Processing disdrometer raindrop spectra time series from various climatological regions using estimation and autoregressive methods

Abstract

A large data set of rain drop size distribution (RSD) measurements collected with Joss-Waldvogel (JWD) and 2D video disdrometers (2DVD) in UK, Athens, Japan and USA are analyzed. The objective of this work are manifold: i) show the differences of a wide climatological DSD-derived moments; ii) retrieve from this disdrometer data set the driving parameters of the normalized gamma RSD and perform a sensitivity analysis of these results by using different best-fitting techniques; iii) exploit the correlation structure of the estimated RSD parameters as input of a vector autoregressive stationary model in order to simulate time series (or horizontal profiles) of RSDs and, consequently, of either rain rate or path attenuation; iv) characterize the distribution of the inter-rain duration (or dry periods: DP) and rain duration (or wet periods: WP) to design a simple semi-Markov chain to represent the intermittency feature of rainfall process. The overall stochastic procedure to randomly synthetize (or generate) RSD time series is named Vector Autoregressive Raindrop Markov Synthesizer (VARMS) model. This stochastic RSD generation tool may find useful applications both in hydro-meteorology and radio-propagation.