Akademik Veri Yönetim Sistemi
Journal of the Indian Society of Remote Sensing, cilt.54, sa.7, ss.3037-3054, 2026 (SCI-Expanded, Scopus)
Accurately extracting urban impervious surfaces (UIS) from satellite imagery is essential for assessing environmental impacts and supporting sustainable urban planning. This study presents the Multi-Index Impervious Surface (MIIS) index, a proposed method for effectively extracting UIS from PlanetScope and similar satellite imagery with limited spectral bands. By integrating multiple spectral indices through Principal Component Analysis (PCA), the MIIS index enhances the spectral separability of impervious surfaces, which is subsequently evaluated using a consistent Random Forest (RF) classification framework. The method was rigorously evaluated across diverse urban areas, demonstrating robustness and wide applicability. MIIS consistently outperformed traditional indices and classification algorithms, achieving the highest classification accuracy and Kappa coefficients, with overall accuracy (OA) reaching 0.94, exceeding that of the standalone RF classifier. Statistical significance was confirmed using the McNemar test (p < 0.05). Application of the method to Sentinel-2 satellite imagery further demonstrated the applicability of MIIS to other freely available sensors with visible and NIR bands. Separability analysis confirmed MIIS’s superior ability to distinguish various land cover types. These results highlight the MIIS method’s interpretability, applicability across sensors, and minimal training data requirements. These features make it a robust and scalable tool for operational urban monitoring and environmental management worldwide.