Some recommendations for testing oil spill computer models

摘要

According to a recnet state-of-the-art review of modelling transport and fate of spills [1], more han 50 oil spill models have been developed in the last 30 years. Even though some of these models are used for spill response actions during accidents, environmetnal impact assessment, contingency planning and response training, there are not standard methodologies to evalaute their quality. Questions like "how good is a model at predicting real oil spill events?" are not generally answered objectively. This article presents some ideas that may contribute to design a set of standard benchmarks that would allow users and developers to assess models on a rational basis.