Mapping corroded and non-corroded roofs with worldview-3 imagery: A study of semi-permanent buildings in fire-prone urban areas

Rapid urbanization in the West Banjarmasin District, Banjarmasin City, South Kalimantan Province, Indonesia, has led to the expansion of dense informal settlements, where semi-permanent housing with vulnerable roofing materials significantly increases fire risk and hampers emergency response efforts. To address this issue, this study investigates whether remote sensing techniques can effectively differentiate building roof types, particularly corroded and non-corroded corrugated iron roofing commonly used in fire-prone structures. The Normalized Difference Iron Oxide Index (NDIOI), formulated as (NIR2 – Blue) / (NIR2 + Blue), was applied using very high-resolution WorldView-3 imagery (0.5 m) to detect spectral signatures of roof degradation. A total of 200 surface reflectance samples (100 corroded, 100 non-corroded) were used to calibrate and validate NDIOI through threshold-based binary classification (threshold = 0.1). Validation demonstrated strong performance, with accuracy of 93.0 , precision of 100 , recall of 86.0 , and an F1-score of 92.5 . Additionally, spatial overlay between NDIOI-classified zones and verified fire incident data acquired during the same time frame as the 2021 imagery revealed a strong spatial correlation between high NDIOI values and actual fire-prone areas. © 2025