A statistical methodology for the evaluation of long-range dispersion models: An application to the ETEX exercise

This paper describes the statistical methodology applied to evaluate the performance of the long-range dispersion models that were used in the modelling activities of ETEX (European Tracer EXperiment). The availability of a large number of models makes this exercise rather unique. These models are used for the practical purpose to quantify the contamination effects over a vast area, following a hypothetical accidental release of harmful material. This makes the quality judgement that could be attributed to the results of each model particularly important. The statistical indicators considered to be the most effective for the evaluation of long-range dispersion models are introduced and commented, with specific examples in the frame of ETEX simulations. The importance of using several indices and critically interpreting the results is discussed. (C) 1998 Elsevier Science Ltd. All rights reserved.