INTEGRASI TEKNOLOGI PENGINDERAAN JAUH DAN MACHINE LEARNING PADA WEB GIS UNTUK PEMETAAN POTENSI BANJIR

Abstract

Flooding is a natural phenomenon that has frequently posed significant challenges in various regions of Indonesia, driven by factors such as rainfall, river conditions, upstream landscapes, land use patterns, and sea-level rise. These events often lead to severe consequences, including the spread of waterborne diseases, destruction of infrastructure, depletion of natural resources, and economic disruption. One proactive measure to mitigate such impacts is mapping potential flood risk areas. This study utilized Landsat 8 satellite imagery Level 2, Collection 2, Tier 1 processed on the Google Earth Engine (GEE) platform to derive indices such as the Digital Elevation Model (DEM), Topographic Position Index (TPI), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Water Index (NDWI). These indices served as input variables for a Random Forest model, classifying areas into high, medium, and low flood risk categories. The developed model achieved 86% accuracy when evaluated using a confusion matrix, with precision, recall, and F1-score metrics validating its performance. The integration of this model into a WebGIS service was implemented through Flask, offering an API that supports real-time flood risk data retrieval by third-party applications. The front-end interface, built using LeafletJS, provides an interactive and user-friendly map visualization of flood risk levels. The results demonstrate that the Random Forest model effectively classifies flood risk, while the WebGIS service offers a practical tool for visualizing and disseminating flood risk information. This service has the potential to support disaster management efforts and enhance community preparedness against flooding.