Track 1: AI and Data-Driven Decision Making

• Geological data: Regional geology and structural features, including faults and contacts • Geochemical data: Multi-element geochemical datasets • Remote sensing data: Sentinel-1 Synthetic Aperture Radar (SAR), including VV and VH polarizations From these datasets, additional features were derived to enhance geological representation. In particular, a distance-to-plutonic-intrusions layer was generated from regional geological maps, capturing spatial relationships between intrusive bodies and potential mineralization. All raster datasets were standardized to a common spatial grid of 2199 × 2367 pixels, with a uniform spatial resolution of 100 m per pixel. 3.2 Selection of Layers The selection of layers was guided by geological knowledge of porphyry copper systems and supported by previous studies in mineral prospectivity mapping. The following six raster layers were selected: • Distance to plutonic intrusions • Digital Elevation Model (DEM) • RTP (magnetic intensity) • RTP-TDR (magnetic Tilt Derivative) • SAR VV polarization • SAR VH polarization The inclusion of distance to plutonic intrusions is critical, as porphyry deposits are genetically associated with magmatic-hydrothermal systems and are typically spatially related to intrusive bodies (Sillitoe, 2010). Magnetic data, including RTP and its derivatives, are widely used to delineate subsurface structures, alteration zones, and lithological contrasts associated with mineralization (Bahi et al., 2024). Topographic information derived from DEMs can provide indirect insights into structural controls and geomorphological expressions of mineral systems (Zhao et al., 2023). Synthetic Aperture Radar (SAR) data, particularly dual-polarization (VV and VH), have been shown to enhance the detection of surface roughness, structural features, and alterationrelated patterns, making them valuable inputs for mineral exploration models (Guha et al., 2013).

RkJQdWJsaXNoZXIy MTM0Mzk2