Spherical spline interpolation - basic theory and computational aspects (Q762383)

In recent years the method of spherical spline interpolation has been analyzed in depth, both in theoretical and computational aspects [cf. the author, Veröff. Geod. Inst. RWTH Aachen, 27 (1979; Zbl 0446.65009) Math. Meth. Appl. Sci. 3, 551-575 (1981; Zbl 0481.41007), \textit{G. Wahba}, SIAM J. Sci. Stat. Comput. 2, 5-16 (1981; Zbl 0537.65008)]. Spherical spline functions turn out to be natural generalizations to the sphere, of periodic splines on the circle [cf. e.g. \textit{I. J. Schoenberg}, J. Math. Mech. 13, 795-825 (1964; Zbl 0147.321)] and surface splines in Euclidean spaces [cf. e.g. \textit{J. Meinguet}: Numerische Integration, Tag. Oberwolfach 1978, ISNM Vol. 45, 211-220 (1979; Zbl 0422.41004)]. Spherical spline functions are more adaptable approximating functions than the spherical harmonics; they are ''smooth'' functions avoiding larger oscillations and severe undulations [cf. \textit{R. Reuter}, Veröff. Geod. Inst. RWTH Aachen 33 (1982), the author and \textit{R. Reuter}, Methoden Verfahren Math. Phys. 27, 73-104 (1983), the author, Z. Angew. Math. Mech. 65, T 260-T262 (1985)]. Moreover, in analogy to the interpolation method due to \textit{J. Meinguet} [Z. Angew. Math. Phys. 30, 292-304 (1979; Zbl 0428.41008)] the whole solution process can be made surprisingly simple and reasonably efficient for numerical applications [cf. the author, OSU Rep. No.341, Dept. Geodetic Sci. and Surveying, The Ohio St. Univ. (1982)]. In this paper the convergence of the solution obtained by spherical spline interpolation is developed by use of estimates for the Legendre polynomial. The ''smoothness'' of the spherical splines is assessed by a seminorm using iterated (Laplace-) Beltrami derivatives on the sphere. A modified approach based on (invariant) pseudodifferential operators on the sphere is given by the author [Z. Vermessungswes. 109, 569-576 (1984)].