Analisis Pembentukan Sebaran Bivariat Berbasis Copula Antara Luas Area Terbakar dan Curah Hujan di Sumatra Bagian Selatan

Forest and land fires in Indonesia have a close relationship with the surrounding climatic conditions, such as rainfall. One model that can be used to analyze the relationship between the two variables is the copula. However, apart from rainfall, global climate phenomena such as the El Nino-Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD) also have an influence on forest and land fires. Therefore, this study analyzes and models the copula-based co-distribution between rainfall and burned area by partitioning the data based on ENSO and IOD phenomena. The method used is copula-based joint distribution analysis which is estimated using the Inference of Function for Margins (IFM) method. Several copula functions are used to form joint distributions, such as Gaussian, student’s t, Clayton, Gumbel, Frank, Joe, Galambos, BB1 (Clayton-Gumbel), BB6 (Joe-Clayton), BB7 (Joe-Gumbel), and BB8 (Joe-Frank). The results showed that the highest correlation to the burned area occurred for the two months cumulative rainfall data based on the Kendall-Tau correlation. Each ENSO and IOD condition has different characteristics, indicated by the differences in the selected univariate distribution and copula function. Probabilities of burning areas are higher when rainfall is low. In addition, the higher the ENSO and IOD indices, the higher the probability of burned area, during low rainfall. Based on the conditional probabilities, the Positive IOD condition has relatively more significant influence than the Moderate-Strong El Nino. Apart from the Moderate-Strong El Nino and Positive IOD, another condition that has a conditional probability for a relatively high is Weak El Nino conditions. Other conditions, such as La Nina, normal ENSO, negative IOD, and Neutral IOD, have a conditional probability of a very small burn area.