Porphyry Copper Prospectivity Mapping Using GIS and Machine Learning

Where should an exploration company look for the next porphyry copper deposit? This project builds a 1 km grid over the whole of Arizona (295,138 cells) from free USGS data - the porphyry copper deposit database, the Mineral Resources Data System, the State Geologic Map Compilation, the national magnetic anomaly grid and the isostatic residual gravity grid - and turns them into ten evidence layers with GDAL. It compares the classical mineral- prospectivity methods (weights of evidence with conditional-independence tests, a fuzzy-logic overlay and logistic regression) with machine learning (Random Forest, XGBoost and bagging positive-unlabelled learning) under honest 50 km spatial-block cross-validation, shows how much a random-cell split would overstate the result, explains the maps with SHAP and ranks exploration target areas. A web application on five databases lets a geologist query any cell, follow up targets with field observations and check new discoveries. GNU Octave re-checks the classical computations, a QGIS project with GeoTIFF layers and a complete MATLAB version are included.

Category: Mining Engineering

Mining Engineering Project MN05 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.