Análisis de eficiencia de la distribución Bi-Gumbel

Abstract

MOST HYDROLOGICAL EXTREME STUDIES IN THE PAST HAVE BEEN ANALYZED THROUGH USE OF UNIVARIATE DISTRIBUTIONS. THE LARGE VARIABILITY OF THE T-YEAR FLOOD ESTIMATES HAS PROMPTED EXPLORATION OF JOINT ESTIMATION MODELS, SUCH AS THE BIVARIATE DISTRIBUTION WITH EXTREME VALUE TYPE I MARGINALS, NAMED BI-GUMBEL DISTRIBUTION. TO INVESTIGATE WHETHER THE ESTIMATES OF THE QUANTILES BASED ON BIVARIATE DISTRIBUTION ARE BETTER THAN THOSE ON UNIVARIATE PROCEDURES A DISTRIBUTION SAMPLING TECHNIQUE WAS USED. A SIGNIFICANT IMPROVE MENT OC CURS WHEN THE PA RAM E TERS ARE ES TI MATED US ING THE BIVARIATE DIS TRI BU TION IN STEAD OF UNIVARIATE FORM AND SUCH AGAIN IS MORE SIGNIFICANT IN RELATION TO THE SHORTER SAMPLES.