Initialization of a fine-scale model for convective-system prediction: A case study

This paper focuses on the problem of the initialization of a fine-scale (2.5 km resolution) model with observational data. An approach combining optimal interpolation (GI) analysis and bogussing methods was designed. The information provided by surface mesonet observations (relative humidity, temperature and winds) is incorporated by means of a mesoscale (OI) analysis. Simple cloud and precipitation analyses based on conventional radar reflectivities and infrared satellite data help to adjust the humidity and hydrometeor fields of the initial state. With this initialization procedure the resulting initial state describes the signature of mesoscale convective systems. This technique is applied to one observed mesoscale convective system to produce an initial state describing a one-hour-old developing convective system. From this initial state, the mesoscale nonhydrostatic model, named MESO-NH, succeeds in simulating a convective line organization which is close to observations. Experiments to determine the sensitivity to the initial state definition and to the model characteristics are carried out and the initialization procedure is found to be instrumental in a successful simulation; both the surface-data analysis and the cloud analysis are required.