Generating monthly rainfall amount using multivariate skew-t copula

This study aims to generate rainfall data in cases where the data is not available or not enough for a certain area of study. In general, the rainfall data is rightly skewed, so the multivariate skew-t copula is used as it able to model rainfall amount and capture the spatial dependence in the data. To illustrate the methodology, three rainfall stations in Kelantan are used. Firstly, the observed data is transformed to uniform unit. The Spearman's correlation coefficient is calculated between the three stations. It is found that the correlations between the stations are significance at α = 0.05. The next step involved generating the synthetic rainfall data using the multivariate skew-t copula. The data is then transformed to uniform unit and the correlation coefficient is calculated for the generated data. Finally, the correlation coefficient of the observed and the generated data are compared. The Kolmogorov-Smirnov goodness of fit test is used to assess the fit between theoretical and empirical copula and supported by graphical representation. The results show that there is no significant difference between empirical and theoretical copula at 5% significance level. Thus, the multivariate skew-t copula is suitable to generate synthetic rainfall data that can mimic the observed rainfall data. It can also be used to present different rainfall scenarios by changing the value of the parameters in the mode