Modeling weathering processes of spilled oil on the sea surface based on dynamic Bayesian network

Abstract

Oil spills pose a serious threat to the ecological environment and human health. When the spilled oil enters seawater, oil weathering processes like evaporation, emulsification, biodegradation, photooxidation and dissolution will occur. These processes have high uncertainty and directly affect the spatial distribution and fate of oil spills. This paper presents a stochastic modeling method for weathering process of spilled oil on the sea surface based on Dynamic Bayesian Network (DBN). The physical equations of evaporation, emulsification, biodegradation, photooxidation and dissolution are converted into DBN, and then all the individual DBNs are integrated to obtain the entire DBN of the oil weathering process. The developed DBN could intuitively and graphically express the logical relations and conditional probabilities among physical variables. This method is helpful to find out the influencing mechanism of multi-random variables on the oil weathering process. The proposed method is applied to one type of light oil that is spilled on the sea surface of the Bohai Sea. Based on the developed DBN model, the oil weathering process in the first 18 h are studied by quantitative analysis. Through sensitivity analysis, it shows that the photooxidation and dissolution have greater impact on the heavy oil fraction than biodegradation. For oil-water mixture, it is greatly affected by evaporation in the first few hours, but emulsification becomes the leading weathering process in the 18th hour. The effects of dissolution are low, but higher than biodegradation and photooxidation. The key parameters affecting each oil weathering process are determined. Uncertainty analysis is carried out to reveal the possible upper and lower limit values of some physical quantities caused by the weathering process.