Metodologías de análisis de frecuencia en variables hidrológicas y el fenómeno del niño

Abstract

The dependence of the frequency of occurrence of extreme events of hydrological variables with El Niño and climate variability is analyzed. Daily maximum levels of the Parana river, maximum daily flows of the Uruguay river and maximum daily rainfall at several gages located in Entre Rios were used. The classical approach under assumptions of stationarity was considered. Also the possibility that the maximum values are generated by distributions of two or more populations. This was done applying a distribution of mixed probability to model the probability of exceedance considering the presence of ENSO phenomenon to characterize the population belonging. The phenomenon of non-stationarity on the data series was considered from the assumption that the extreme behavior of the series depends on an exogenous variable. Three cases for the GEV distribution are considered: (i) the position parameter depends linearly on a covariate, (ii) the position parameter is a quadratic function of a covariate and (iii) the position and scale parameters depends on a covariate. The extension of return period concept to nonstationary process is discussed.