Assessing oil spill risks in the northern Baltic Sea with Bayesian network applications
Year: 2015
Type: dissertation
Abstract: Coastal and marine ecosystems across the globe are heavily impacted by various anthropogenic stressors, which has led to a significant loss of biodiversity and ecosystem services in recent decades. In order to find means to counteract this trend, there is a need to develop methods for assessing the environmental impacts of human activities and the effectiveness of management practices to mitigate the harmful effects. However, this is a challenging task due to the complex interactions within and between the ecosystems and human components, and various uncertainties related to them. Bayesian networks (BNs) are graphical models for reasoning under uncertainty. A BN consists of a set of probabilistic variables connected with links describing causalities within the system. As the states of the variables are described with probability distributions, uncertainty can be described in an explicit manner. BNs also enable integration of qualitative and quantitative knowledge from various sources such as observational data sets, models and expert knowledge. In this thesis I have developed BN models to study environmental risks related to anthropogenic stressors in the Gulf of Finland and the Finnish Archipelago Sea. The main aim is to quantify human impacts on the environment, and to assess the ability of different management measures to lessen these impacts. I focus especially on oil spills resulting from potential tanker accidents and I set out to fill various information gaps related to this recently emerged threat. The thesis includes five papers. In paper I, the main aim is to assess the spatial risk posed by oil spills in the Gulf of Finland and the Finnish Archipelago Sea, and identify species and habitat types with the highest risk. In paper II, I focus on the effectiveness of different oil combating methods to mitigate the negative impacts of oil spills on the ecosystem, and paper III widens the approach to a probabilistic cost-benefit analysis of preventive and post-spill measures. Paper IV deals with multiple risks as, in addition to oil spills, eutrophication and harvesting of species are studied. Paper V reviews and discusses various methods that can be applied to evaluate the uncertainty related to deterministic models, which could increase their usefulness in decision-making. The results suggest that risks related to tanker accidents are distributed unevenly between areas, habitats and species. Furthermore, the results support the current Finnish strategy to base oil combating primarily on offshore recovery vessels instead of chemical dispersants. However, as the efficiency of mechanical recovery is dependent on several factors, there is also a need to develop preventive measures. Although major oil accidents are estimated to be rare events, the costs can be very high, if a spill occurs. The work offers new insights to the oil spill risks in the study area and provides examples how Bayesian networks can be applied in environmental risk assessment. The thesis is a part of the work needed in order to develop comprehensive decision support tools related to environmental risk management in the northern Baltic Sea.
Author: Inari Helle
Institution: University of Helsinki
Language: English
Subfield: Pollution
Field: Environmental Science
Domain: Physical Sciences
Keywords: oil spill risk assessment, tanker accidents, Gulf of Finland, Bayesian networks, environmental risk assessment, oil spill response, mechanical oil recovery, chemical dispersants, cost-benefit analysis, eutrophication, multiple stressors, environmental management decision-making, marine ecosystem impacts, coastal ecosystems, anthropogenic pressure, expert knowledge
Open Access status: green
OpenAlex ID: W1484782661
