Interpolation of incompletely and non-uniformly sampled survey data for buried radioactive point source detection

At the US Army´s test sites, fired penetrators made of Depleted Uranium (DU) have been buried under ground and become hazardous waste. The first step for the detection and mitigation of DU contamination is surveying a site. Typically a survey is performed in an incomplete and non-uniform manner, causing difficulties for the subsequent detection techniques. In this research, we study how to interpolate such survey data into a uniform grid. We derive an inverse distance covariance matrix Ȃ, to map an incomplete and non-uniform sample vector v into g. We demonstrate that the information contained in Ȃ and g̑ can be used for interpolation through linear Kriging, to yield a uniform grid. We verify the effectiveness of this approach through simulation experiments.