Fire Forest - Celllular Automaton

DECOURT Colin

This repository contain a school project simulating cellular automatons on a fire forest. This was done as part of the course Intelligence distribuée given by Laurent Simon at Bordeaux INP - ENSEIRB MATMECA.

Description

There is two Jupyter Notebooks.

First Jupyter Notebook

The second notebook called percolation-fire-forest.ipynb is a cellular automaton where a map with different density of trees is generated (from 0% of tree to 95% of tree). This notebook is divided in several sections :

• ▶️️ Utils functions : contain functions for getting valid cells' coordinates and the neighbour cells coordinates
• ▶️️ Parameters : parameters of the simulation and function for generating a forest with different density of trees (percolation threshold).
• ▶️️ Transition functions : contain state transition functions. Transition rules are :
  • An empty cell remain an empty cell
  • A ash cell stay ash
  • A fire cell become ash
  • A tree cell :
    • Become fire if one of them neighbour is in fire. The Moor neighbourhood was used.
    • Remain a tree otherwise
• ▶️️ A step function which one propagation step. Get the actual state and output a new state depending on transition rules.
• ▶️️ Main cell : initialization of the forest, launch fire forest with different grid sizes until there is no more burning tree.
• ▶️️ Data analyse : some plots to see impact of percolation threshold...

Tree type burning simulation

50x50 grid - 70% of trees

Impact of percolation threshold

Second Jupyter Notebook

The second notebook called tree-density-fire-forest.ipynb is a cellular automaton where a map with different tree type is generated. A tree will burn depending on the density (size for instance) of the tree. This notebook is divided in several sections :

• ▶️️ Utils functions : contain functions for getting valid cells' coordinates and the neighbour cells coordinates
• ▶️️ Parameters : parameters of the simulation and function for generating a forest with different tree densities (four tree types) and some clearing.
• ▶️️ Transition functions : contain state transition functions. Transition rules are :
  • An empty cell remain an empty cell
  • A ash cell stay ash
  • A fire cell become ash
  • A tree cell :
    • Become fire if one of them neighbour is in fire depending on the density (the type of the tree). The more the density is high the more the tree will burn (non intuitive I know...). The Moor neighbourhood was used.
    • Remain a tree otherwise
• ▶️️ A step function which one propagation step. Get the actual state and output a new state depending on transition rules.
• ▶️️ Main cell : initialization of the forest, launch fire forest with different grid sizes until there is no more burning tree.
• ▶️️ Data analyse : some plots to see impact of probability of burning...
• ▶️️ Pray for Australia section : this is the same simulation but adapted to Australia forest map. Each type of forest have a density and will burn depending on it. For this simulation different type of forest exist :
  • Woodland : d = 0.75
  • Herbland : d = 0.1
  • Open forest : d = 0.55
  • Closed forest : d = 0.35
  • Desert : can't burn
  • Ocean : can't burn

Tree type burning simulation

Some plots...

Australia burning simulation...

Some plots