Improved Urban Impervious Surface Mapping Using Multi-Sensor Feature Fusion: A Cross-City Analysis

• DOI: 10.20944/preprints202409.1033.v1
• Publication Date: 2024-09-13T01:17:23Z
• Authors: Muhammad Nasar Ahmad; Fahad Almutlaq; Md. Enamul Huq; George D. Bathrellos; Akib Javed; Fakhrul Islam; Hariklia D. Skilo...
• Keywords: Python; Impervious surface; Sensor Fusion; Land Cover

The study put forward a data fusion approach for urban remote sensing that combines SAR (Synthetic Aperture Radar) and optical satellite data. By integrating datasets from different sensors spatial-temporal scales, the technique aims to extract more accurate information. utilizes two methods: feature-based fusion, where relevant features are extracted fused, simple layer stacking (SLS), original directly stacked as multiple layers. This using textures (using Sentinel-1) modified indices Landsat-8), then classified these an XGBoost algorithm implemented in Python Google Earth Engine. Researchers examined five cities, each representing distinct climatic zone dynamic: Cape Town, Guangzhou, Los Angeles, Mumbai, Osaka. An accuracy assessment was conducted random validation points, achieving overall of 89.5% proposed MSFI method. A comparison also performed with three well-known global products. approach, outperformed all products achived 89% while ESA (84%), ESRI (81%) Dynamic World (82%). Additionally, Land surface temperature analysis accomplished investigate relationship between UIS Surface Temperature (LST) across selected cities show practical use warm temperate city, showed highest LST among cities. datasets, along GEE codes, available at https://github.com/mnasarahmad/sls.