Statistical distributions and sampling strategies for the analysis of extreme dry spells in Catalonia (NE Spain)

Statistical distributions of annual extreme and long dry spells, using daily rain amount thresholds of 0.1, 1.0 and 5.0 mm/day, are analysed from a database of 39 rain gauges in Catalonia (NE Spain), with a daily recording period extending from 1950 to 2000 and with little missing data. The generalised extreme value (GEV) and generalised Pareto (GP) distributions are considered to model the series of annual extreme (AE) dry spells. The same distributions are also assumed for the partial duration (PD) series, which are derived from dry spells exceeding the 95% percentile. In both cases, the three parameters of the GEV and GP distributions are fitted by means of the L-moments method, which offers a robust estimation of them. Additionally, the fit between empirical data and theoretical distributions is evaluated in terms of the empirical and theoretical skewness and kurtosis. Even though AE spells are commonly well fitted by the GEV model, the GP model is a better option for some rain gauges. Similarly, in the case of the PD series, a few cases are better modelled by the GEV distribution. This study establishes the basis for climatic drought risk assessment in Catalonia, given that dry spell lengths associated with return periods of 2, 5, 10 and 25 years are accurately revised by comparing results deduced from the GEV and GP models, both for the AE spells and PD series.