Bayesian inference modeling to rank response technologies in arctic marine oil spills

Marine oil spills have a detrimental effect on aquatic systems. Yet, it is challenging to select appropriate tech-nologies in the Arctic because of limited logistics support, inclement weather conditions, and remoteness, and limited research has been conducted in this direction. This article suggests a method to rank the oil response technologies, including mechanical recovery, chemical dispersant, and in-situ burning, for use in Arctic oil spill risk assessment and preparedness planning. The proposed Preference Learning based Bayesian Inference Modeling offers data-driven ranking of systems by learning a label function and considers factors such as ice covered sea areas, cold weather, and spill volume. A data generation system is developed to produce numerous oil spill scenarios, using a state-of-the-art engineering tool. Results demonstrate that the model, while simple, can effi-ciently and accurately select the best available technique, making it suitable primarily for marine pollution preparedness and response planning in strategic risk assessments.