The states of the variable are described as intervals, which are quite large but can be easily modified if necessary. There is one specific amount for the oil spills, of 30 000 ton, which is the largest oil spill considered by the authorities in Finland, and reflects the preparedness level for Finland, see SYKE (2011). It is an independent variable, which exists in three states: spring (Mar.–May), summer (Jun.–Aug.) and autumn (Sept.–Nov.). Winter

is excluded for several reasons, first as oil-spill combating during ice season is different than during the other seasons. Second, some of the oil-combating vessels are not capable of operating in ice conditions. Third, there is no reliable MEK inhibitor prediction model for the movement of oil in ice conditions in the GOF, (Helle et al., 2011). The prior distribution for the variable Season is presented in Table 2 and informs about the probability that an accident resulting in an oil spill would occur on the Gulf of Finland specifically during this season of the year. The distribution was gained from the compiled accident statistics of HELCOM between the years 1989 and 2005 – ( HELCOM, 2013). It is one of the most important factors affecting the cost of the clean-up operation. It affects the cost in a multitude of ways, starting from the C59 wnt chemical structure way that the spilled

oil spreads in water, which affects the time it takes for the spill to reach the shoreline. In addition, heavier oil has the tendency to sink; this in turn affects the possible recovery Adenosine percentage of the oil-combating vessels.

The oil type also affects the efficiencies of the combating vessels, due to the fact that some oils are less likely to adhere to the brushes used by the combating vessels. In the presented model, this variable exists in three states: light, medium and heavy. The probabilities for each state are given in Table 3. They are based on an estimation made by experts from the Finnish Environment Institute considering the oil tankers traffic in the Gulf of Finland, see for example Juntunen et al. (2005). For the Gulf of Finland, it is estimated that an oil slick would arrive ashore quite quickly. In the case of an accident taking place in the middle of the sea, it could take between one to nine days for the oil to reach the shoreline, see for example Andrejev et al., 2011, Viikmäe and Soomere, 2013 and Soomere et al., 2011. Therefore the variable is set to consist altogether of ten intervals, ranging from zero to ten days. We assume, the prior distribution for this variable follows the Gaussian distribution, with μ = 5 days and σ = 2 days. However, if the spill takes place in Finnish waters of the Gulf of Finland, it is estimated that it would take a maximum of three days before the oil reaches the shore, ( Hietala and Lampela, 2007).