Earthquake hazard data by location: the ground's own ledger, delivered

The phrase covers two different products, and mixing them up is how site screens go wrong. Event catalogs record what actually happened: one row per earthquake, carrying origin time, epicenter, hypocenter depth and magnitude, selectable inside any circle or rectangle drawn anywhere on Earth. Hazard maps and ground-motion models are derived products - probabilistic statements about future shaking - and they live in separate analytical services that rarely reduce to rows you can join. An event catalog tells you what occurred where; it cannot tell you the annual exceedance probability at your footing line.

Datadory delivers the first product as typed rows: the global earthquake catalog maintained by the U.S. Geological Survey's National Earthquake Information Center and its contributing regional networks - millions of cumulative event records reaching into the early 1900s for larger magnitudes. Within Datadory's twenty-eight-record construction & engineering slice it is the only record that measures physical loads rather than human activity; everything else profiles permits, filings, spending or safety paperwork. Which makes the pairing obvious: Web Soil Survey (WSS) supplies the ground-condition half, and events-over-soils is the classic pre-design screen.

Twenty-five documented fields attach to each event, and the definitions were verified against live responses during Datadory's August 2026 research pass, not inferred from documentation. The essentials come first: origin time in ISO8601 UTC, epicenter latitude and longitude, hypocenter depth in kilometers, magnitude with the scale that produced it (magType distinguishes moment tensors mww, local magnitudes ml and body-wave readings mb), a human-readable place string, the contributing network's code and a stable event id, and a review status separating automatic solutions from analyst-reviewed ones.

Location-quality diagnostics ride along on the same row - stations used, largest azimuthal gap, distance to the nearest station, RMS travel-time residual, and horizontal, depth and magnitude uncertainties - so a screening model can weight a well-stationed reviewed event above a thin automatic one without extra enrichment. Impact layers attach where products exist:

The complete dictionary ships with your sample, including composite point geometry that packs longitude, latitude and depth onto one coordinate for direct GIS joins.

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 with ids you can cite in a report. Multiply by a century of records and the table becomes the loading history of the planet, one typed row at a time.

Seven decisions separate a defensible screen from a casual lookup:

1. Fix the area of interest. Start from the parcel or corridor coordinates you already hold; the same definition carries straight into the soil layer described below.
2. Choose the geometry. A radius circle suits point assets - a parcel, a bridge crossing, a plant footprint. A bounding rectangle suits linear corridors - pipelines, transmission lines, highways.
3. State the time window explicitly. An unstated window collapses to the most recent month, which quietly amputates the fifty-year history you meant to study. Say the dates.
4. Add engineering thresholds. A magnitude floor, a depth band and a reviewed-only filter remove the automatic tail and leave events a reviewer would sign.
5. Pick the shape that lands. Point geometry carrying longitude, latitude and depth joins directly into GIS; flat tabular rows suit models and spreadsheets.
6. Overlay the soil half. The identical area of interest resolves to SSURGO map units with engineering suitability ratings - dwellings with basements, septic tank absorption fields, local roads.
7. Let the volume be someone else's problem. Multi-decade pulls arrive paginated into clean windows with vintages stamped, so nothing rides on a single oversized request.

Name the region, the magnitude window and the period in a sample request and the extract arrives cut to all seven decisions.

A location screen rarely stops at faults, and three sibling records finish the picture inside the same industry slice.

Ground condition. Web Soil Survey (WSS) carries USDA NRCS's authoritative SSURGO database: roughly 100,000+ map unit polygons nationwide covering more than 95% of US counties, each keyed by mukey, muname and musym, with component-level and horizon-level attributes underneath and engineering suitability interpretations on top - ratings for dwellings with basements, septic tank absorption fields and local roads, resolved against your area of interest.

Together the four records cover loads, ground, approvals and money - the four questions any site decision quietly asks.

Four records, four different questions - and choosing among them decides what your screen can claim:

Only the first row measures the earth. The rest measure what humans did to it - and a screening memo that quotes one while implying the other is borrowing authority it has not earned. The trade-off between a hazard feed and a permit ledger is written up head-to-head in USGS Earthquake Hazards API (FDSN) vs Chicago Building Permits.

Four limits belong in any methodology note, and all four favor teams that know them in advance.

An event catalog is not a probabilistic hazard map. It tells you what occurred where; design-level ground motion and hazard curves are derived products living outside the record. Treat the feed as the loading-input starting point, not the deliverable.

Completeness varies by era and network. Minimum magnitude completeness depends on when each regional network came online, so confirm thresholds before letting any pre-1970s stretch carry analytical weight. Modern instrumented decades are uniformly strong; older ones are richer for large events than for small.

Magnitude scales differ by design. magType records whether a figure is a moment tensor, a local magnitude or a body-wave reading, so segment or convert before averaging across decades - comparing raw numbers across scales is the classic mistake the field exists to prevent.

Consequence fields are conditional - and so is the soil layer. ShakeMap intensity appears only where a map product exists, PAGER alerts stay empty when none was issued, and felt counts depend on reports being filed; an empty cell documents absence of a product, not absence of shaking. On the soil side, a map unit describes an association of soils across a polygon, not a measurement at your footing line. Borings still win.

Five personas get outsized value, and each works a different corner of the record.

Structural and geotechnical engineers draw the circle, read the magnitude range, depth distribution and largest recent event inside it, and frame design context before borings are budgeted. Catastrophe-risk analysts read event frequency by magnitude band and geography for exposure screens, accumulation studies and cat-bond work. Data scientists train hazard models on labeled event history where uncertainty arrives as columns rather than guesses. Developers wire location-keyed lookups into property dashboards and due-diligence tools without building a seismology pipeline. Journalists and academics cite id-stamped specifics - magnitude, depth, alert level - traceable to one row of one catalog.

The persona-by-persona workflow breakdown lives on data scientists use cases for construction engineering.

Because the hard part was never knowing the catalog exists - it is keeping a century of heterogeneous rows usable. Uncertainty measures arrive as columns, not footnotes; review status labels solution quality so conservative analyses can filter without touching the physics; magnitudes carry their scale beside the number; and every extract ships with its retrieval date stamped, because a catalog that accumulates reviewed solutions over time will not look identical next quarter.

Start with the construction engineering data hub, which indexes all twenty-eight records in the slice and shows where a seismic screen sits among them. Then go to the products themselves: USGS Earthquake Hazards API (FDSN) for the event core and Web Soil Survey (WSS) for the ground-condition half. The best construction & engineering datasets ranking scores the leaders side by side, the walkthrough of pulling SSURGO map units lives in our soil survey data guide, and the construction engineering data guide maps the wider pool - output statistics, permit records, safety enforcement and site intelligence.