New methods for detection and estimation of anisotropy was developed at the Technical University of Crete. These methods are very fast, and are used to detect geometric anisotropy in potentially large data sets. This enables us to significantly speed up the likelihood based estimators in the psgp and copula methods, by correcting for the anisotropy that is detected.
A description of the method can be found in the following papers and reports:
Deliverable 2.2 Bayesian spatial analysis and anisotropies
Chorti, E., and D. T. Hristopulos (2008) Nonparametric identification of anisotropic (Elliptic) correlations in spatially distributed data sets. IEEE Transactions on Signal Processing, 56 (10 I): 4738-4751.