Wildfire risk is often estimated with tools like the Canadian Forest Fire Weather Index (FWI), which uses weather conditions to predict how likely fires are to start and spread. While FWI is commonly used to assess fire danger today and in future climate scenarios, it has not been widely applied to climate model simulations of the past.

We explore how wildfire risk in the Western US has changed through time by calculating FWI using atmospheric data from climate model simulations run for the last 21,000 years. We find that as the climate warmed, FWI increased across the region and that changes in FWI align with shifts in weather patterns like temperature and rainfall during periods of rapid climate change.

To see how well our FWI calculations reflects actual wildfire activity in the past, we compare our results with charcoal from lakes in California, Oregon, and Washington – a common way to measure past wildfires. We find that FWI tends to match the charcoal data better than any individual climate variable, like temperature or precipitation, alone.

Applying FWI to past climate simulations helps inform how climate has shaped wildfire activity over thousands of years and may improve predictions for the future.