Hydrological frequency analysis based on robust statistical theory

In hydrological frequency analysis, the basic assumption that hydrological extremes are identically distributed from a given probability distribution with unknown parameters, is somewhat challenged. It is more likely to recognize that data collected from observation are from a dasiacontaminatedpsila distribution. In this paper, a new robust statistical method was designed to provide quantile estimates by data from the dasiacontaminatedpsila probability distribution. The method is constructed by combining Huber´s M-Estimator (M-E) and Minimum Distance Estimator (MD-E). It has been shown that M-E is capable of obtaining a robust estimation of the location parameter of a distribution, and MD-E is able to achieve the robust estimations of its scale and shape parameters. Statistical properties of the proposed method have been studied based on Monte-Carlo experiment to the Person type three distribution (P-3). Results show that the method presented is robust and has advantages in preserving good statistical properties.