Calamity

Calamity Research Dataset

Multi-hazard event data with cascade risk models — built for disaster science.

What's in the dataset

Calamity aggregates, normalizes, and enriches disaster events from 250 scientific data sources worldwide. Every event is processed through an 11-step pipeline that assigns a CalamityScore (0-100), calculates population exposure using a gaussian decay model, and detects cascade interactions using 20 probabilistic rules derived from peer-reviewed literature.

250
Data Sources
16
Disaster Types
20
Cascade Rules
20K+
Archived Events

Data Dictionary

FieldTypeDescription
idstringUnique event identifier
typeenumDisaster type (16 categories: earthquake, fire, flood, volcano, ...)
titlestringHuman-readable event title
severityenumlow | medium | high | critical
calamity_scorefloatComposite risk score 0–100 (intensity 40%, population 30%, cascade 20%, historical 10%)
latitude / longitudefloatWGS84 coordinates
event_timestampISO 8601Original source timestamp
reported_atISO 8601First detection by Calamity pipeline
sourcestringPrimary data source identifier
population_exposedintegerPEI — Population Exposure Index (gaussian decay model)
population_radius_kmfloatExposure radius in km
country_codeISO 3166Two-letter country code
nearest_citystringClosest populated place
cascade_dataJSONArray of cascade risk assessments with probability, type, radius, timeframe
type_metadataJSONHazard-specific fields (magnitude, VEI, wind speed, AQI, ...)

Export Formats

CSV

One row per event. Cascade data serialized as JSON column. Compatible with R, Python pandas, Excel.

JSON

Full event objects with nested cascade arrays. Ideal for programmatic analysis and web applications.

GeoJSON

FeatureCollection with Point geometries. Direct import into QGIS, ArcGIS, Mapbox, Kepler.gl.

API Access

GET /archive/bulk-export?format=csv&from=2025-01-01&to=2026-01-01&type=earthquake

# Filters: format, from, to, type, severity, min_score, include_cascades, limit
# Max 500K events per export
# API key required for authenticated tiers

Full API documentation available at /docs.

Risk Assessment Dataset — CFI Time Series

The Calamity Forecast Index (CFI) produces a risk assessment time series per country, hazard type, and day. Each record contains a 0-100 risk score, confidence level, and driver decomposition (cascade, return period, seasonal, trend, news velocity). The CFI estimates risk levels based on historical frequencies, seasonality, and current activity — it is not a prediction engine.

GET /api/v2/forecast

→ 95 countries × 16 hazard types × daily resolution
→ Fields: country_code, hazard_type, date, cfi (0-100),
  level (critical/high/elevated/moderate/low),
  drivers (text), components (cascade, return_period,
  seasonal, trend_anomaly, news_velocity)

# Citable as time series for multi-hazard risk assessment
# BibTeX available at /research#cite
95
Countries
16
Hazard Types
5
CFI Components
Daily
Resolution

How to Cite

If you use the Calamity dataset in your research, please cite it using the following BibTeX entry:

@dataset{calamity_dataset_2026,
  author    = {Leone Ventures},
  title     = {Calamity Multi-Hazard Event Dataset},
  year      = {2026},
  publisher = {Leone Ventures},
  url       = {https://calamity.live/research},
  note      = {197-source aggregated disaster dataset with cascade risk models}
}

Pricing

Academic
$29/month

Available with .edu email

  • Full dataset access (10K events/download)
  • CSV + JSON export
  • 5,000 API requests/day
  • 14-day free trial
Research
$79/month
  • Full dataset access (500K events/download)
  • CSV + JSON + GeoJSON export
  • 50,000 API requests/day
  • Cascade data included
  • DOI-citable dataset
  • 14-day free trial

Methodology

Our cascade detection rules are based on peer-reviewed literature including Gill & Malamud (2014), the UNDRR Sendai Framework, and OASIS CAP v1.2 standards. The CalamityScore formula, population exposure model, and deduplication engine are fully documented in our methodology page.