Spatial Point Process Modeling for Forest Fires in New Brunswick

The locations of forest fires in New Brunswick were treated as a point pattern and modeled by an inhomogeneous Poisson process and an Baddeley-Geyer process. Three geographic variables were used as covariates in the modeling, which was conducted using the package spatstat in R.;The forest fire data and the covariate data were converted into formats that can be used by spatstat. Inhomogeneous Poisson process models were fitted to model the spatial trend and to provide an intensity estimate for calculating the inhomogeneous K-function. Since both a Q-Q plot and the inhomogeneous K-function showed that there is an attraction interaction in the data, the Baddeley-Geyer process was introduced as an alternative model to the inhomogeneous Poisson process. The Baddeley-Geyer model was found to provide a better fit to the data than the inhomogeneous Poisson process.