gstat

Gstat is an open-source software package for geostatistics implemented for the R programming language. Developed primarily by E. Pebesma and collaborators, the package provides tools for exploratory geostatistics, variogram analysis, and spatial interpolation, including cokriging and multivariate models.

gstat supports univariate and multivariate geostatistics, implementing variogram estimation, variogram modelling with common models (spherical, exponential, Gaussian, and Matérn), and methods for fitting model parameters.

It offers kriging variants such as ordinary, universal, and simple kriging, as well as block and cokriging, and supports linear models of coregionalization for multivariate data.

It integrates with spatial data structures from the sp package (and more recently sf), accepting data frames and spatial objects, and providing a formula-based interface to specify variables and relations.

Outputs include kriging predictions and their estimation variances, residual diagnostics, cross-validation, and variogram parameter estimates.

Applications span environmental, mining, hydrological, agricultural, and ecological studies where estimation at unsampled locations is required.

gstat is distributed under an open-source license and is available on CRAN, where it is maintained alongside documentation and vignettes.

See also: sp, sf, and rgdal packages in R that commonly accompany geostatistical workflows.