System for determining lahar disaster status using machine learning method

Abstract

Lahar disaster is an event of material transport such as sand, gravel, and rocks following volcanic eruption that is triggered by intense rainfall. The disaster on the slope volcano induces a potential loss that include casualty, damage or loss of property, and environmental disruption. Therefore, a system of lahar flood warning system is needed to help determining the status of flood disasters on the volcano slope. In this study the system of lahar vulnerability estimation is developed. The target area is a river on Merapi volcano Indonesia. Naïve Bayes Classifier Method is applied to classify areas categorized as flood-prone zones or safe zones. The determining factors are spatially distributed rainfall intensity from X-band weather radar, topographical factor, and soil type. This research has produced a flood disaster status determination system on the slopes of Merapi with an accuracy rate of 84.6%, from the results of taking 10% of the training data. The output of this system is an information system shown in vulnerability map that provides information about the status of susceptible zones to lahar flow.