Code

Data sets

• Daily US precipitation data (R workspace) provided by the United States Historical Climatology Network
• Climate model output (R workspace) provided by the North American Regional Climate Change Assessment Program
• Gun laws and homicides data (R workspace and spatial coordinates) taken from Kalesan et al. (Supplement and story)
• GPS observations from Reveal Mobile (R workspace and Documentation)

Visualization

• Geostat plotting using fields
• Geostat plotting using ggplot2
• Geostat plotting using google maps
• Areal data plotting using ggplot2

Mean and covariance functions

• Fitting the mean function
• Drawing realizations from a nonstationary Gaussian process (R code)
• Drawing realizations from a Matern covariance (R code)
• A basic spatial analysis of ozone data

Representations of a Gaussian process

• Empirically orthogonal functions
• Realizations of a 1D Gaussian kernel convolution process (R code)
• Compute spectral density and plot sample paths
• Filter a map using FFT (code)

Estimation and prediction

• The empirical variogram (R code)
• MLE estimation (R code)
• Kriging derivation
• Kriging weights (R code)
• Frequentist asymptotics in a simple case

Methods for large datasets

• Timings of matrix operations
• Estimating Matern parameters using a variogram
• Evaluation of the predictive process approximation (R script)
• Parameter estimation using a spectral approximation

Nonstationarity

• Plotting convolution kernel non-stationary processes and R code
• Kernel convolution derivation

Spatial GLMs + Bayesian methods

• Spatial counts with random effects and a copula (R script)
• Non-spatial Gibbs sampling
• Illustration of 1D Metropolis sampling
• Metropolis sampling for non-spatial logistic regression
• Bayesian Kriging with MCMC
• Spatial GLM with MCMC (R script)

Spatiotemporal methods

• Exploratory analysis of precip data
• Spectral analysis of precip data
• Fit a separable model via MCMC
• Realizations of a non-separable process (thanks to Joe Guinness)

Multivariate methods

• Exploratory analysis of climate model output

Areal data models

• Fit a CAR model using MCMC to simulated data for NC
• JAGS code for the CAR model
• Realizations from an autologistic/Potts model

Spatial point patterns

• K functions for simulated point patterns
• Point pattern regression
• Hot spot detection

Spatial point patterns

• GPD distribution
• GPD analysis (CLIMDEX data)
• GEV distribution
• GEV analysis (CLIMDEX data)
• Realizations from a max stable process