Abstract: Analysis on low flow features is of practical significance in the management of lake water resources in dry season. According to the theory of runs which is widely used in drought analysis, low flow duration, averaged low flow intensity and maximum low flow intensity were determined as the characteristic variables from the data of averaged daily lake level at Duchang hydrologic station from 1958 to 2007. Weibull, Logistic, and Generalized Extreme Value distributions were used to simulate low flow duration, averaged low flow intensity, and maximum low flow intensity, respectively. Several asymmetric Archimedean copulas were applied to construct three-dimensional joint distributions and the Clayton copula was selected as the best one. The joint probabilities and return periods of low flow variables were calculated and analyzed. It was concluded that the three-dimensional joint distribution could reveal the probability characteristics of low flow more comprehensively in the Poyang Lake.

Key words: Copula function, Poyang Lake, Duchang station, low flow, theory of runs, multivariate joint distribution