Wavelets, Filters and Spatial Stochastic Processes

Wavelet methods have proven useful in the analysis and synthesis of one-dimensional processes (e.g., time series) and a natural extension is to spatial processes. The key to wavelet-based simulation and estimation of stationary spatial processes is the ability to produce approximately uncorrelated wavelet coe#cients. Using standard filtering techniques, we explore the spectral properties of the two-dimensional wavelet decomposition of a popular class of spatial covariance functions