Loss & Historical Databases

The most comprehensive global natural and technological disaster database, over 22,000 events since 1900 with standardised data on deaths, affected populations, and economic losses across all perils and all countries. The primary benchmark dataset for historical disaster loss calibration. Used by cat modellers, researchers, and reinsurers globally.

The reinsurance industry's most widely cited catastrophe loss database tracking global natural catastrophe losses with insured and economic loss splits going back decades. Provides annual statistics, regional analyses, and event-specific data.

Swiss Re's annual sigma reports on natural and man-made catastrophe losses are the industry benchmark publication read by every reinsurance professional. Provides insured and economic loss totals by event, peril, and region for each calendar year the standard reference for annual loss benchmarking and trend analysis.Required reading for anyone working in the global cat reinsurance market.

Long-run interactive charts on natural disaster deaths, economic losses, and insured losses globally across all perils, sourced from EM-DAT and presented in accessible visual form. Best used for non-technical communication: when explaining the scale of a peril or the growing protection gap to an underwriter or client, OWID charts are more immediately accessible than raw EM-DAT tables. Also tracks the growing protection gap across perils and regions over decades.

A curated directory of disaster loss and social vulnerability databases maintained by the Natural Hazards Center at CU Boulder. Aggregates links to ComCat, EM-DAT, Dartmouth Flood Observatory, Global Terrorism Database, FEMA data, and dozens of others. The best starting point when you need to find the right dataset for a specific peril or geography and are not sure where to look a meta-database of databases.

Aon's annual catastrophe report one of the most widely read free publications in the global reinsurance market. Provides a comprehensive annual review of natural and man-made catastrophe losses globally with insured and economic loss totals, event-by-event summaries, and regional trend analysis. Every senior underwriter and cat analyst reads this report annually. Free download, no registration required.

The definitive historical hurricane database for the Atlantic and Eastern Pacific basins; 6-hourly position, maximum sustained wind speed, and central pressure for every named storm going back to 1851. The standard trigger data source for Caribbean and US parametric hurricane products and the primary calibration dataset for Atlantic and Gulf of Mexico stochastic hurricane catalogues. Use HURDAT2 for Atlantic and Eastern Pacific it is more carefully quality-controlled than IBTrACS for these basins.

The global tropical cyclone archive covering all ocean basins Atlantic, Eastern and Western Pacific, Indian Ocean, Southern Hemisphere from multiple national meteorological agencies merged into a single standardised dataset. Use IBTrACS when HURDAT2 does not cover your basin Western Pacific typhoons, Indian Ocean cyclones, Australian cyclones. Note that data quality varies by basin and historical period; the Atlantic and Eastern Pacific data in IBTrACS is sourced from HURDAT2 and is equivalent quality, but other basins have shorter records and greater uncertainty in older data.

Global active flood archive tracking flood events worldwide since 1985 event dates, affected countries, severity classification, reported fatalities, and displacement data. Used for historical flood frequency research and for benchmarking modelled flood event sets against observed flood occurrence in global markets where commercial cat model coverage is limited. Freely downloadable dataset with GIS-compatible files.

Downloadable earthquake catalogue covering global seismicity the same dataset that forms the backbone of earthquake stochastic catalogue construction. Download event lists by magnitude threshold, date range, and geographic bounding box in multiple formats. Essential for anyone building or validating earthquake hazard models from historical seismicity data provides the raw event record from which Gutenberg-Richter relationships are calibrated.

Global database tracking political violence and protest events in near-real time battles, explosions, riots, and protests with precise location, date, and fatality data. Updated weekly. Used for country risk assessment in political violence underwriting and for SRCC accumulation analysis. The interactive map allows filtering by event type, country, and date range. The primary data source for operationalising the country risk scoring frameworks discussed in the Man-Made Perils module.

The most comprehensive open-source database of global terrorism events over 200,000 attacks since 1970 with data on attack type, target, perpetrator group, casualties, and property damage. Used for terrorism cat model calibration and for threat assessment in political violence underwriting. Provides geographic and time-series filtering and downloadable datasets. Free access with registration. The stationarity limitation discussed in the Terrorism lesson is directly relevant when using this data for model calibration.

The eXtreme Wind Storms (XWS) catalogue provides historical European windstorm track data, footprints, and modelled loss estimates for 50 of the most significant windstorms affecting Europe from 1979 to present. The HURDAT2 equivalent for the European windstorm market used for stochastic catalogue calibration and for benchmarking modelled losses against historical windstorm events. This is a research dataset, not a live tracking tool intended for historical analysis and model validation rather than real-time event monitoring.

European industry-wide insured loss data for natural catastrophes the most authoritative source for European windstorm, flood, and hail insured loss benchmarks. Expanding coverage to include Japan and New Zealand SCS from 2025. The standard reference for European modelled loss validation if your model produces a 1996 Lothar loss materially different from PERILS, that is a calibration conversation worth having.

Data-driven statistics on US insurance losses across all major cat perils hurricane, severe convective storm, earthquake, flood, wildfire, and winter storm. Provides historical insured loss series, market statistics, and publicly accessible fact sheets. Best used for US market context and non-technical communication when presenting modelled cat losses to a non-specialist audience, Triple-I historical loss data provides the market context that anchors the numbers. Broader scope than just hurricanes covers the full US cat peril suite.

Lloyd's publishes the Realistic Disaster Scenario (RDS) definitions and current threshold values for the major cat scenarios used for Lloyd's market aggregate monitoring — US Hurricane, US Earthquake, European Windstorm, Japan Earthquake, and others. The benchmark against which every Lloyd's syndicate measures its portfolio PML. Publicly available documents download directly from Lloyd's website. Essential reference for anyone working on Lloyd's syndicate cat management or exposure management frameworks.