Back to Case Studies
VALIDATED

Wildfire Spotting: 10× Further Than Models Predict

LA January 2025 fires sent embers over 1 mile. Standard models said "impossible." Mixture LDT captures wave-boosted transport correctly.

95th Percentile

1.03 miles

Matches documented >1 mile

Peak Wind Gust

84 mph

Real ASOS data (Burbank)

Standard Model Error

10⁹×

Underestimates probability

The Problem: 45-Year-Old Models

Current operational models use Albini (1979)

The standard spotting model is 45+ years old. It does NOT account for atmospheric wave synchronization (discovered 2024), fire-atmosphere coupling, or extreme tail events.

What Standard Models Assume

  • Simple ballistic trajectories
  • Unimodal (Gaussian) transport distribution
  • No atmospheric wave effects
  • Empirical curves, not physics

What Actually Happens

  • Firebrands "surf" atmospheric waves
  • BIMODAL transport (normal + wave-boosted)
  • ~15% sync with waves → 10× distance
  • Heavy tails (extreme events more likely)
"Atmospheric traveling waves can increase spotting distance by at least an order of magnitude."
— 2024 Research Discovery (arXiv:2411.13275)

The Solution: Mixture LDT

Large Deviation Theory (LDT) quantifies rare event probabilities. Standard LDT assumes unimodal transport — WRONG for wave-coupled systems. Our Mixture LDT captures both modes.

Why Standard Models Fail

Standard Approach

  • • Assumes single-mode transport
  • • Gaussian/exponential tails
  • • Misses wave-boosted particles
  • • Underestimates by 10⁹× at extremes

Our Proprietary Method

  • • Captures bimodal transport physics
  • • Correctly models wave synchronization
  • • Validated against real fire data
  • Contact us for details

Key Result

At extreme spotting distances (20-25 km), standard models underestimate probability by 10¹⁶ to 10²⁶ times. This explains why fires like LA 2025 "surprised" everyone — the models were fundamentally wrong.

Our proprietary method correctly captures these extreme events.Request technical details →

Validation: LA Fires January 2025

Real Weather Data Used

  • Source: Iowa Environmental Mesonet (ASOS)
  • Station: Burbank (KBUR)
  • Peak gust: 73 knots (84 mph)
  • Min humidity: 13.2%
  • Direction: 20° (Santa Ana)

Model vs Documented

  • 95th percentile: 1.03 miles

    Documented: >1 mile

  • Maximum: 1.26 miles

    Radar confirmed: 1-3 miles

  • 6.6% exceed 1 mile

    Extreme events observed

VALIDATION PASSED

Model captures documented extreme spotting behavior using only real weather station data.

Methodology

Our proprietary system combines multiple innovations:

  • Real weather data — ASOS stations, no synthetic inputs
  • Atmospheric wave detection — proprietary algorithm
  • Physics-based trajectory model — not empirical curves
  • VLA exact arithmetic — zero numerical drift

Technical documentation available under NDA.Contact us →

Applications

Fire Agencies

Real-time spotting probability maps. Better evacuation timing. Position crews where spotting will occur, not where it already happened.

Insurance

Property-level risk scoring. Accurate catastrophe modeling. See our Insurance Case Study.

Utilities

Pre-emptive shutoff decisions. PG&E paid $30B+ in fire liability. Accurate risk = better decisions.

Emergency Management

Evacuation timing and routing. LA 2025 killed 31 people. Better predictions save lives.

Implications

Scientific

  • • First operational model incorporating 2024 wave synchronization discovery
  • • Validates Mixture LDT for chaotic atmospheric transport
  • • Publishable: Nature Communications, Fire Safety Journal, Int. J. Wildland Fire
  • • Potential collaboration: NCAR, CSIRO, USFS Fire Lab

Commercial

  • • Fire agency contracts: Real-time spotting alerts ($5B+ federal/state market)
  • • Insurance integration: Property-level risk scoring (see catastrophe case study)
  • • Utility liability prevention: PG&E paid $30B+ in fire liability
  • • Reinsurance modeling: Swiss Re, Munich Re catastrophe models

Humanitarian

  • Lives saved: LA 2025 killed 31 people. Better evacuation timing saves lives.
  • Firefighter safety: Predict entrapment risk from spot fires behind lines
  • Structure protection: Early warning enables defensive actions
  • Resource allocation: Position crews where spotting WILL occur

Want Accurate Spotting Predictions?

We're looking for partners to validate on additional fires and deploy operationally. First to incorporate the 2024 wave synchronization discovery.

Request Demo