Locally supported kernels for spherical spline interpolation (Q1356799)

The author proves several results for spherical spline interpolation in the following sense. It is assumed that a set of scattered data on the sphere satisfies a certain admissibility condition. The desired interpolant to this data set consists of a spherical harmonics term and a component generated by a reproducing kernel, and it minimizes a certain seminorm. In this general setting, the focus of this paper is the introduction of reproducing kernels which are both strictly positive definite and locally supported, thus ensuring a uniquely determined solution of the interpolation problem and computational advantages. Polynomial precision, i.e., the reproduction of spherical harmonics in this interpolation procedure, is also an issue.