Year
2006
Abstract
In addition to high-resolution panchromatic imagery, multispectral and hyperspectral imaging are now beginning to be used by safeguards regulators to help characterize nuclear-related materials. Advances in hyperspectral remote sensing have resulted in faster pre-processing times, better calibrated datasets, and improved mapping techniques. However, in the absence of reliable ground truth data and incomplete nuclear-based spectral libraries, mapping nuclear-related materials from hyperspectral imagery is still a challenge. This paper proposes a systematic approach to mapping uranium mines and deposits from hyperspectral data in the absence of local ground data. The method is based on classical uranium deposition models and supporting materials obtained from large, known operating uranium mines and processing plants. The primary features of each model and mine are identified and tabulated, and a custom spectral library is then compiled for uranium ores, host rocks, rock assemblages, noneconomic rocks and minerals, and alteration/weathering products. Secondary materials and byproducts produced or needed by mines and mills are also included. Using this concept, a preliminary hyperspectral examination of a uranium mine for one depositional model is presented. High-grade ore at Ranger Mine, Australia is differentiated from lower grades on the basis of their spectral signatures, and tracked to different locations on the mine site.