Analisis Tingkat Kerawanan Banjir Menggunakan Machine Learning Artificial Neural Network (ANN) dan Geospasial

Abstract

Floods pose a significant threat to the Asahan Toba Watershed in North Sumatra, primarily driven by high rainfall, land use changes, and geomorphological conditions. This study aims to identify flood-prone areas and predict flood risk using an Artificial Neural Network (ANN). Key parameters include rainfall, river proximity, elevation, soil type, slope, and land cover. Data were processed with Geographic Information Systems (GIS) and modeled in TensorFlow using Python. The ANN achieved an accuracy of 85% and was validated against real flood events in 2024. Flood risk zoning maps highlight high-risk areas, particularly in Bandar Pulau and Simpang Empat sub-districts. The findings confirm that ANN is an effective tool for spatial flood risk prediction and early warning, supporting disaster mitigation and watershed management strategies.