Data Methodology
How we source, validate, and maintain the data behind every document SkyCanvass generates. We believe transparency about our data sources and limitations builds more trust than marketing claims.
Data Sources
Building Codes
IRC, IBC, IECC, IEBC โ verbatim code section text with full section references
State-specific amendments, adoption dates, and edition overrides (e.g., Florida Building Code)
Local amendments and special requirements beyond state adoption
Storm Events
Official U.S. severe weather event records including hail, wind, tornado, and flood events
Dual-polarization Doppler radar data from 160 WSR-88D stations for reflectivity and MESH analysis
Severe weather reports, storm surveys, and spotter reports for ground truth verification
NOAA gridded hail analysis combining radar with surface observations
NCEI pre-computed MESH detections from the HDA algorithm
Geographic & Zones
ZIP-to-county-to-municipality resolution for jurisdiction hierarchy
Flood zone designations (AE, VE, X) via WMS tiles
Wildland-Urban Interface fire zone boundaries
High-Velocity Hurricane Zone (HVHZ) boundaries
Validation Pipeline
Source Ingestion
Raw data imported from official sources with full provenance tracking. Every record links back to its origin.
Schema Validation
Imported data validated against strict schemas. Type mismatches, missing fields, and out-of-range values are flagged.
Cross-Reference
Storm events are checked against available radar, field-report, warning, and wind-observation sources. Building codes are cross-referenced against ICC published editions.
Human Review
Flagged anomalies reviewed manually. State-specific amendments verified against official gazette publications.
Versioned Output
All generated documents include data version timestamps. Reports are reproducible from the same data snapshot.
Known Limitations
No data platform is perfect. Here is what our data can and cannot tell you.
Building codes are reference material, not legal advice
SkyCanvass displays verbatim code text with section references. Local amendments, administrative interpretations, and variance rulings may modify how codes apply to a specific project. Always verify with your local Authority Having Jurisdiction (AHJ).
Storm reports document weather events, not property damage
Storm data proves that severe weather was reported near a location. It does not prove that specific damage occurred at a specific address. Physical inspection is required to establish damage causation.
NOAA data has inherent latency
NOAA Storm Events Database entries are typically published 60-90 days after an event. SkyCanvass imports data within 48 hours of NOAA publication, but the underlying events may have occurred months prior.
Radar-derived hail sizes are estimates
MESH (Maximum Estimated Size of Hail) values are algorithmically derived from radar reflectivity data. They represent estimates of potential hail size, not ground-truth measurements. Ground truth comes from spotter reports and ASOS stations.
Municipal code coverage varies
State-level codes have comprehensive coverage. Municipal amendments are tracked for high-volume markets (FL, TX, CO, VA, MD) and expanded quarterly. Some smaller municipalities may not have local amendments captured yet.
Current Coverage
Questions about our data or methodology? We are happy to discuss.