Copulas analysis of victorian precipitation data: a nonparametric approach

This study used data from rain gauges at six (6) Victoria stations with some of the highest annual mean rainfalls. Empirical copulas are used to model the bivariate rainfall severity and duration distribution. A nonparametric density function is used to estimate the Standard Precipitation Index (SPI) prior to fitting copulas. The marginal distributions of each of the rainfall characteristics in the copula function's argument are then fitted using empirical distributions. Based on the values of Mean Absolute Error, the empirical copula can be used to represent the results obtained with standard copulas.