An implementation of fuzzy inference system for onset prediction based on Southern Oscillation Index for increasing the resilience of rice production against climate variability

Rice production system in Indonesia is very sensitive to global phenomena, especially the El Nino phenomenon. Information regarding the onset of rainy season is important to increase the resilience of the production system. This paper is focused on the implementation of Fuzzy Inference System (FIS)...

Full description

Saved in:
Bibliographic Details
Main Authors: Buono, A., Mushthofa
Format: Conference Proceeding
Language:English
Subjects:
El Nino
Fuzzy Inference System
Fuzzy logic
Indexes
Mathematical model
Meteorology
Onset
Oscillators
Production systems
Rice production system
Southern Oscillation Index
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Rice production system in Indonesia is very sensitive to global phenomena, especially the El Nino phenomenon. Information regarding the onset of rainy season is important to increase the resilience of the production system. This paper is focused on the implementation of Fuzzy Inference System (FIS) as a technique for predicting the onset of rainy season based on the Southern Oscillation Index (SOI) data in the months of July, August, September and October. There are two data sets used: the SOI data from the year 1877 to 2011 and the rainy season onset in the District of Indramayu. Fuzzy set memberships and the set of rules are designed by investigating the two sets of data (via visualization and clustering). The prediction system is verified by using the actual data from the district. The result of the verification shows that the correlation between the rainy onset and its predicted value is 0.68. Even though the prediction accuracy is relatively competitive compared to the existing methods, the requirement of the use of the SOI variables from the month of October renders the model less useful in practice, since it would be mostly too late to wait until October to perform the prediction. Further research can be developed which integrates Markov Chain method to overcome this problem.