Oil spill forecast assessment using Fractions Skill Score

In the event of an oil spill, emergency responders must quickly deploy cleanup and protection equipment using guidance provided by a forecast trajectory. Forecasting the location of the surface oil over time is standard practice; however, current performance metrics used for assessing the quality of the spill forecast lack both an appropriate numerical model accuracy score and specification of the expected spatial resolution limit for useful forecast information. This paper adapts the Fractions Skill Score method, commonly used in weather forecasting, to oil forecasting. A subset of satellite images and trajectory forecasts from the Deepwater Horizon oil spill are used as an example of the method.