ABSTRACT
Many observed annual flood series exhibit reverse curvatures when plotted on lognormal probability paper. The occurrence of these curvatures may be attributed to seasonal variation in flood-producing storm types, relative dominance of within-the-channel or floodplain flow, and variability in antecedent soil moisture and cover conditions. A mixed distribution model is needed to analyze such flood series because of the mixed population of floods. A versatile flood frequency methodology has been developed which considers the probability distribution of an observed flood series as a mixture of the probabilities of two lognormal distributions. Objective detection and modification of any outliers and inliers in an observed flood series in as integral part of the versatile flood frequency methodology. Analyses of about 600 flood series in the USA and other countries in the world show the versatility and superiority of the proposed methodology for satisfactory estimation of design floods for various recurrence intervals.