ASSESSMENT OF CLIMATIC EVOLUTION THROUGH STOCHASTIC ANALYSIS OF RAINFALL TIME SERIES

In the last years a great attention has been devoted to the analysis of long hydrologic time series, looking for assessment on climatic change. Particularly the fluctuations of the internal structure of rainfall process induced by climatic change have been analysed through stochastic analysis of non-stationary properties of time series. Moreover several non-parametric statistical analyses have shown to be able to determine abrupt change in single time series. The present work is mainly focused on the rainfall characterization at different temporal aggregations observed in 25 rain gauges of the Crati river basin, Calabria (Italy) for the time period 1920-2000. Firstly, both at annual and monthly time scale, non-parametric statistical techniques have been adopted to identify significant shifts and trends in precipitations series. Particularly the Kendall’s rank correlation test and some procedures of the Mann-Whitney test have been used to determine direct and residual trend in annual and monthly rainfall series. Successively stochastic analysis of daily rainfall data has been performed through the simple Poisson rectangular pulse model. The dynamic characteristics of daily precipitation patterns have been pointed out by means of the spatial distribution of model parameters and their temporal evolution, thus showing the combined effect of the distance from the coast and the orographic shields from the prevalent direction of the storms.