Modelling maximum precipitation in a mountainous area of Greece under global warming

We investigated the fit of generalized extreme value (GEV) distributions to maximum precipitation over the Mesochora mountainous catchment in central-western Greece under present and future climate scenarios. Precipitation was modelled as a stochastic process coupled with atmospheric circulation models. Automated objective classification based on optimized fuzzy rules was used to classify observed circulation patterns (CPs) and ECHAM4 General Circulation Model-generated CPs for IxCO(2) and 2xCO(2) climate scenarios. The GEV distribution was fitted by maximum likelihood, allowing for non-stationarity over time in its location and scale parameters. The stationary model was adequate for historical data on annual daily maxima for 1972-1992 and also for 1xCO(2) for the period 1961-2000. However, the 2xCO(2) series for 2061-2100 required a cubic time trend in location to obtain a satisfactory fit (P<0.0001 by likelihood ratio test). This series declined to a minimum around 2080, followed by an increase to a maximum around 2092, and subsequently a further decline.