Object-oriented processing of CRM precipitation forecasts by stochastic filtering

International audience; In order to cope with small-scale unpredictable details of mesoscale structuresin cloud-resolving models, it is suggested in this paper to process the modeloutputs following a fuzzy object-oriented approach to extract and trackprecipitating features (associated with a higher predictability than the directmodel outputs). The present approach uses the particle filter method torecognize patterns based on predefined texture or spatial variability of themodel output. This provides an ensemble of precipitating objects, which arethen propagated in time using a stochastic advection-diffusion process. Thismethod is applied to both deterministic and ensemble forecasts provided bythe AROME-France convective-scale model. Specific case studies support theability of the approach to handle precipitation of different types.