This paper presents a method for approximating spherical functions from discrete data of a block-wise grid structure. The essential ingredients of the approach are scaling and wavelet functions within a biorthogonalisation process generated by locally supported zonal kernel functions. In consequence, geophysically and geodetically relevant problems involving rotationinvariant pseudodifferential operators become attackable. A multiresolution analysis is formulated enabling a fast wavelet transform similar to the algorithms known from classical tensor product wavelet theory.
Main IdeaSeveral different concepts for the construction of wavelets are known in the literature. For an overview, see [8]. Within these different methods, two essential approaches can be identified: The first one uses certain grids on the sphere and tries to transfer concepts of the one-dimensional wavelet analysis to the sphere. The approaches with longitude/latitude grids or triangulations fall into this category. The second approach put the emphasis on the solution of rotation-invariant pseudodifferential equations on the sphere (as proposed, e.g., in [3]). For this concept, zonal functions are an appropriate tool. In this case, there is -at first glance -no need for a particular point distribution on the sphere.In this paper we are interested in a new compromise for the construction of spherical wavelets. In fact, our wavelets are based on zonal functions but also relate to structured grids in order to obtain fast algorithms that are easy to implement.The key points of our paper can be characterized as follows:• The wavelets and the scaling function are based on zonal kernel functions so that our approach is well-suited for the solution and regularization of rotation-invariant pseudodifferential equations.• The wavelets are locally supported so that the integration can be made efficient.• The construction is based on a biorthogonal system of zonal functions, which gives us almost all advantages of an orthogonal approach.• A new block-wise grid structure is used on the sphere: The points lie on circles of constant latitude to ensure that the construction of a biorthogonal system is not expensive. But the grid gets sparser in the polar regions so that there are not too large differences in the distances between neighboring points.• The grid is organized in blocks of size 2 p by 2 p−1 so that the algorithms can be formulated easily.• Finally, a scaling equation is established with only a few coefficients. In fact, we end up with a fast wavelet transform which is completely similar to the algorithms known from tensor product approaches of Euclidean wavelet theory.What is the prize that we have to pay for all these advantages? It is by no means dramatic: First, we do not develop a multiresolution analysis of the whole space of square-integrable functions but on finite dimensional subspaces. In fact, we have -at the beginning of the analysis -to choose a finest level of resolution. From numerical point of view, this is what has to be done for p...