Space-Time Gaussian Process Approximations

This repository provides the R code for the simulation studies and real-world applications presented in the paper "Scalable non-separable spatio-temporal Gaussian Process Models for large-scale Climate Data."

The iterative methods for full-scale approximations are implemented in the GPBoost package, available here: https://github.com/fabsig/GPBoost.

Repository Structure

• Data
  Provides scripts for generating data and the actual data for both the simulation studies and real-world experiments.

• Simulation_Studies
  Contains scripts to run the simulation experiments. Refer to the README.md file in the simulation_studies folder for a detailed description of the various simulations and how to execute them.

• Real_World_Application
  Includes scripts for the real-world application on daytime land surface temperature data.