Forest fire

Trees grow, lightning strikes, fire spreads to every touching tree. Most fires are tiny. A few burn half the forest. Can you stop them?

← all systems
–of the forest has trees now (one fire can cross it all above 59%)
–trees in the biggest fire lately
–of burned trees went in the big fires (over 1% of the forest each)
–trees in a typical (median) fire
tree fire ash firebreak

Every fire

one dot per fire: when it ended and how many trees it burned (log scale)something you did

How often each size

needs a few dozen fires…

Trees

share of the forest with trees59%: one fire can cross it all

Why big fires are rare but never impossible

The "How often each size" chart shows a power law (also called scale-free or heavy-tailed): fires 10× bigger are always about the same factor rarer, here roughly 12×. On a log-log chart that is a straight line. An exponential decline would instead be straight on a semi-log chart (tick the box) and would make giant fires practically impossible: with a typical fire of about 35 trees, a 20,000-tree fire would have a chance of about e−570. Here it happens every few thousand ticks.

Why: a fire burns exactly one connected patch of trees, so its size is the size of the patch the lightning hits. Slow growth and sudden fires turn the forest into a patchwork of thin and thick areas, always near the point where patches just start joining up. There, patches come in all sizes with no typical size, and lightning hits big patches more often because they have more trees. Too thick, and a huge fire clears it; too thin, and only small fires happen while it regrows. The forest keeps itself at this edge with nobody tuning it: self-organized criticality (Bak, Tang and Wiesenfeld 1987; Drossel and Schwabl 1992). The last points bend down because no fire can be bigger than the forest. Real forest fires (Malamud, Morein and Turcotte, 1998), earthquakes (the Gutenberg–Richter law), avalanches and blackouts follow power laws too.