|
Abstract:
|
With the recent advancements in GIS software and related tools, spatial modelers are increasingly encountering massive spatial datasets which entail computations that become prohibitive. There is a burgeoning literature on approaches for analyzing large spatial datasets. In this article, we propose a mutiscale spatial regression model that employs a divide-and-conquer strategy within the Bayesian paradigm. We develop a multiscale spatial kernel convolution technique with higher order functions to capture fne scale local features and lower order terms to capture large scale features. To achieve parsimony, the coefficien in the multiscale kernel convolution model is assigned a new class of "Tree shrinkage prior" distributions. Excellent empirical performances are illustrated using several simulation experiments and a geostatistical analysis of the sea surface temperature data from the pacific ocean.
|
