Fire risk: alpha, leverage, and the curve

The vocabulary FRED uses to describe wildfire risk and the procurement tradeoff that lets you weight it. This page covers everything you'll see in a fire-aware response.

#Burn probability

Every cluster in FRED has an associated annual burn probability, sourced from the USDA FSim 2024 wildfire simulator. The values range from approximately 0 (negligible risk) to about 0.08 (8 percent annual probability) for the most fire- prone clusters.

Burn probability is computed at 30-meter pixel resolution by FSim, then area-weighted to each cluster polygon by FRED. A cluster with a burn probability of 0.03 has, on average, a 3 percent chance of burning in any given year under current climate and fuel conditions.

#The four fire-risk tiers

For interpretive purposes, FRED groups clusters into four tiers based on burn probability percentiles:

  • Very Low: clusters in the bottom half of statewide burn probability.
  • Low: clusters in the 50th to 75th percentile.
  • Moderate: 75th to 90th percentile.
  • High: top 10 percent of statewide burn probability.

FRED's responses will sometimes break supply down by tier ("65 percent of your procurement comes from Very Low and Low tiers"). This is useful context for programs like BioRAM that target specific risk levels.

#Alpha: the weighting knob

Alpha is the single parameter that controls FRED's cost-versus-fire tradeoff. It ranges from 0 to 1:

  • alpha = 0: pure cost optimization. FRED minimizes $/BDT regardless of fire risk in the contributing clusters.
  • alpha = 0.05: modest fire weighting. The leverage knee for most California regions. Roughly 5.8 percent fire reduction per 1 percent cost premium.
  • alpha = 0.25: meaningful fire weighting. Notable cost premium, declining leverage. Suitable for BioRAM-style contracts with explicit risk targets.
  • alpha = 0.5: aggressive fire weighting. Substantially higher cost; smaller marginal fire reduction. Used for high-priority fire treatment programs.
  • alpha = 1.0: pure fire optimization. FRED ignores cost. Useful as a theoretical benchmark.

You can set alpha explicitly in a query ("at alpha = 0.1") or let FRED recommend the leverage-optimal alpha for your region. The default behavior when you say "prioritize fire risk" is to find the leverage knee, typically near alpha = 0.05.

#The leverage ratio

The leverage ratio at any alpha is defined as:

NOTE
leverage = (fire reduction percentage) / (cost premium percentage)

At alpha = 0.05, the typical leverage ratio across California is approximately 5.8x. This is the headline number from FRED's fire-aware analysis.

A 5.8x leverage means a 1 percent cost premium buys roughly 5.8 percent fire-risk reduction. Both numbers are measured relative to the cost-only optimum: cost premium is the percentage increase in LCOE versus the alpha = 0 baseline, and fire reduction is the percentage decrease in the procurement-weighted average burn probability.

Leverage declines monotonically with alpha. Past alpha ≈ 0.1, the ratio drops below 4x. Past alpha = 0.5, it's typically 1.5x or less. Past alpha = 1.0, leverage is below 1x, meaning each additional percent of cost is buying less than a percent of fire reduction. The economics are not favorable past the knee.

#Fire score

When FRED reports a "fire score" or "fire gain," this is the normalized fire-risk reduction relative to the cost-only baseline. A fire score of 1.19 means the procurement plan has 119 percent the fire-risk reduction of the cost-only optimum. The number is unitless and only meaningful as a comparison.

Most useful when comparing two alpha settings or two sites. If site A's fire score is 1.19 and site B's is 0.84, site A is doing substantially more fire-risk reduction for whatever cost premium it has incurred.

TIP
For policy-facing presentations, the cost premium / fire reduction pair is usually clearer than the leverage ratio alone. "+20% cost, +119% fire reduction" communicates the tradeoff more directly than "5.8x leverage."