Digital Mapping of Soil Vanadium Across Australia

Vanadium (V) is increasingly recognised as a critical mineral due to its potential for decarbonisation technologies. Australia holds an estimated quarter of global V resources, yet there is limited knowledge of how these resources are distributed across the country. Here, we employed digital soil mapping techniques to map the distribution of soil V content at a 90 m resolution. Using remotely sensed data as environmental covariates (i.e., barest Earth Landsat imagery, gamma‐ray spectrometry maps) and layers representing soil, climate, and topography, we calibrated Cubist machine learning models on 1315 nationally sampled data points of aqua‐regia‐extracted V content. Our models performed reasonably well in predicting V content in top outlet sediment (0–10 cm) and bottom outlet sediment (on average ~60–80 cm), with an average concordance correlation coefficient of 0.49 and 0.52 on the validation data. Independent validation using top outlet sediment data from northern Australia ( n = 780) demonstrated that our models maintained consistent accuracy. Feature importance ranking revealed that spectral images at barest soil condition, climate and soil texture were the most influential predictors. High V contents predominantly occurred on Kandosols and iron‐rich Tenosols, particularly in Western Australia. This first explicit prediction of soil V content provides essential baseline information for sustainable resource planning and management across Australia.