The New French Operational Radar Rainfall Product. Part II: Validation

Abstract A new operational radar-based rainfall product has been developed at Météo-France and is currently being deployed within the French operational network. The new quantitative precipitation estimation (QPE) product is based entirely on radar data and includes a series of modules aimed at correcting for ground clutter, partial beam blocking, and vertical profile of reflectivity (VPR) effects, as well as the nonsimultaneity of radar measurements. The surface rainfall estimation is computed as a weighted mean of the corrected tilts. In addition to the final QPE, a map of quality indexes is systematically generated. This paper is devoted to the validation of the new radar QPE. The VPR identification module has been specifically validated by analyzing 489 precipitation events observed over 1 yr by a representative eight-radar subset of the network. The conceptual model of VPR used in the QPE processing chain is shown to be relevant. A climatology of the three shape parameters of the conceptual VPR (brightband peak, brightband thickness, and upper-level decreasing rate) is established and the radar-derived freezing-level heights are shown to be in good agreement with radiosonde data. A total of 27 precipitation events observed by three S-band radars of the network during the winter of 2005 and the autumns of 2002 and 2003 are used to compare the new radar QPE to the old one. Results are stratified according to the distance to the radar and according to the height of the freezing level. The Nash criterion is increased from 0.23 to 0.62 at close range (below 50 km) and from 0.35 to 0.42 at long range (between 100 and 150 km). The relevance of the proposed quality indexes is assessed by examining their statistical relationship with long-term radar–rain gauge statistics. Mosaics of QPE and quality indexes are also illustrated.