Skip to content

Generative Models

This page covers the simplified models used to test which mechanisms could plausibly generate observed wildfire scaling patterns.

Why generative models matter

The project does not need to simulate every physical detail of a fire to ask useful mechanistic questions. Minimal models are useful because they isolate candidate processes.

Candidate mechanism classes

  • diffusion-like spread
  • connectivity-driven spread through heterogeneous fuels
  • anisotropic wind-driven propagation
  • long-range spotting or jump processes

What these models are for

They allow the project to ask:

  • which mechanisms can produce similar exponents?
  • which mechanisms can generate fractal perimeter roughness?
  • which combinations of local rules produce the observed geometry?