Climate regimes and yearly streamflow frequency analysis in Sicily

Abstract

Yearly streamflow frequency analysis is performed in Sicily, Mediterranean’s largest island, using the index flow procedure, i.e. fitting a single probability distribution function (pdf) to the available yearly flow data scaled by their mean. This requires 1) the identification of homogeneous regions where L-moments’ observed variability can be ascribed to sample variability only, and 2) the estimation of mean annual streamflow by multiple regression analysis with some significant climatic and morphologic covariates. Rather than having defined geographical boundaries, the identified homogeneous regions should gather basins sharing similar morphologic and climatic features. Steps 1) and 2) are commonly kept separated; however, classic multiple regression analysis has a potential to identify regions with different average streamflow levels through measures of heteroscedasticity. This can be related to the presence of climatic regimes and is useful to direct analysis in the identification of basin typologies that can be characterized by a single pdf. In Sicily, within a general Mediterranean climate, different climatic regimes co-exist ranging from humid to semiarid. Tests for heteroscedasticity on the residuals of a regression of mean yearly flow versus the morphologic and climatic covariates indicate that groupwise heteroscedasticity exists at a regional scale and therefore suggest splitting the sample in two subsamples of semiarid and subhumid basins. What is significant, however, is that such subdivision, based on Thornthwaite’s humidity index, also proves effective from the standpoint of the homogeneity of the pdf of yearly flows. Further analysis is carried out to test the goodness-of-fit of different candidate distributions of the scaled yearly streamflow.