The purpose of a statistical geophysical survey is to clearly delimit clean and contaminated areas with an acceptable level of confidence at an acceptable cost. In addition to the survey specifications, the survey approach must quantify the detection confidence, the potential risks, and the cost savings. Protocols must be mathematically supportable as the foundation of an overall engineering process and risk assessment that is acceptable to external regulators and to the general public whose safety depends upon it. This project makes use of the unique advantages of sensor arrays in order to reduce the required sampling area. Rather than dividing a homogeneous sector into a scattering of small grids for detailed investigation, a more uniform distribution of array swaths is utilized to delineate the boundaries of contamination with higher confidence and lower cost. The advantage to this approach is that ordnance density calculations can be based on actual measurements across the entire sector. This removes the statistical assumption of homogeneity between grids within a sector. Although the technology to conduct these surveys exists, the mathematical foundations and statistical protocols for this survey methodology have never been derived.
This project seeks to identify the mathematical foundations and statistical protocols in the domain of point process theory of spatial statistics by focusing on three objectives: (1) develop the statistical spatial models required to produce the mathematical foundation for UXO distribution characterization, (2) develop optimal sampling strategies through experimental survey design, and (3) improve confidence levels for contamination estimates from measured data by improving discrimination techniques.
The technical approach for the investigation and development of the statistical sampling methodology consists of the following tasks/phases: (1) mathematical development of spatial point pattern models, (2) forward modeling of UXO shapes and orientations, (3) mathematical development of optimal survey design models, (4) analysis of existing surface and airborne data sets for detection limitations and statistical reduction of geologic and cultural noise effects, (5) evaluation of design models by comparison to existing geophysical data, (6) evaluation of geophysical attributes to distinguish UXO-related sources from other man-made and geologic sources, (7) classification of anomalies in the point pattern data set, (8) integration of geophysical response parameters with the spatial distribution models and refinement of the density distribution map, and (9) package of the entire methodology into a suite of software routines.
This project will provide the Department of Defense with a set of engineering specifications and confidence levels for a statistical geophysical sampling program that makes use of the most recent technological advances in cost effective survey equipment and that is defensible to regulators and the public. The methodology will estimate point process models from partial point map data and provide a complete map of UXO contamination density on large geographic sites. Additionally, the methodology will enable optimal sample placement providing for more cost effective evaluation of the data and results.