Time-Varying Bivariate Modeling for Predicting Hydrometeorological Trends in Jakarta Using Rainfall and Air Temperature Data

Changes in rainfall patterns and irregular air temperature have become essential issues in analyzing hydrometeorological trends in Jakarta. This study aims to select the best copula of the stationary and non-stationary copula models and visualize and explore the relationship between rainfall and air temperature to predict hydrometeorological trends. The methods used include combining univariate Lognormal and Generalized Extreme Value (GEV) distributions with Clayton, Gumbel, and Frank copulas, as well as parameter estimation using the fminsearch algorithm, Markov Chain Monte Carlo (MCMC) simulation, and a combination of both. The results show that the best model is the non-stationary Clayton copula estimated using MCMC simulation, which has the lowest Akaike Information Criterion (AIC) value. This model effectively captures extreme dependence in the lower tail of the distribution, indicating a potential increase in extreme low events such as cold droughts. Visualization of the best model through contour plots shows a shifting center of the distribution over time. This study contributes to developing dynamic hydrometeorological models for adaptation planning of changing hydrometeorological trends in Indonesia.