Interpolation of Large and Noisy Scatter Data

One often deals with a large number of measurements of natural phenomena in environmental sciences. These data can be frequently irregularly distributed and affected by different type of errors. The problem of interest is to recover the variable field with certain regular properties and a reasonable computational effort. To solve this problem, we use block interpolation followed by a projection onto the wavelet space of cubic splines. In this article, we focus our attention on the block interpolation using Thin Plate and Kriging methods.