This report describes the results obtained from deployment of the Enhanced Site Characterization System (ESCS) at the Hanford Site's 618-4 Burial Ground. The objective of this deployment was to use advanced geostatistical methods to integrate and interpret geophysical and ground truth data, to map the physical types of waste materials present in unexcavated portions of the burial ground. One issue of particularly interest was the number of drums (containing depleted uranium metal shavings or uranium-oxide powder) remaining in the burial ground and still requiring removal. Fuzzy adaptive resonance theory (ART), a neural network classification method, was used to cluster the study area into 3 classes based on their geophysical signatures. Multivariate statistical analyses and discriminant function analysis (DFA) indicated that the drum area as well as a second area (the SW anomaly) had similar geophysical signatures that were different from the rest of the burial ground. Further analysis of the drum area suggested that as many as 770 drums to 850 drums may remain in that area. Similarities between the geophysical signatures of the drum area and the SW anomaly suggested that excavation of the SW anomaly area also proceed with caution. Deployment of the ESCS technology was successful in integrating multiple geophysical variables and grouping these observations into clusters that are relevant for planning further excavation of the buried ground. However, the success of the technology could not be fully evaluated because reliable ground truth data were not available to enable calibration of the different geophysical signatures against actual waste types.
PDF
download
878987797
028-85220240
