v0.3.4

• Add empirical omnidirectional semi-variogram
• Add plot recipes for empirical and theoretical variograms

v0.3.3

• Fix spherical variogram model
• Add more tests

v0.3.2

• Add ellipsoid distance in 2D and 3D as a way of modeling anisotropy.
• Add haversine distance between two latitude/longitude pairs.
• Replace immutable by struct and type by mutable struct in Julia v0.7.

v0.3.1

• Use Parameters.jl until default field values become part of the language (see JuliaLang/julia#10146).
• Add anisotropic variogram models by means of custom distance functions.
• Add composite (additive) variogram model.

v0.3.0

This is the first release of the package with support to Julia v0.6. We kindly ask users to update to the current version of the language. Below is the list of changes:

• Add Matérn family of variogram models
• Require SpecialFunctions.jl on Julia v0.6 and later
• Require Julia v0.6

v0.2.1

• Fix remaining issues with Float64 vs. Float32 precision as discussed in #9.

v0.2.0

In this release, all covariance models were replaced by the corresponding variogram models. This is a breaking change in preparation to future developments with Julia v0.6+.

The documentation was updated accordingly. We kindly ask users to update their code to the new API.

v0.1.3

• Pass vector of observations to fit! method
• Add documentation for Kriging estimators

v0.1.2

• Fix issue #6 by explicitly choosing a factorization (e.g. Cholesky or LU) for the different Kriging variants.
• Separate fit! and estimate methods. The former builds the Kriging system once and saves the factorization. The latter is used to estimate at different locations without overhead.

v0.1.1

Fix issue #2 by exporting the types.