Simultaneous Confidence Intervals for the Ratios of Log-normal Means Using Bayesian Two-steps MOVER with EEC Rainfall Application

Thailand frequently faces flooding during the summer monsoon, which is frequently caused by powerful tropical storms that cause extensive damage. This study attempts to address this problem by establishing simultaneous confidence intervals for the ratios of log-normal means, which are crucial for comparing rainfall intensities across geographical areas. Three proposed methods—M-BIND, M-BR, and M-BU—were evaluated against three existing methods—M-M, M-F, and SFGCIs—to assess their effectiveness. Monte Carlo simulations were used to assess the performance of these techniques, with coverage probabilities and expected widths performing as the evaluation criteria. Findings showed that, especially in situations with small to moderate variance and moderate to large sample sizes, the suggested M-BR approach continuously performed better than the others by obtaining the shortest predicted widths while maintaining appropriate coverage probabilities. The results have been confirmed by applying the computational methods to weekly maximum rainfall data from the Eastern Economic Corridor of Thailand.