Update (2019-10-24): I totally messed up the definitions the first time I posted this! I’ve fixed it now. Many thanks to commenter szcfweiya to pointing this out!
Let be a stochastic process, where is the index set for the stochastic process. (Most often we have to index time or to index space). The stochastic process has an associated covariance function such that for any , .
In general, a covariance function must satisfy two properties:
- It is symmetric, i.e. for all , and
- It is positive semi-definite, i.e. for all , , , .
A covariance is isotropic if depends only on the distance . A covariance is said to be anisotropic if it is not isotropic. That is, either does not depend on the distance , or it depends on as well as some other functions of and .
Clearly the class of isotropic covariances is much smaller than that of anisotropic covariances. To model anisotropic covariances, one usually has to make an assumption on how depends on and .
In my little googling around, anisotropic covariance modeling seems to be popular in geostatistics and more broadly, spatial statistics. One popular example of anisotropic covariance is called geometric anisotropy. In that setting, the index set of the stochastic process is (typically with or ). Isotropic covariance in this setting would have the form , where is some function and is the Euclidean metric. (See this earlier post for some examples. Not all kernels there are isotropic but it should be obvious which are.) Geometric anisotropy refers to a covariance of the form , where is some other distance metric. Some examples (from Reference 1) are the Mahalanobis distance and the Minkowski distance.
A special case of using the Mahalanobis distance is where for some diagonal matrix . This corresponds to giving each axis a different scale before computing the Euclidean distance.
- Haskard, K. A. (2007) An anisotropic Matérn spatial covariance model: REML estimation and properties.