Mathematical models for extreme monthly precipitation

Abstract

In this article we present the results of modelling the extreme monthly precipitation series collected in the period January 1965 – December 2005, at ten hydro-meteorological stations situated in Dobrudja, a region from the South - Eastern part of Romania. The main model used was the Generalised Pareto Distribution. For the non-stationary series, determined after performing KPSS test, the existence of a cyclical (harmonic) trend has also been studied and a GPD re - parameterisation by the orthogonal method has been done, using the scale as Poisson parameter, in the case of cyclical trend existence. The comparison between the models, by the likelihood – ratio test, proved that only in two cases (Medgidia and Corugea series) the reparameterized model was better.