An AI-Assisted Multi-Spectral Remote Sensing Framework for Change Detection and Road Damage Index (RDI) in Flood-Prone Regions

Abstract. Road infrastructure in flood-prone regions is highly vulnerable to degradation driven by environmental, hydrological, and climatic stressors. However, quantitative and spatially explicit assessments of such vulnerabilities remain limited, particularly for rural and intercity road networks. This study introduces a comprehensive Road Damage Index (RDI) framework that integrates multi-source satellite observations, meteorological records, and soil-related parameters to quantify road susceptibility in West Azerbaijan Province, Iran. All geospatial layers were preprocessed, normalized to a common scale, and weighted according to their relative contribution to road degradation. The resulting RDI was computed through pixel-wise aggregation, generating a spatially continuous map of road damage potential. Quantitative analysis indicates that over 95% of the 3,652.99 km road network falls within Moderate to Very High vulnerability levels, including 59.3% Moderate, 23.2% High, and 12.5% Very High, while only a small fraction (5.0%) exhibits low susceptibility. Spatial variability analysis reveals that northern and western counties show higher RDI values due to unfavorable soil and hydrological conditions, while southern and central regions demonstrate relatively lower vulnerability. The proposed framework provides a scalable, cost-effective, and reproducible methodology for regional road monitoring, integrating multi-dimensional environmental factors into a single decision-support index. Overall, the RDI framework enhances the capacity for data-driven infrastructure planning, targeted maintenance prioritization, and disaster preparedness, offering a transferable tool for improving road resilience under current and projected climate variability.