FUZZY LOGIC APPROACH TO FOREST FIRE RISK ASSESSMENT IN TANJUNG PUTTING NATIONAL PARK

Abstract

In implementing fuzzy logic, the Sugeno fuzzy method faces several challenges, such as issues in determining the fuzzy rule base and the occurrence of undefined outputs (defuzzification) with values of 0/0. This study examines the application of the Sugeno fuzzy method in identifying the level of forest fire risk by considering various variables. The variables are temperature, humidity, and wind speed. The model is developed using fuzzy rules constructed based on the relationships among the variables. The test results show that after modifying the membership function boundaries to decimal values approaching the original lower bounds, the Zero-Order Sugeno fuzzy method can produce an average forest fire risk level of 68.83 (high category) in Tanjung Puting National Park. In addition, applying the First-Order Sugeno fuzzy method produces a multiple linear regression model that can be applied within the rule base, resulting in an average forest fire risk level of 68.89 (high category) at the same location. During the evaluation phase, the First-Order Sugeno model achieved a lower RMSE value (15.47) than the Zero-Order model (16.03), indicating that it is more suitable for handling extreme conditions such as dangerous spikes in risk. Therefore, this approach has the potential to serve as an effective early warning system for forest fire mitigation, supporting decision-making processes.