CLASSIFICATION OF WINDSTORMS OVER FRANCE

A method of semi-automated classification of windstorms over France is proposed, in order to provide insurance companies with meteorological statistical values, that could lead to an evaluation of storm risks. Because not all storm events have the same origins or the same feature s, and do not always hit the same regions, a classification is necessa ry in order to assemble storm events with meteorological and geographi cal similarities. The classification is based on the expert knowledge of a weather forecaster, who first analyses 44 storm events and draws out the meteorological features of different types of storm. Afterward s, the results of the human typology are automated by determining deci sive rules with criteria, which are calculated with only observed mete orological surface data over France. The automation of the human typol ogy is obtained by taking into account the main results of statistical methods such as objective classification or discrimination and by val idating them with physical considerations. We thus succeed in formulat ing the reasoning of the expert. The classification is then applied to the Meteo-France records since 1962. We succeed in selecting 524 stor m events and in classing them into seven main types, with enough event s in each class to calculate statistical values. (C) 1998 Royal Meteor ological Society.