds of spills occurred only in sheltered settings (e.g., from the vessels Sansinena and Mobiloil). In contrast, a buoyant oil can pick up enough sediment, either after stranding onshore or mixing with sediment suspended by wave action, to become an oil-sediment mixture that is denser than sea water. If the sediment separates from the oil mass, the still-buoyant oil can then re-float, as was observed at the Morris J. Berman spill in 1991 in Puerto Rico.
Recent studies on sedimentation of spilled oil have focused on the interaction of fine particles (clay) and oil stranded on the shoreline as a mechanism that speeds natural removal of residual stranded oil (Bragg and Owens, 1995). This process involves oil-fine interaction of micron-sized mineral fines with oil droplets in the presence of water containing ions. Once processed, the oil droplets do not coalesce, and the oil is readily removed from the shoreline by tidal flushing and wave action. The oil-clay oil particle clusters are of near-neutral buoyancy and are easily kept in suspension. The oil attached to the fine particles is more available for biodegradation. This process was first described during the Exxon Valdez oil spill and has been proposed as occurring at other recent spills (Bragg and Owens, 1995). No field or laboratory measurement techniques, however, have been developed that enable immediate identification of