Above the Impact Zone: Aerial Imaging with AI-Based Rapid Flood Inundation Assessment Using EXIF-Tagged Imagery and Drone-to-Satellite Comparison

Abstract

Flood assessment in local disaster response is often delayed by limited access to inundated areas and reliance on ground-based observation. This study developed and evaluated a UAV-based aerial imaging system integrated with AI-assisted analysis for rapid flood inundation assessment using EXIF-tagged imagery and drone-to-satellite comparison. The system captures aerial images using a quadcopter platform, embeds GPS-based EXIF metadata, processes images through an AI analysis module, and compares drone imagery with ESRI World Imagery for geospatial validation. Controlled flight tests were conducted at 20 m, 40 m, 60 m, 80 m, and 100 m above ground level, while AI models were evaluated based on flood detection, response time, flood level estimation, and drone-to-satellite alignment. Results showed 100% successful flight trials across all tested altitudes, usable aerial image quality at 1920×1080 resolution, and increased ground coverage at higher altitudes. AI-based flood and non-flood detection achieved 100% accuracy across tested models, while GPT 4.1 Nano produced the fastest response time and Llama-4 obtained the highest mean satellite alignment. Expert evaluation by barangay and MDRRMO personnel indicated that the system was acceptable for flood monitoring and decision support. The developed system demonstrates potential as a rapid aerial assessment tool for local disaster response.