Wildfire risk layers
Five layers describing relative wildfire risk across Wyoming, all from
the USDA Forest Service's Wildfire Risk to Communities (WRC)
program. They model how often a place is likely to burn, how intense
those fires would be, and the consequences for buildings — built on
large-scale simulations against the LANDFIRE fuels-and-vegetation map.
These are strategic, long-term indices, not seasonal
forecasts: they don't incorporate current weather or fuel-moisture
conditions and they aren't a substitute for locally-calibrated risk
assessments. The continuous layers (bp, cfl,
rps) live on a 30 m grid; the hazard-potential layers
(whp, whpcls) come from the same program at
270 m.
How WyldFire processes every layer
Each WRC raster arrives as a CONUS-wide GeoTIFF. WyldFire clips every
layer to the USGS Wyoming state boundary plus a 24-km buffer,
which provides edge tolerance for AOIs that straddle the state line.
Where it's storage-efficient, continuous values are scaled to compact
integers using a documented scaleFactor that the viewer
re-applies on display; raw integer values you read directly from the COG
must be multiplied by scaleFactor to recover the real-world
quantity. Each year of a layer is stored as a band of a single
multi-band Cloud Optimized GeoTIFF on S3, addressable via
/vsicurl/ HTTP range reads.
Layers at a glance
| Layer | What it measures | Native res | scaleFactor | Years |
|---|---|---|---|---|
bp |
Annual burn probability | 30 m | 0.0001 |
2014, 2020 |
cfl |
Conditional flame length (ft) | 30 m | 0.01 |
2014, 2020 |
whp / whpcls |
Wildfire hazard potential (continuous & classified) | 270 m | 1.0 |
2014, 2020 |
rps |
Risk to potential structures | 30 m | 0.0001 |
2014, 2020 |
Burn probability bp
What it is
"The annual probability of wildfire burning in a specific location."
A dimensionless probability between 0 and 1, expressing the likelihood that a wildfire burns at this 30 m pixel in any given year.
How upstream computes it
Burn probability comes from FSim, the Forest Service's large-fire simulation system. FSim repeatedly simulates tens of thousands of wildfire seasons against the LANDFIRE fuels-and-vegetation landscape, drawing weather from historical records and ignitions from observed patterns, then divides the number of simulated burns at each pixel by the number of simulated years.
How WyldFire processed it
Clipped to the Wyoming state boundary plus a 24-km buffer, scaled
from a dimensionless probability to a compact integer by
scaleFactor = 0.0001 (a stored 50 means
0.005, i.e. 0.5% annual probability), and re-encoded as
a multi-band COG hosted on S3 — one band per year (2014, 2020).
Storage vs display
Native dtype is a small unsigned integer; multiply raw pixel values
by 0.0001 to recover probability. The viewer applies a
class-breaks renderer for display (the bins are a viewer convention,
not part of the raster).
How to interpret
A BP of 0.005 at a pixel means roughly a 0.5% chance
that a wildfire burns through this 30 m location in any single year.
Over a 30-year horizon that compounds to roughly a 14% chance.
Compare values across an AOI to find the relatively hottest /
coldest pockets — BP is most meaningful as a relative ranking
within Wyoming.
Caveats
"The datasets presented here are the product of modeling, and as such carry an inherent degree of error and uncertainty. These datasets are intended to provide nationally-consistent information for the purpose of comparing relative wildfire risk among communities nationally or within a state or county. Data included here are not intended to replace locally-calibrated state, regional, or local risk assessments where they exist."
BP is a long-run statistical expectation, not a forecast for the upcoming fire season. It does not incorporate live weather, drought, or fuel-moisture conditions.
Year availability
2014 and 2020.
Provider & citation
USDA Forest Service, Wildfire Risk to Communities (RDS-2020-0016-2). Scott, Joe H.; Dillon, Gregory K.; et al. Wildfire Risk to Communities: Spatial datasets of landscape-wide wildfire risk components for the United States. Fort Collins, CO: Forest Service Research Data Archive. doi.org/10.2737/RDS-2020-0016-2
Conditional flame length cfl
What it is
"Conditional Flame Length: Most likely flame length at a given location if a fire occurs, based on all simulated fires; an average measure of wildfire intensity."
Units are feet.
How upstream computes it
CFL comes from FlamMap simulations run against the same LANDFIRE landscape FSim uses. For every pixel, FlamMap estimates the flame length under each modeled weather scenario; those flame lengths are then averaged across scenarios weighted by the conditional probability that the pixel actually burns under each scenario. The result is a single expected flame length per pixel, conditional on a fire occurring.
How WyldFire processed it
Clipped to Wyoming + 24-km buffer; scaled by
scaleFactor = 0.01 for compact integer storage (a
stored 475 means 4.75 ft of flame length); re-encoded
as a multi-band COG on S3.
Storage vs display
Native is integer; multiply by 0.01 to recover feet.
The viewer renders CFL with a continuous color stretch.
How to interpret
Flame length is a proxy for the difficulty of suppression. Roughly: under 4 ft, hand-crews can attack directly; 4–8 ft requires engines and dozers; above 8 ft, direct attack typically fails and only air resources and indirect strategies are effective. CFL says nothing about whether the pixel will burn — only what the fire would look like if it did.
Caveats
Subject to the same WRC use-limitations quoted in the BP section. Flame length is sensitive to the weather-scenario weighting and to the fuel model assigned by LANDFIRE; the value at a single pixel carries meaningful uncertainty.
Year availability
2014 and 2020.
Provider & citation
USDA FS WRC, RDS-2020-0016-2. doi.org/10.2737/RDS-2020-0016-2
Wildfire hazard potential whp & whpcls
What it is
"An index that quantifies the relative potential for high-intensity wildfire that may be difficult to manage, used as a measure to help prioritize where fuel treatments may be needed. This layer integrates wildfire likelihood and intensity, but does not include generalized susceptibility of homes."
How upstream computes it
WHP combines four ingredients into a single relative-hazard index: burn probability, conditional flame-length distribution, historic small-fire ignition density (a layer not modeled by FSim), and fuel-type control-resistance factors from Dillon et al. (2015). The control-resistance step is what distinguishes WHP from a simple likelihood × intensity product — it adds the difficulty of suppressing fire in a given fuel type to the calculation.
How WyldFire processed it
Both delivered surfaces — the continuous whp and the
classified whpcls — are clipped to Wyoming + 24-km
buffer and re-encoded as multi-band COGs on S3 with
scaleFactor = 1.0 (no scaling needed; values are
already small integers).
Storage vs display — whp vs whpcls
Same upstream program, two delivered surfaces. The
continuous whp is the raw 1–100 hazard index;
whpcls bins it into five ordinal hazard classes plus
two special non-burnable codes. In the WyldFire viewer
only whpcls is shown — the continuous
whp is included in the API catalog for power users
reading COGs directly, but is not exposed in the in-app map because
a 1–100 stretch reads as false precision; the classified version
communicates the same information more honestly.
The classified codes are:
| Code | Hazard class |
|---|---|
1 | Very Low |
2 | Low |
3 | Moderate |
4 | High |
5 | Very High |
6 | Non-burnable |
7 | Water |
How to interpret
WHP is a relative hazard ranking, not a probability. A High pixel adjacent to a Very Low pixel says something about fuel and control-resistance differences, not about which pixel is more likely to burn this year. Treat the classes as ordinal — the gaps between them are not equal.
Caveats
WHP does not include home vulnerability — it
characterizes the landscape, not the assets on it. For that, use
rps. Subject to the same WRC use-limitations as the
other hazard layers.
Year availability
2014 and 2020.
Provider & citation
Continuous whp: USDA FS WRC, RDS-2020-0016-2.
doi.org/10.2737/RDS-2020-0016-2
Classified whpcls: Dillon, Gregory K.; Gilbertson-Day,
Julie W. 2020. Wildfire Hazard Potential for the United States
(270-m), version 2020. 3rd edn. Fort Collins, CO: USDA Forest
Service Research Data Archive.
doi.org/10.2737/RDS-2015-0047-3
Risk to potential structures rps
What it is
"A measure that integrates wildfire likelihood and intensity with generalized consequences to a home on every pixel. For every place on the landscape, it poses the hypothetical question, 'What would be the relative risk to a house if one existed here?'"
Also referred to in earlier WRC publications as "Risk to Homes."
How upstream computes it
RPS applies a flame-length response function (Scott et al. 2013) to the per-scenario flame-length distribution that produces CFL. The response function converts flame length into a Net Value Change (NVC) estimate for a hypothetical home — capturing the rapid escalation in damage between low-intensity ground fire and high-intensity crown fire. NVC is then weighted by burn probability, giving an expected-loss index per pixel.
How WyldFire processed it
Clipped to Wyoming + 24-km buffer; scaled by
scaleFactor = 0.0001; multi-band COG on S3, one band
per year (2014, 2020).
Storage vs display
Multiply raw integer values by 0.0001 to recover the
index. Viewer renders with a class-breaks color ramp.
How to interpret
RPS is a hypothetical: every pixel carries an RPS
value answering "what if a house existed here?" — even a desert
pixel with no actual structures will have one. This is intentional;
it lets planners and insurers compare the wildfire-risk landscape
independent of where homes happen to be today. To weight RPS by
actual development, intersect with the housing-unit (hu)
or building-count (bc) layers.
Caveats
The "what if a house existed here" framing is easy to misread. RPS does not say "this house, right here, has risk X" — the home is hypothetical and uniformly described. Subject to the same WRC use-limitations quoted in the BP section.
Year availability
2014 and 2020.
Provider & citation
USDA FS WRC, RDS-2020-0016-2. doi.org/10.2737/RDS-2020-0016-2