OIL SPILL DETECTION AND MODELLING: PRELIMINARY RESULTS FOR THE CERCAL ACCIDENT

Abstract

Oil spill research has significantly increased mainly as a result of the severe consequences experienced from industry accidents. Oil spill models are currently able to simulate the processes that determine the fate of oil slicks, playing an important role in disaster prevention, control and mitigation, generating valuable information for decision makers and the population in general. On the other hand, satellite Synthetic Aperture Radar (SAR) imagery has demonstrated significant potential in accidental oil spill detection, when they are accurately differentiated from look-alikes. The combination of both tools can lead to breakthroughs, particularly in the development of Early Warning Systems (EWS). This paper presents a hindcast simulation of the oil slick resulting from the Motor Tanker (MT) Cercal oil spill, listed by the Portuguese Navy as one of the major oil spills in the Portuguese Atlantic Coast. The accident took place nearby Leix˜oes Harbour, North of the Douro River, Porto (Portugal) on the 2nd of October 1994. The oil slick was segmented from available European Remote Sensing (ERS) satellite SAR images, using an algorithm based on a simplified version of the K-means clustering formulation. The image-acquired information, added to the initial conditions and forcings, provided the necessary inputs for the oil spill model. Simulations were made considering the tri-dimensional hydrodynamics in a crossscale domain, from the interior of the Douro River Estuary to the open-ocean on the Iberian Atlantic shelf. Atmospheric forcings (from ECMWF - the European Centre for Medium-Range Weather Forecasts and NOAA - the National Oceanic and Atmospheric Administration), river forcings (from SNIRH - the Portuguese National Information System of the Hydric Resources) and tidal forcings (from LNEC - the National Laboratory for Civil Engineering), including baroclinic gradients (NOAA), were considered. The lack of data for validation purposes only allowed the use of the two-dimensional surface plume transport model VOILS [1] with the oil spreading formulation enabled. The remaining oil weathering processes (evaporation, emulsification, dispersion and dissolution in the water column) and shoreline retention were disregarded. The computational structure of the model is based on Eulerian-Lagrangian formulations, horizontal unstructured mesh discretization and it is soft-coupled with the tri-dimensional hydrodynamic model SELFE - Semi-Implicit Eulerian Lagrangian Finite Element [15] that uses hybrid sigma-Z coordinates in the vertical. The preliminary results of this hindcast simulation for the Cercal oil spill are presented and compared with available satellite SAR images. The forcings used play an important role in the final results. During the late stage spreading phases of the oil, about one month after the spill, the Douro River outflow is best seen in the SAR images. The morphology of the river outflow is discussed according to traditional coastal dynamics, and compared with model results. In addition to several interesting physical features that were identified, we report on the generation of Internal Solitary Waves (ISW) in the vicinity of the Douro River Plume (DRP). It is well known that trains of short-period internal waves can be generated by river plumes (such as the Columbia River). The internal structure of the observed internal waves (elevation waves or mode-2 versus mode-1 internal waves) is discussed based on the SAR signatures and available stratification. The present work has been conducted under an FCT - Fundac¸ ˜ao para a Ciˆencia e a Tecnologia / MCTES - Minist´erio da Ciˆencia, Tecnologia e Ensino Superior (PIDDAC - Programa de Investimentos e Despesas de Desenvolvimento da Administrac¸ ˜ao Central) Portuguese funded project entitled PAC:MAN Pollution accidents in coastal areas: a Risk management system (PTDC/AACAMB/ 113469/2008).