ML-Enhanced 3D Probabilistic Path-Tracing and Volumetric Viewshed Analysis

Traditional archaeological modeling often relies on "least-cost paths" that only account for terrain steepness, treating human movement like water flow. Similarly, standard visibility analysis is typically 2-D and planimetric, ignoring building heights, window apertures, and translucent partitions. This project proposes an ML-enhanced framework for the Necropolis of El Bagawat. By using multispectral ground surface classification (SAR-MS.tif) to weight movement costs and incorporating building entrance orientations from architectural plans (Site_Plan.pdf), I will develop a probabilistic path ensemble that reflects social and physical realities. Furthermore, I will implement a true 3D ray-casting viewshed that accounts for 3D geometry and window openings, delivering high-fidelity GIS vector layers and 3D volumes to rewrite the embodied history of this late antique site.