Problems of using rock volume data in predictive resource studies

Abstract

The use of rock volume data and an index of diversity of rock type within 5.5 X 5.5-km map cells is compared with the use of an integrated score (based on a weighted combination of geochemical, geological, and geophysical attributes derived from the hydrogeochemical and Stream Sediment Reconnaissance phase of the National Uranium Resource Evaluation program of the U.S. Department of Energy) as regression-based predictors for potential uranium production in the 19,700-km 2 area of the Pueblo quadrangle, Colorado. Ridge regression is shown to be a powerful tool for the identification of stable predictors in data with inherent correlation, such as the rock volumes. While diversity, characterized by an entropy measure, is not significantly correlated with uranium past production from map cells containing uranium occurrences, it is correlated with the weighted scores for the map cells which themselves form a useful predictor for potential uranium yield per cell.