Data Sources & Provenance
CivilSense is built on authoritative, publicly verifiable data sources. Every feed, every historical dataset, and every model parameter is traceable to its origin. This page documents what powers the platform, how often it updates, and why you can trust it.
Live Data Sources
Real-time and near-real-time feeds ingested by automated workers. Each source is independently monitored for uptime and freshness.
USGS Earthquake Hazards Program
Real-time, 2 minEarthquake events, magnitude, depth, location
Gold standard for seismological data. Global feed, US-filtered for display. Source of all earthquake consequence rings.
NOAA National Hurricane Center
Real-time during season, 30 minHurricane advisories, track forecasts, wind radii
Official US tropical cyclone authority. Advisories issued every 6 hours during active storms, supplemented by intermediate advisories.
NASA FIRMS VIIRS
Near real-time, 15 minActive fire hotspots, thermal anomalies
Satellite-based fire detection from VIIRS instrument on Suomi NPP. US-filtered. Complements NIFC official perimeters with early detection.
NIFC Active Fire Perimeters
Twice dailyOfficial wildfire perimeters, acreage, containment
Authoritative wildfire boundary data from the National Interagency Fire Center. Used for official consequence ring geometry.
NOAA AHPS River Gauges
Real-time, 10 minRiver stage, flood stage, gauge trends
NOAA Advanced Hydrologic Prediction Service. Gauge readings from thousands of US river stations with flood stage thresholds.
NOAA Storm Prediction Center
Continuous, 5 minSevere weather outlooks, tornado/hail watches
Official US severe weather authority. Convective outlook polygons and active watch geometries for tornado, hail, and wind threats.
NOAA Storm Surge (SLOSH Model)
Per advisoryStorm surge forecasts, inundation zones
NOAA Sea, Lake and Overland Surges from Hurricanes model. Rendered as a distinct surge overlay layer on hurricane events.
GDELT DOC 2.0
15 minOSINT news articles, media coverage
Global Database of Events, Language, and Tone. Aggregates worldwide news for disaster OSINT intelligence. Items filtered by credibility score before display.
Historical Data Sources
70 years of US-scoped disaster records used for frequency analysis, return period calculation, and model calibration.
| Dataset | Time Span | Records | Scope |
|---|---|---|---|
| FEMA Disaster Declarations | 1953 -- present | 68,000+ | US counties |
| NOAA Storm Events Database | 1950 -- present | 1.8M+ | US states and counties |
| IBTrACS Best Track Archive | 1842 -- present | 100,000+ track points | North Atlantic basin |
| NIFC Historical Fire Perimeters | 1984 -- present | 25,000+ | US federal and state lands |
| USGS Earthquake Catalog | Full history | 3M+ | CONUS + AK + HI + territories |
| FEMA NFIP Claims | 1978 -- present | Aggregate by census tract | US flood insurance program |
| Microsoft Building Footprints | Current release | 130M+ structures | United States |
Scoring Model Sources
Peer-reviewed publications and government technical manuals that calibrate our hazard scoring models. Every coefficient is traceable to these references.
USGS NSHM 2023
Seismic hazard parameters, Gutenberg-Richter a/b values, fault geometry
Petersen, M.D., et al. (2024). The 2023 US National Seismic Hazard Model. Bulletin of the Seismological Society of America.
FEMA Hazus v6.1
Vulnerability functions, depth-damage curves, construction type damage ratios
FEMA (2024). Hazus Earthquake/Hurricane/Flood Model Technical Manuals v6.1 / v5.1.
Kaplan & DeMaria (1995)
Hurricane wind decay after landfall
Kaplan, J. & DeMaria, M. (1995). A Simple Empirical Model for Reducing Tropical Cyclone Intensity After Landfall. Journal of Applied Meteorology.
Abatzoglou & Williams (2016)
Climate-adjusted wildfire frequency multiplier
Abatzoglou, J.T. & Williams, A.P. (2016). Impact of anthropogenic climate change on wildfire across western US forests. PNAS 113(42). DOI: 10.1073/pnas.1607171113
Brooks et al. (2003)
Tornado frequency climatology, EF2+ spatial distribution
Brooks, H.E., Doswell, C.A., & Kay, M.P. (2003). Climatological Estimates of Local Daily Tornado Probability for the United States. Weather and Forecasting 18.
Data Quality Commitments
Every ingested record stores: source, source_url, ingested_at, confidence_score.
Every model parameter references its source publication with source_description, validation_note, and effective_date.
Workers never crash the application. A failed source returns empty data gracefully -- no cascading failures.
3+ consecutive failures on any data source trigger automated Sentry alerts for immediate investigation.
Every hazard score exposes all sub-components in score_components. No black boxes.
Model parameter changes are logged with old_value, new_value, change_rationale, and supporting_paper_ids.
All data sources listed are publicly accessible or available under government open-data policies. CivilSense does not modify source data — raw records are stored alongside derived outputs for full auditability. For situational awareness only — not for emergency response.