USGS Earthquake Hazards API (FDSN)

It is the global earthquake catalog opened up row by row: millions of cumulative event records maintained by the U.S. Geological Survey's National Earthquake Information Center and its contributing regional networks, reaching back into the early 1900s for larger magnitudes. One row per event, worldwide, selectable by bounding rectangle or radius circle around any point on Earth.

Each record carries the essentials - origin time in ISO8601 UTC, epicenter latitude and longitude, hypocenter depth in kilometers, magnitude with its scale (magType: mww, ml, mb among the values), a human-readable place string and the contributing network's event id - and then keeps going: three separate uncertainty measures, station geometry, a review status separating automatic solutions from reviewed ones, Did You Feel It? response counts, community and ShakeMap intensity figures, the PAGER alert level, a tsunami flag, a 0-1000+ significance score and an event-type field that distinguishes earthquakes from quarry blasts and explosions. Twenty-five documented fields per event, all definitions verified during Datadory's August 2026 research pass.

Get a sample of this dataset and we return rows shaped exactly like the dictionary below, cut to whatever region, magnitude window and period you name.

Two real events from the August 2026 research pass, exactly as they land:

# event 1 - magnitude 7.7, reviewed solution, yellow PAGER alert
time    : 2026-08-14T21:58:21.564Z
place   : 68 km NNW of Ende, Indonesia
coords  : lat -8.3101   long 121.3517
depth   : 10 km         mag : 7.7 (magType mww)
id      : us6000tkt2    net : us      status : reviewed
alert   : yellow

# event 2 - magnitude 7.4, reviewed solution, red PAGER alert
time    : 2026-08-10T12:34:28.125Z
place   : 5 km S of San Jose del Palmar, Colombia
coords  : lat 4.8436    long -76.2422
depth   : 110.285 km    mag : 7.4 (magType mww)
id      : us6000tjl2    net : us      status : reviewed
alert   : red

Read what the pair already says. Four days apart, two magnitude-7-plus events on opposite sides of the Pacific rim - one shallow at 10 km beneath Indonesian waters drawing a yellow alert, one deep at 110 km under Colombia rated serious enough for red. Both carry analyst-reviewed solutions on the us network with stable event ids you can cite. Multiply by a century of records and the table becomes the loading history of the planet, one typed row at a time. Request a sample and rows come back filtered to the geometry and magnitude band you name.

Fifteen core fields carry most analytical weight; definitions below were verified against live responses during Datadory's August 2026 research pass, not inferred from documentation. The remaining ten - ShakeMap and community intensity, the seven-field measurement-quality block and composite point geometry - fold out on request with your sample.

Geography - the whole planet. Events are filterable by bounding rectangle or by radius circle centered on any coordinates on Earth, so a parcel, a corridor or a hemisphere are all one selection away. Nothing in the catalog is aggregated away: the unit is the individual event, wherever it occurred, including offshore and deep-slab events most commercial hazard extracts quietly drop.

Temporal - the catalog reaches into the early 1900s for larger magnitudes, with completeness dependent on which regional network covered an area. Across all eras and catalogs the record accumulates millions of event rows - enough instrumented history that probabilistic screening is statistics rather than decoration.

Granularity - strictly one record per seismic event, with uncertainty and review status attached at the same grain. Sub-event phase-level detail sits in companion product data outside this record. That discipline is why joins hold: one event, one stable id, one row, forever.

To put ground conditions under the same footprint, pair this feed with Web Soil Survey (WSS); for the built environment on top of it, Chicago Building Permits and NYC DOB Job Application Filings name what humans actually built.

API, files, or your warehouse. Daily, weekly, or hourly.

Name the region, the magnitude window and the period when you request the sample; the sample ships first either way, with the full twenty-five-field dictionary attached so your engineers can validate shape before anything recurring starts.

Cadence is yours to set - tighten it during an active sequence, loosen it when your study is historical. Switching later is a settings conversation, not a re-integration project.

• Seismic site and corridor screening - draw a radius around a parcel, bridge crossing or pipeline segment and read the event history inside it before committing to a foundation design. See worked patterns on our data scientists use cases page.
• Catastrophe-risk and exposure modeling - event frequency by magnitude band and region feeds insurer exposure screens and cat-bond analysis. More on our investors quants use cases page.
• Geotechnical baselining with honest bounds - depth, magnitude and per-event uncertainty together replace single-number folklore in design memos.
• Blasting and induced-event context - the event-type field separates quarry blasts and explosions from tectonic events, so a site's neighbors explain their own signatures.
• Hazard-aware product features - location-keyed lookups drop into property and resilience platforms; integration patterns sit on our developers builders use cases page.
• Citation-grade reporting - stable event ids let journalists and academics pin every claim to one specific row; see our journalists academics use cases page and the citation grade research use cases playbook.

Data scientists and ML engineers and journalists, academics and students rate top relevance - one trains hazard and risk models on a century of labeled events where uncertainty arrives as columns, the other cites id-stamped facts no news roundup can match. Investors and quant researchers read catastrophe exposure behind insurers and builders; developers and data-product builders pipe location-keyed lookups into property platforms; market researchers and consultants fold regional seismicity into feasibility work. Competitive intelligence teams use it selectively, reading seismicity around rival facilities when resilience becomes part of the pitch. Persona-by-persona workflows sit on our construction engineering data hub, and the ranked shortlist lives on best construction engineering datasets.

Nothing else in Datadory's construction engineering slice measures physical loads. Every other record profiles human activity - permits, filings, spending, safety paperwork - while this one profiles the planet. The natural pairing is Web Soil Survey (WSS), which supplies SSURGO map units with engineering suitability ratings: events over soils is the classic pre-design screen, and the two together cover ground motion and ground condition in one pass. Against the administrative registries, the split is loads versus ledgers: NYC DOB Job Application Filings and Chicago Building Permits tell you who built what and how long approval took, while OSHA Enforcement & Establishment Data tells you what happened to their people. The national aggregates - Value of Construction Put in Place (VIP), New Residential Construction (Building Permits, Housing Starts & Completions) and Eurostat Construction Statistics & Building Permits - size markets rather than measure ground. Head-to-head with the permit ledger: USGS Earthquake Hazards API (FDSN) vs Chicago Building Permits.

Four things worth knowing upfront.

First, completeness varies by era and network. Minimum magnitude completeness differs across the contributing regional networks, so confirm thresholds before relying on any pre-1970s stretch for engineering analysis - modern instrumented decades are uniformly strong, older ones are richer for large events than for small.

Second, magnitude scales differ by design. magType records which scale produced each figure - moment tensors (mww), local magnitudes (ml), body-wave readings (mb) - so segment or convert before averaging across decades; comparing raw numbers across scales is the classic mistake this field exists to prevent.

Third, consequence fields are conditional. ShakeMap intensity appears only where a map product exists, PAGER alerts stay empty when none was issued, and felt-response counts depend on reports being filed. An empty cell documents absence of that product, not absence of shaking.

Fourth, status is a quality label. The review-status field separates automatic solutions from analyst-reviewed ones, so conservative analyses can filter to reviewed events without touching the underlying physics. All fifteen core field definitions above carry verified confidence - the schema is not among the open questions.

Get a sample of this dataset and we route rows scoped to your region, magnitude window and period, with the complete twenty-five-field dictionary attached.