Prospectivity model for magmatic Nickel, Copper, Cobalt, and Platinum-Group Elements in the Midcontinent Rift

Bezali Kwame Danso , Adwoa Aduaba Aidoo, Saadatu Abdullah

Much of the Midcontinent Rift is concealed beneath glacial sediment, which limits direct geological observation and makes it difficult to evaluate the potential for undiscovered magmatic Ni–Cu–PGE deposits with publicly available data. The 1.1 Ga rift hosts the Eagle Ni–Cu–PGE deposit of northern Michigan, one of the few domestic sources of nickel, a critical mineral of the United States. We address this problem with public geophysical and geochemical data from the Upper Peninsula of Michigan and northern Wisconsin. First, we develop a knowledge-driven prospectivity model from the new U.S. Geological Survey Earth MRI airborne magnetic and radiometric survey and apply it to 984,215 grid cells of 200 m (39,368.6 km2). The model scores each cell by its similarity to the geophysical signature of a small mafic–ultramafic intrusion, propagates the uncertainty of the evidence weights by Monte Carlo simulation, and down-weights mapped iron formation and Keweenawan volcanic rocks. The Eagle deposit, withheld from model construction, ranks at the 98.9th percentile of the survey area (97.6th to 98.8th across 200 random weightings), so a search restricted to the highest-ranked 1.14% of the area would have included it. Second, we test whether public geochemical fertility indicators (Cu/Pd, Ni/Cu, MgO, and Mg-number) distinguish rocks near known Midcontinent Rift deposits from rocks farther away under spatially independent validation. Although Cu/Pd is lower near deposits (median 21,736 versus 31,568), predictive performance is poor: the area under the receiver operating characteristic curve (AUC) is 0.29 for a 3-km deposit buffer, and 0.48 and 0.35 for alternative buffers. We release 1,629 ranked exploration targets (751.8 km2) with uncertainty estimates, an open-source Python workflow, and an interactive map. The results show that public airborne geophysical data can prioritize concealed critical-mineral targets, whereas currently available regional geochemical data provide only weak predictive discrimination.