Using a classical weight-of-evidence approach for 4-years' monitoring of the impact of an accidental oil spill on sediment quality.