Fusing Sentinel-1 and Sentinel-2 Data in Google Earth Engine for Road Infrastructure Mapping in Data-Scarce and Conflict Environments Mali Shadrack Paul1, Mitsuharu Tokunaga2
1Department of Civil and Environmental Engineering, Graduate School of Engineering, Kanazawa Institute of Technology
2Department of Civil and Environmental Engineering, Graduate School of Engineering, Kanazawa Institute of Technology
* mwashorg2020[at]gmail.com
Abstract
Accurate road infrastructure data is vital for planning, mobility analysis, and disaster response, yet in conflict-affected and data-scarce environments such as South Sudan, authoritative sources are scarce and often outdated. Traditional field surveys remain difficult due to insecurity and cost, while optical imagery is frequently obscured by persistent cloud cover. This study presents a cloud-resilient, low-cost workflow for road mapping in Juba County that fuses Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical imagery within Google Earth Engine (GEE). Imagery from December 2024 to March 2025 (Sentinel-2) and the full 2024 year (Sentinel-1) was composited using pixel-wise median stacking with selected bands (B2, B3, B4, B8, B11, B12). Four fused tiles were exported to QGIS for marging, visualization, digitization, and comparison with OpenStreetMap (OSM), Geofabrik, and Google Satellite data. A Select-Zoomed-In road network visibility analysis (RNVA) demonstrated enhanced detection of paved and unpaved roads compared to single-source data. Results revealed numerous unmapped segments and outdated classifications in existing datasets. The integration of Informed Volunteered Geographic Information (IVGI), derived from the researcher^s local knowledge of Juba roads, further improved classification accuracy. The outputs provide a GeoAI-ready dataset for future automated road surface detection, contributing to closing data gaps in fragile regions.
