COMPUTER-ASSISTED DECISION AID FOR THE ESTIMATION OF MINERAL ENDOWMENT: URANIUM IN THE SAN JUAN BASIN, NEW MEXICO, A CASE STUDY.

Abstract

The Arizona Appraisal System is a totally integrated methodology that uses a series of interactive computer programs to translate subjective geologic opinion into a probabilistic estimate of mineral endowment. By "totally integrated methodology" is meant a unified, conceptually complete approach. This methodology comprises two main sections, each executed on a different computer system. The first section, the Geologic Decision Model, has been computerized as an interactive PLATO program. Using the PLATO system, the geologist describes probabilistically the perceived states of geologic processes and conditions. The decision model analyzes this information and computes a probability distribution for mineral occurrence. The second section, the Endowment Simulation Model (program MASTER), is run on the DEC 10 and Cyber 175 computers. Program MASTER takes the product of the Geologic Decision Model, combines it with other data, and produces a probabilistic estimate of mineral endowment for the region being evaluated. Development and testing of the Arizona Appraisal System were carried out simultaneously over a period of about three years. During this period, four geologists from government and industry were called upon four or five times over a period of about a year for a study of the uranium (U₃O₈) endowment in the San Juan Basin of northwestern New Mexico. The results produced by the system consist, for each geologist, of a probability distribution for tons of U₃O₈ endowment for (1) each partition of each stratigraphic unit, (2) each stratigraphic unit as a whole, (3) "formations" or "merged units" (groups of stratigraphic units), and (4) the San Juan Basin as a whole (all stratigraphic units). The system also calculates the average distribution across all geologists for the various merged units and for the basin as a whole. The result for the basin as a whole (in thousands of tons) is: mean 3,855, variance 4,108 x 10⁹, and 95th percentile 6,541. The author believes that his major contribution has been to design and implement a working resource estimation methodology that is flexible with respect to commodity and geographic location.