Forest Fire Modeling using Cellular Automata and Dynamic Agents: Comparing Modular and Non-Modular Representations

Cellular automata (CA) have been widely used to model fire behavior by discretizing space and providing a localized description of fire dynamics within each CA cell. Both static and dynamic CA models have been employed to simulate fire propagation. The dynamic CA approach, in particular, allows for the representation of large-scale systems by only simulating active cells, thereby reducing computational complexity. In this paper, we use the π HYFLOW formalism to model forest fires through the application of cellular automata. In this paper, we employ the π HYFLOW formalism to model forest fires using CA. This formalism supports the process-interaction worldview, allowing for the definition of dynamic processes or agents that can be created and destroyed during runtime. At the network level, π HYFLOW facilitates time-varying model topologies, enabling the dynamic modification of model composition and coupling. We present both modular and non-modular representations for dynamic-structure CA. Simulation results demonstrate that the number of active fire agents is only a small fraction of the total number of CA cells, showing the efficiency of our method in simulating large-scale forest fire scenarios. The results also reveal that the non-modular models achieve a speedup of 7.12 compared to the modular models in a CA grid of 1200 × 1200 cells, simulating a fire spread over a uniform terrain under non-wind conditions. This highlights the effectiveness of the non-modular representation for simulating fire scenarios across large areas.