Our Methodology

Data Collection

Atlas United aggregates data exclusively from authoritative federal sources. We do not use proprietary models or third-party data vendors. All data is publicly auditable and traceable to its source agency.

Data is collected via official APIs and bulk download services maintained by NOAA, FEMA, USDA, and USGS. Update frequencies range from real-time (severe weather reports) to annual (wildfire hazard potential).

Spatial Matching

For property-level assessments, we use haversine distance calculations to match addresses to nearby weather events. Events within a configurable radius (typically 5-10 miles) are associated with the property based on reported coordinates.

Flood zone determination uses FEMA's National Flood Hazard Layer (NFHL) with point-in-polygon analysis. Soil data is matched using USDA SSURGO map unit boundaries.

Climate Projections

County-level climate projections use USGS CMIP6-LOCA2 downscaled data. This dataset includes 27 global climate models run under 3 SSP scenarios (SSP2-4.5, SSP3-7.0, SSP5-8.5).

We calculate ensemble medians across all models for each variable and time period. Projections are available for 2030, 2050, 2070, and 2100 relative to a 1991-2020 historical baseline.

Limitations

• Storm event locations are reported at varying precision levels; some events may be county-centroid approximations
• Historical data extends to 2002; events before this date are less consistently documented
• Climate projections represent model consensus, not predictions of specific outcomes
• Soil and elevation data reflects most recent survey; local conditions may have changed

Quality Assurance

All reports include source citations with timestamps. Data anomalies are flagged during processing. Reports that cannot be generated due to data gaps are rejected rather than filled with estimates.