Theme-2-2-Comparative study of different parameter estimation techniques for EV-1 distribution.

Abstract

Design of water resources structures needs estimation of hydrologic and meteorologic events for different return periods. This can be obtained through frequency analysis approach. The extreme value type 1(EV-1) distribution is widely used for frequency analysis of extreme events in hydrology and meteorology. Its successful application depends upon the accuracy with which its parameters can be estimated. However, there is no universal agreed method of estimating its parameters.

This paper explains and statistically compares various parameter estimation techniques cited in literature using data of different sample sizes generated by Monte Carlo Simulation. The analysis of results reveals that method of mixed moments and incomplete mean procedures are least accurate methods for estimating the parameters and quantiles of EV-1 distribution. The methods based on probability weighted moments, principle of maximum entropy and maximum likelihood give nearly similar results and are recommended for general use.