Abstract Application of the generalised extreme value (GEV) distribution in extreme event modelling requires the estimation of its three parameters (location, scale and shape). Although different techniques for estimating the parameters are available, the comparison of the available parameters’ estimation techniques is yet to be investigated rigorously for the determination of the probability of occurrence of extreme rainfall with higher degree of accuracy. This research presents the assessment of four parameters estimation techniques, maximum likelihood estimation (MLE), generalised maximum likelihood estimation (GMLE), Bayesian and L-moments. A key contribution of this research is the systematic evaluation of block length (in extreme rainfall extraction) sensitivity in GEV distribution parameters estimation and associated return levels which are evaluated consistently across several temporal time windows. The evaluation is accomplished within a regional-scale applying long-term rainfall recordings from 29 meteorological stations in Tasmania, Australia. The outcome of the study suggests that the method of L-moment method is more appropriate for fitting the GEV distribution to the time-series data with outliers. The selection of block length in extreme rainfall extraction has considerable influence on the magnitude of the GEV distribution parameters.
Hossain et al. (Fri,) studied this question.