Software

R software: Package GeoModels (on GitHub)

Coauthored with Moreno Bevilacqua.

This package provides a set of procedures for:

a) Simulation of Gaussian and some non Gaussian spatio (temporal), bivariate random fields.

b) Analysis of spatial, spatio-temporal and bivariate random fields using standard and composite likelihood. Data analysis involves estimation of both regression and covariance parameters for Gaussian and some non Gaussian random fields with marginal distribution of different types such as SkewGaussian, Student T, Weibull, Gamma, LogGaussian, Binomial, Negative binomial.

c) Kriging using optimal (linear) prediction.

Some feature of the package:

- Estimation of both regression and covariance parameters.

- Analysis of space time data with temporally dynamic spatial locations.

- Analysis of bivariate spatial data for both isotopic and heterotopic case.

- Analysis of georeferenced data given in lon/lat format, using great circle or chordal distance.

- Computation with GPU parallelization through OpenCL

- Support for sparse matrices algorithms using R spam package.