Plume Characterization with Natural Neighbor Interpolation
Engineering Computer Graphics Laboratory,
Brigham Young University, Provo, UT.
Abstract
A relatively new interpolation scheme called "natural neighbor interpolation" is described. Natural neighbor interpolation has several features that make it attractive for use in three-dimensional contaminant plume characterization. The subset of data used for interpolation is defined by topological rather than distance-based relationships. As a result, natural neighbor interpolation is easy to use and performs very well with the clustered data sets that are typically encountered in plume characterization. A three-dimensional version of natural neighbor interpolation developed by the authors is compared to inverse distance weighted interpolation and kriging using an elliptical control function. For this particular case, natural neighbor interpolation is shown to be more accurate than the other two schemes with respect to mean residual error.