BC Wildfire Predictive Service Labs#

← Home Page

Welcome to the BC Wildfire Predictive Service (wps) Labs β€” a growing collection of hands-on lab tutorials that bridge the gap between advanced technology and practical wildfire management.

Our mission is to increase awareness, collaboration, and education on the applications of remote sensing and machine learning to wildfire research.

By increasing exposure to the challenges of wildfires amongst the wider Data Science community, we strive to empower leaders with the tools to make optimal decisions which minimize the loss that communities and wildlife suffer.

Each tutorial is carefully crafted to build foundational knowledge while tackling real-world challenges, such as early fire detection, perimeter mapping, and burn severity assessment. Whether you’re a data scientist looking to apply your skills to environmental challenges, or a wildfire researcher seeking to leverage machine learning, you’ll find resources here to support your journey.

πŸ“š What’s Inside#

This book collects the notebooks from the repository’s two labs:

Intro Wildfire Mapping Lab

  • Mapping BC Wildfires with Google Earth Engine β€” get started building wildfire perimeters with Earth Engine, geemap, and folium.

Bayesian Wildfire Stats Lab

  • Bayesian Simple Linear Regression with PyMC β€” a first pass at Bayesian regression on wildfire data, and the limits of what a simple model can tell you.

  • Part 1 β€” Rethinking Predictors β€” why causal reasoning matters before you reach for a regression.

  • Part 2 β€” Putting DAGs to the Test β€” what regression reveals (and hides) about wildfire drivers.

  • Explaining Markov Chain Monte Carlo β€” MCMC by way of wildfire forensics.

  • Reverse-Engineering Hamiltonian Monte Carlo β€” the engine behind modern Bayesian inference, built up from scratch.

πŸš€ How to Use This Book#

Read online#

Navigate the tutorials using the sidebar. Each tutorial renders with its saved outputs, so you can read the narrative, code, and figures without running anything.

Run it yourself in Colab#

Every notebook carries an Open in Colab badge. Click it to open the notebook in Google Colab and run the cells interactively. Note that some labs require credentials (for example, an Earth Engine project via the CYBERSE environment variable) β€” see the individual lab README for setup.

Preview locally#

pip install "jupyter-book<2.0"
jupyter-book build book
# then open book/_build/html/index.html

βž• Adding a Tutorial#

Adding a new notebook is nearly zero-effort. Drop the notebook into one of the lab folders, add a two-line entry to book/_toc.yml, and push to main β€” the GitHub Actions workflow rebuilds and republishes the book automatically.

Full instructions, including the bracket-quoting and URL-encoding gotchas for filenames with special characters, are in ADDING_A_NOTEBOOK.md.

πŸ™ Acknowledgments#

This book builds on the tutorials contributed by the BC Wildfire Predictive Service community, and reuses the Jupyter Book scaffolding from the Statistical Rethinking with PyMC book.


Questions or feedback? Open an issue on our GitHub repository