A nudging scheme for the assimilation of precipitation data into a mesoscale model

A nudging procedure for the assimilation of rainfall data into a mesoscale model (BOLAM) has been developed in order to improve short-range forecasting.The scheme modifies the model specific humidity profiles at every time step, according to the difference between observed and forecast precipitation. Different relaxation procedures are applied depending on the precipitation type (large-scale or convective rain), as estimated by the model itself. Optimizations of the nudging parameters and assessment of the scheme's performance was carried out in an idealized framework (OSSE-Observing System Simulation Experiment type strategy), implementing a Lagged Forecast scheme. Two events were selected for this purpose, both characterized by heavy precipitation in the Mediterranean basin. The first was a severe orographic rainfall event, associated with the passage of a frontal system over the Alps during the Mesoscale Alpine Programme (MAP) field-phase, in September 1999. The second was characterized by the development of a very deep low close to the Algerian coast where heavy precipitation caused a disastrous flood in the city of Algiers in November 2001. The effects of rainfall assimilation were evaluated, both qualitatively and quantitatively, in terms of precipitation forecasts and modification of dynamical fields, with particular attention to the impact on cyclone development. Finally, sensitivity tests were performed in order to assess the dependence of the nudging procedure on rainfall data characteristics (length of the accumulation period and associated error) and model error.