A new multiscale prediction-error filter for sparse data interpolation

William Curry and Morgan Brown

bill@sep.stanford.edu, morgan@sep.stanford.edu

ABSTRACT

Prediction-error filters (PEFs) have been used to successfully interpolate seismic data. Using conventional methods, PEFs often cannot be estimated on sparse, irregularly sampled data. We implement an algorithm in which we resample the data to various scales to estimate a single PEF. We show that compared to PEFs estimated from single data scales, our PEF provides a robust first guess for a nonlinear interpolation scheme.