Drop size distribution is a fundamental property of rainfall for two main reasons: the shape of the distribution reflects the physics of rain formation processes, and it is of basic importance in determining most parameters used in radar-meteorology. Therefore, several authors have proposed in the past different parameterizations for the drop size distribution (DSD). The present work focuses attention on the gamma DSD properties, assumed to be the most suitable for describing the observed DSD and its variability. The data set comprises about 3 years of data (2001-2004) for about 1900 ruin of rain, collected in Ferrara in the Po Valley (Northern Italy) by a Joss and Waldvogel (JW) disdrometer. A new method of moments to determine the three gamma DSD parameters is developed and tested; this method involves the fourth, fifth and sixth moments of the DSD, which are less sensitive to the underestimation of small drops in the JW disdrometer. The method has been validated by comparing the observed rainfall rates with the computed ones from the fitted distribution, using two classical expressions for the hydrometeor terminal velocity. The 1-min observed spectra of sonic representative events that occurred in Fen-am are also presented, showing that with sufficient averaging, the distribution for the Ferrara rainfall can be approximately described by a gamma distribution. The discrimination of convective and stratiform precipitation is also an issue of intense interest. Over the past years, several works have aimed to discriminate between these two precipitation categories, on the basis of different instruments and techniques. The knowledge of the three gamma DSD parameters computed with the new method of moments is exploited to identify some characteristic parameters that give quantitative and useful information oil the precipitation type and intensity. First, a key parameter derived from the knowledge of two gamma DSD parameters (m and Lambda), the peak (or modal) diameter D, defined as m/Lambda, is identified. A theoretical relationship between the m and Lambda parameters is successively derived, conducing to a new convective/stratiform discrimination algorithm: in an m-A plot the line (1.635 Lambda-m)= 1 can be considered as the discriminator; the stratiform events fall in the upper part, the convective ones in the lower. A classical tropical oceanic convective/stratiform discrimination algorithm is also tested, showing that it is not suitable to correctly discriminate the mid-latitude precipitations analyzed here. (c) 2005 Elsevier B.V. All rights reserved.
