Communicated by H. NeunzertVector spherical interpolation is discussed from both the theoretical and computational points of view. The theory of vector spherical harmonics is an essential tool. An estimate is given for the absolute error of the interpolation process; an efficient algorithm is developed for the computation of a vector spherical interpolant. The displacement boundary value problem of determining the elastic field from a finite number of discretely given displacement vectors is solved by the use of vector splines.
IntroductionNumerous papers concerned with the scalar interpolation theory of spherical splines have appeared in the last decade. Scalar spherical splines (SSS) have been analyzed in depth both in their theoretical and computational aspects (cf., e.g., [8-10, 12,213 and the references therein) and have been found to be natural generalizations of polynomials, that is spherical harmonics, having desirable characteristics as interpolating functions; SSS can be recommended for the numerical solution of various interpolation and best-approximation problems in geosciences. In particular, S S S are best suited for the macromodelling and micromodelling of the Earth's gravitational field from discretely given data on the Earth's suface (cf., e.g., [4,5, 10, 16, 171). Moreover, based on the ideas of the scalar theory, a successful application of spherical splines has been given to the analysis of meteorological data In this paper we establish a general vectorial concept of spherical spline interpolation. More explicitly, our focus will be on the determination of a spherical vector field from discretely given normal and tangential components. It turns out that interpolation by vector spherical splines (VSS) essentially amounts to solving a well-posed (cf. [22]).