• Injection of data;
  • QPF


This paper examines the benefit on quantitative precipitation forecasts of assimilating a mesoscale network of surface observations. The analysis is performed with an optimal interpolation system tuned for meso-β scale (10 km resolution). The case-study is chosen in the framework of the Mesoscale Alpine Program (MAP). It corresponds to a frontal system crossing France on 2 November 1999 followed by moderate to strong convective activity which developed over the Alps and Italy on 3 November (MAP Intensive Observation Period (IOP) 14).

Model sensitivity to the various initial conditions (European Centre for Medium-Range Weather Forecasts analysis, mesoscale analyses with dense surface observations) and to the resolution (meso-β versus meso-γ (2.5 km scale)) is investigated. A comprehensive network of rain-gauges is used for validation. Conventional objective scores (correlation and equitable threat) are produced for precipitation accumulated over short durations (6 and 1 h) for consistency with the target horizontal scales (meso-β, meso-γ). The positive contribution of the surface network to precipitation forecasts is confirmed by a novel and powerful validation method based on the comparison of METEOSAT synthetic radiances (infrared and water vapour channels) computed from the simulations to the corresponding satellite observations. Copyright © 2005 Royal Meteorological Society