README

This repository is associated with the paper Non-separable spatio-temporal Poisson point process models for fire occurrences by Nicoletta D’Angelo, Alessandro Albano, Andrea Gilardi, and Giada Adelfio. It contains the code and the data necessary to reproduce the analyses presented in the manuscript.

In particular:

• The R code used to create the exploratory analyses included in Sections 2 and 4 and estimate the model detailed in Section 3 can be found in summary.R.

• The file summary.html (generated by summary.qmd) contains a notebook-like summary of the analyses in HTML format which already includes all code and figures. The HTML file is too large to be displayed directly on Github, but it can be downloaded and opened in a browser.

• The data (~1.5GB) can be downloaded from the data-release of this repository using the following code:

library(piggyback)
data_names <- c(
  "DL_FIRE_J1V-C2_510187.zip", "confini-regioni.zip", "land-use.zip", "NDVI.zip",
  "environmental-variables.zip", "INGV-elev.zip"
)
pb_download(
  file = data_names,
  dest = tempdir(), # Specify the path where you want to save the data
  repo = "agila5/ppm-fire-occurrences",
  tag = "v1-data",
  overwrite = TRUE,
  show_progress = TRUE
)

A small variation of this code chunk is already included in summary.R and summary.html.

• We have already cached the long running computations in the qcache folder using the qs::qcache function. This folder is not fully included in the Github repository, but summary.R and summary.html include the relevant code to download the missing objects from the data-release of this folder. Feel free to remove the cache directory or the qs::qcache function from the code if you want to re-run such long computations.

• Finally, we have also prepared a Github Codespace that can be used to automatically set-up an Rstudio Server and explore the code and the data in an interactive way.

Instructions to access the Github Codespace¹:

1. Click on the following link to set-up your own version of the Codespace: https://codespaces.new/agila5/ppm-fire-occurrences. Warning: the Machine type option needs to be set equal to 4-core.
2. Click on the green button named Create codespace.
3. Wait for the Codespace to be created. This operation takes approximately 5/10 minutes. At the end you should see something like:
4. Click on the Ports tab (which is highlighted in the previous image). You will see something like
5. Drag the mouse over the Forwarded Addresses field in the 8787 port and click on the globe icon (as displayed in the previous image). This will open a new tab in your browser with the Rstudio Server interface.
6. Login into the Rstudio server using the following credentials:
   • Username: rstudio
   • Password: rstudio
7. These operations will create an Rstudio session in your browser. Now you can explore the code and the data interactively!

Warning: Please notice that

GitHub Codespaces is paid for either by an organization, an enterprise, or a personal account. The Free and Pro plans for personal accounts include free use of GitHub Codespaces up to a fixed amount of usage every month.

We refer to the official docs for more details.

Footnotes

1. Please note that you need a Github account to access the Github Codespace. ↩