Modelling of dispersant application to oil spills in shallow coastal waters

Application of dispersants in shallow water remains an issue of debate within the spill response community. An experimental oil spill to evaluate potential environmental impacts and benefits of applying dispersants to spills in shallow water has therefore been under consideration. One site being considered was Matagorda Bay, on the Texas coast. Coupled three-dimensional oil spill and hydrodynamic models were used to assist in the design of such an experiment. The purpose of the modeling work was to map hydrocarbon concentration contours in the water column and on the seafloor as a function of time following dispersant application. These results could assist in determining the potential environmental impact of the experiment, as well as guiding the water column sampling activities during the experiment itself. Eight potential experimental oil spill scenarios, each of 10 bbl in volume, were evaluated: 4 release points, each under two alternate wind conditions. All scenarios included application of chemical dispersants to the slick shortly after release. Slick lifetimes were under 5 h. Due to the shallow depths, some fraction (2-7%) of the released hydrocarbons became associated with bottom sediments. The algorithms used for the oil droplet-sediment interactions are theoretical, and have not been verified or tested against experimental data, so the mass balances computed here must be considered tentative. Currents computed by the hydrodynamic model are consistent with previous observations: the circulation is largely tidally driven, especially near the ship channel entrance. In the center of the bay, the circulation appears relatively weak. The use of water column drifters with surface markers during the experiment would augment model results in assisting activities to monitor concentrations. These simulations suggest that the eventual behavior of an oil droplet cloud in the middle of the bay will be relatively insensitive to release point or time in the tidal cycle. A limited analysis was run to evaluate model sensitivity to the oil-sediment sorption coefficient. Increasing this coefficient by a factor of 10 results in an approximately linear increase in the fraction of oil in the sediments. Sensitivity of estimated time-to-zero-volume for the 0.1-ppm concentration contour demonstrated that the model prediction of 3.5 days was associated with an uncertainty of +/-12 h for a release of 10 barrels. This time estimate is also a function of the oil-sediment interaction rate, since more oil in the sediments means less oil in the water column. (C) 2003 Elsevier Ltd. All rights reserved.