PREDIKSI HUJAN BULANAN PADA PERIODE ENSO (El NINO SOUTHERN OSCILLATION) MENGGUNAKAN ANFIS (ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM) DI BANJARMANGU, BANJARNEGARA

High accuracy in rainfall prediction is needed to obtain appropriate and useful information for natural disaster mitigations by the community. To get higher accuracy, we need a set of predictor variables related to the rainfall, and we can use rainfall anomaly patterns affected by The El Niňo Southern Oscillation (ENSO). The total column water (TCW), as the chosen predictor variable, is the potential for water vapor in the atmosphere, which has the possibility to become droplets falling to the earth's surface. TCW data are the re-analyzed Global Circulation Model data obtained from the European Mid-Term Weather Forecast Center (ECMWF). The correlation analysis was carried out to evaluate the relationship between predictors and rainfall. Adaptive Neuro-Fuzzy Inference System (ANFIS) was used to predict the chaotic rainfall. The results showed that the climatological pattern of TCW predictors was in accordance with the climatological rainfall pattern with the correlation strength (r) 0.79. The results showed that the climatological pattern of TCW predictors followed the climatological rainfall pattern with a relation strength of 0.79. The correlation between prediction and observation rainfall is 0.82. The lowest correlation at the time of the Normal, El Niňo and La Niňa patterns in 2016 was 0.69, followed in 2014 when Normal Phase and El Niňo were at 0.77, and when EL Niňo was dominant, it reached the highest correlation at 0.93 in 2015. In the 10day scale, rain predictions show a level of reliability that is not much different from the prediction of monthly rainfall with a value of r (0.65) in the La Niňa period and 4 (0.80) in the El Niňo period.