Length of Rainy Season Prediction Based on Southern Oscillation Index and Dipole Mode Index Using Support Vector Regression

Abstract

The location of Indonesia which is between the Pacific and Indian Ocean causes the climate to be influenced by the global condition in bath oceans. The goal of this research is to use modelling to predict the length of the rainy season. To do this we use support vector regression (SVR), and the southern oscillation index (SOl) from the Pacific Ocean and the dipole mode index from the Indian Ocean as the predictor variables. The predictive value of this model is evaluated with determination coefficient (R2) and root mean square error (RMSE). The data used in this research is the length of rai ny season data from three weather stations in Pacitan district (Arjosari, Kebon Agung, Pringkuku) between 198211983 and 2011/2012 periods as the observation data. SOl and DMI between 1982 and 2011 used in this research as the predictor data. The rcsult of this research is a prediction model for each climate station. The best R 2 for Arjasari, Kebon Agung, and Pringkuku weather stations are 0.73,0.63, and 0.58 respcctivcly. Meanwhilc, the best RMSE for Arjasari, Kebon Agung, and Pringkuku weather stations are 2.45, 3.23, and 2.86 respectively.