On selection of a parent probability distribution for describing the distribution of daily annual maximum rainfall series in Vietnam

Abstract

     Information on the probability of extreme rainfall occurrence and amount is of critical importance for the design and management of various hydraulic structures. This information is commonly obtained using the rainfall frequency analysis (RFA) method - a statistical approach where an appropriate statistical model is selected based on the best fit to the observed data. The common issue is that many statistical models are available, however, there is no general agreement as to which model should be used. Therefore, in practice, a number of popular models are selected and compared for their goodness-of-fit degrees. This paper assesses the performance of 07 statistical models in fitting the daily annual maximum rainfall series (AMS) from a network of 155 raingauges in Vietnam. Results based on various numerical and graphical goodness-of-fit criteria have indicated that no distribution can serve as the best distribution for all stations. However, overall, the Generalized Normal (GNO) were the best models and could be recommended as the most suitable model for describing the distribution of daily AMS in Vietnam. The Pearson Type III (PE3) and the Generalized Extreme Values (GEV) distributions provided almost identical results to the GNO distribution and can also be used for RFA.