How Delphi-1.0 Models Catastrophe Risk
Replacing coarse ZIP-code averages with sub-meter parcel physics, satellite moisture telemetry, and statutory California mitigation discounts.
Four Ground-Truth Data Layers
Every parcel score is computed from verified satellite and topographic measurements.
NASA VIIRS Satellite Telemetry
375-meter active thermal anomaly detection bands refreshed every 3 hours. Tracks active firefront proximity and calculates 25-year historical burn scar overlap.
ECMWF ERA5 Dead Fuel Moisture
Hourly atmospheric reanalysis modeling relative humidity and vapor pressure deficits into 100-hour and 1000-hour dead fuel moisture (FMC) curves.
USGS 3DEP 1-Meter LiDAR
High-density digital elevation models capturing slope angles, canyon wind corridors, and ridgeline exposures that accelerate wildfire spread up to 4x.
California CDI § 2644.9 Mitigations
Parcel-level structure hardening factors including Class A fire-rated roofing, enclosed eaves, and 100-foot defensible space brush clearances credited directly into rates.
How the Loss Calculation Works
Rather than opaque black-box machine learning, Delphi-1.0 uses a transparent compound Poisson-Tweedie distribution (power parameter p=1.5) calibrated across 26 years of historical loss data:
Read the Delphi-1.0 Actuarial Whitepaper
Includes full mathematical proofs, historical backtesting benchmarks (Tubbs, Camp, Glass, Dixie fires), ASOP 56 parameter sensitivity stress tests, and California Insurance Code statutory citations.
Methodology & Science FAQs
Inspect Any California Property
Test sub-meter LiDAR elevation, fuel moisture physics, and instant CDI § 2644.9 ratings in our live report engine.