When the Grizzly Creek Fire torched more than 30,000 acres along Interstate 70 in Colorado's Glenwood Canyon in August 2020, emergency officials knew the danger wasn't over once the flames died down. The following summer, fast-moving debris flows—torrents of mud, rocks, and water—swept across the highway, closing it for weeks and costing the national economy roughly $1 million every hour the road sat empty. Two years on, officials braced for more disruptions, shuttering I-70 several times as light rains approached. But the debris flows never came. The reason? Trees and plants had begun to grow back, restabilizing the hillsides. It was a close call that illustrated exactly what a team of scientists at the U.S. Geological Survey set out to solve.
Andrew Graber, a landslide hazards research geologist at USGS, explained the core problem: when debris-flow warnings go out too often for threats that never materialize, people stop taking them seriously. "The last thing we want is for vital safety warnings to be seen as 'crying wolf' and ignored by people," Graber said. "So we set out to discover what storm intensity is actually needed to trigger debris flows in postfire landscapes where the trees and plants have begun to recover. And we have made great progress towards that answer."
The challenge is that wildfires fundamentally alter steep landscapes in ways that persist long after the fire itself. Burned soil loses its ability to absorb water, and the absence of vegetation means there's nothing to take up rainfall or anchor the earth. That's why debris flows often strike months or years later, catching communities off guard. USGS scientists routinely provide hazard maps and rainfall threshold guidance to agencies like the Colorado Department of Transportation, helping officials decide whether to issue warnings or close highways before storms arrive.
But before Graber's team updated their models, those predictions captured the worst-case scenario immediately after a fire—with no accounting for recovery. The researchers went back over three years of rainfall data across 12 burned areas, incorporated satellite imagery to measure vegetation regrowth, and updated their equations accordingly. The new method appears to reduce unnecessary warnings without increasing the number of debris flows that slip through undetected.
"Once we compared our updated postfire forecasts to real debris-flow data, we were glad to see how much better they reflected what actually happened after the fire," Graber said. The findings, published in the journal Geosphere, represent a meaningful step forward as more people settle in fire-prone terrain across the American West, and as climate change drives more intense storms atop landscapes still recovering from burns. Graber is careful to note the system isn't perfect yet, but the direction is clear: warnings are becoming smarter, and communities in the path of future fires may finally have forecasts they can trust.