Description And Evaluation of a Full Tensor Interpolation Method

Full tensor gravity gradiometers acquire multiple gradient components that are inherently related to one another. One way in which this redundancy of information can be exploited is enhanced interpolation. That is, the estimation of the field between observation points is made more accurate by the presence of many simultaneously measured components. A method of full tensor interpolation based on the relationships between the gradient components in Fourier domain will be presented and its effectiveness demonstrated in both qualitative and quantitative manner. Such an interpolation method can be evaluated qualitatively by gridding the field arising from a random array of buried source objects as is would be acquired at a given line spacing. The resulting map is then compared to one generated from the same source field using a conventional gridding and only the vertical component of the gravity gradient. For the line spacing chosen, all of the source masses could be located using full tensor interpolation whereas none of them could be resolved using conventional gridding. Resolution can be measured quantitatively by use of a source detection algorithm. This objectively determines whether two nearby sources can be resolved as being separate entities. A Monte Carlo technique was used to measure resolution success as a function of line spacing. For the source model chosen full tensor interpolation was able to match the resolution performance of conventional gridding even if the line spacing is greater by a factor of 1.55.