Estimated reading time: 0 minutes
A quarter inch of rain in fifteen minutes. On a freshly burned slope, that can be enough to send mud, ash, and boulders racing downhill. It’s the kind of number behind a “burn scar flood” warning, and if your community was in or near a major wildfire in the past few years, this post is really important to you. If you’ve never heard of burn scar flooding, you might find it informative too.
If you live within or below a burn scar in the Western United States, there is a number that decides when the National Weather Service issues flood warnings. It’s rainfall intensity — or how hard the rain is falling over a fifteen-minute stretch.
When a storm is forecast to exceed that number, the National Weather Service issues a flash flood warning for your burn scar. This warning is based on research done by the U.S. Geological Survey (USGS).
Late 2026, USGS scientists published research showing that for much of the Western U.S., that number has probably been a little off. Not wildly wrong, but enough to change how frequently warnings go out.
Want the short version? The Climate Advisor’s Burn Scar Flood Risk Tool shows which recent burn scars sit above which towns, and gives an estimate of risk.
Why rainfall intensity
Post-fire debris flows do not behave like ordinary floods. Fire strips the vegetation that slows down the flow of water, and intense heat can dry out the soil, making it harder for the soil to absorb water. So fast moving water that doesn’t get absorbed… equals deadly debris flow.
The next hard rain is not slowed down by vegetation and top soil, so it picks up ash and gravel and eventually boulders, and arrives downstream as something closer to a wave of wet concrete than to water.
Crucially, this is triggered by short bursts of rainfall, not by slow steady rain. A gentle all-day soaking can drop far more water than a brief cloudburst, making dangerous runoff less likely.
The Old M1 Workhorse
The model that turns terrain, burn severity, and soil type into a likelihood of flooding is called the M1 model, originally created by Dennis Staley and his coauthors from the USGS in 2016. It has been the operational workhorse since 2017. It was calibrated on a debris-flow inventory from the chaparral canyons of southern California, and then applied to all of the West. In those canyons it takes 24 millimeters (0.95 inches) of water per hour, or about a quarter-inch of rain in 15 minutes to initiate a flood warning.
Obviously, there are different landforms all over the Western U.S. based on the underlying geology, different soil compositions, and different assemblages of plants in any given mountain range. (These are sometimes called “ecoregions”, see the table below.)
If you don’t live in Southern California, then your mountains probably behave differently.
It is tempting to read “the model over-predicted” as good news. More warnings, more caution, no harm done. However, that’s not how warnings should work. A threshold set too low produces warnings for storms that people start to ignore. The price of someone ignoring a warning when they should have paid attention can be life-threatening. These are important considerations in issuing flood warnings. These warnings have to be done just right, not too strict, not too loose. The updated model is a step in the right direction.
Updating the M1 Model (M*)
The USGS has known for a while that applying a southern California model to other mountain places was an approximation. In several regions — Arizona, northern California, Colorado, and New Mexico — M1 overpredicts how likely a debris flow is, and underpredicts how much rain it takes to start one. The fix is obvious…add more places to the model.
The team expanded the calibration inventory to 3,788 observations from 67 burned areas, paired each with short-duration rainfall data, and then fit separate coefficients for three EPA Ecoregions: Mediterranean California, the Upper Gila Mountains, and the Western Cordillera. The updated model (called “M*“) is 15 to 60 percent more accurate for these other Western regions.
The fact that the older M1 model was not super accurate has been known for a while, and discussed extensively in the scientific literature (the numbered green boxes in Figure 1 above!). So the new model is a welcome update. From my perspective, it seems like we should be able to create a hyperlocal model using satellite data of slope changes, geologic data, differences in vegetation and other factors.
EPA Level II ecoregions used in Graber et al. (2026)*
| Level II ecoregion | Where it is | Main features | Main plant communities |
| Mediterranean California | Coastal and southern California, from the Bay Area to northern Baja. Includes the Transverse and Peninsular Ranges, the Coast Ranges and the Sierra foothills. | Hot, dry summers and wet winters, with rain arriving in winter storms and atmospheric rivers. Steep, young, fast-uplifting mountains with weak, fractured rock. Santa Ana winds drive frequent fires that grow fast. This is where the original M1 model was calibrated. | Chaparral (chamise, manzanita, ceanothus, scrub oak), coastal sage scrub, oak woodland and savanna, and mixed conifer at higher elevations. |
| Western Cordillera | The high mountains of the West: the Rockies from Montana to northern New Mexico, plus the Cascades, Sierra Nevada, Klamath, Blue Mountains and Idaho Batholith. | Large, high, glacially carved terrain where most precipitation falls as snow in winter. Summer thunderstorms are common in the southern and interior ranges. Geology, relief and climate vary a lot within the region, which makes it the least uniform of the three. | Ponderosa pine, Douglas-fir, lodgepole pine, spruce–fir (Engelmann spruce, subalpine fir) and aspen, with alpine tundra above treeline. |
| Upper Gila Mountains | Central Arizona along the Mogollon Rim, and west-central to south-central New Mexico, including the Gila and Sacramento Mountains around Ruidoso. | Mid- to high-elevation plateaus and mountains. Precipitation comes in two seasons: winter storms, then summer monsoon thunderstorms that deliver short, intense bursts of rain, the storm type that triggers post-fire debris flows. | Pinyon–juniper woodland at lower elevations, ponderosa pine (the largest continuous stand in North America) with Gambel oak in the middle, and mixed conifer and spruce–fir at the top. |
* Source framework: U.S. EPA, Ecoregions of North America (Level II)
What’s coming…
A very strong El Niño is building for winter 2026–27, and that usually means a wetter southern half of the U.S., from California to Florida. El Niño loads the dice; it doesn’t roll them. The tilt is firmest for Southern California, Arizona, and New Mexico.
That puts the M* update right where the rain is headed: the paper flags Arizona and New Mexico as places where M1 overpredicted debris-flow likelihood, and the Upper Gila Mountains region around Ruidoso gets its own coefficients. Colorado is a shrug — its El Niño signal is mixed, could be wet or dry.
The 2026 fire season, by contrast, was concentrated in the Pacific Northwest and parts of the Great Basin, where El Niño should mean milder and drier weather. That’s a break for those scars this winter, not a pardon. Bare slopes stay primed for two to five years. The Great Basin and Colorado Plateau sit outside the three new regions, so the M* update doesn’t have any big changes for those regions.
What to do with this
Do not adjust your own behavior downward. If a warning is made for a burn scar near you — move to higher ground!
No-cost steps, this week
Turn on alerts. Enable Wireless Emergency Alerts on your phone and sign up for your county’s notification system. These floods arrive in minutes, and the warning is the whole defense.
Know one way out. Pick a route that does not cross a creek or a low-water crossing.
Look up your own scar. The USGS publishes basin-level hazard assessments for selected fires, showing which specific drainages are the dangerous ones. The danger zone is usually a narrow channel, not a whole valley, and knowing which side of the creek you are on is worth an afternoon of web searching to understand your own situation.
Postfire debris-flow hazard maps –> more user friendly. Click on the map graphic on this web page
Burn Severity Viewer –> a little more wonky, lots of data.
Ask your county emergency managers or floodplain administrator what they use. Local county administrators and emergency response personnel are the ones who translate these thresholds into local action. A polite question about which model version is behind your local warnings is a reasonable thing for a resident or a small business to ask. And they might be happy that someone asked, and then tell you everything you might need or would ever want to know.
Low-cost steps, this month
Clear culverts and roadside ditches on your property, especially if you are upslope of a scar. Most rural post-fire damage starts with a plugged culvert.
Bigger investments
Buy the flood policy before the storm. National Flood Insurance Program (NFIP) policies carry a thirty-day waiting period. A flood model revision changes nothing about that clock.
There is one narrow exception. If flooding caused or worsened by a burn scar on federal land damages your property, and you bought your policy within 60 days of the fire’s containment, FEMA can waive the 30-day wait. But that call is made case by case, after the loss, so treat it as a safety net, not a plan. Buy early, and if a scar sits above you, FEMA’s Flood After Fire page explains why the risk can last up to five years.
The science underneath climate adaptation is still being built, and occasionally it gets rebuilt in a way that says the old advice was slightly off. That’s not a reason to distrust it. It’s a reason to keep on reading and learning.
Is your town on the list?
We built a tool for exactly this moment: a running table of every wildfire burn scar over 10,000 acres since 2024, cross-referenced against the towns downstream from them, this winter’s precipitation outlook, and a risk rating for each one — low, medium, or high. It’s not a replacement for your county’s official hazard maps, but it’s a fast way to see whether you, a family member, or a business you rely on is sitting in the path of the next one. (Please read our disclaimer!)
—> Click here for the Burn Scar Flood Risk Tool <—
References Used
- Graber, A.P., et al, (2026), Regional models for postfire debris-flow likelihood and rainfall thresholds across the western United States, Earth Surface Processes and Landforms. doi:10.1002/esp.70393
- USGS Landslide Hazards Program, Hazard Assessment FAQ
- Staley, D.M., et al., (2016), Updated logistic regression equations for the calculation of post-fire debris-flow likelihood in the western United States, USGS Open-File Report 2016-1106.
- EL NIÑO/SOUTHERN OSCILLATION (ENSO) DIAGNOSTIC DISCUSSION, issued by CLIMATE PREDICTION CENTER/NCEP/NWS, 10 September 2026.