Log in Apply for a pilot →
PartnersAboutContactSupport One platform for individuals, businesses and governments
Home / Government & national / Hazard early warning
V4 · HAZARD EARLY WARNING

Hazard news that arrives a step earlier, with its source attached.

Earthquake, hurricane, wildfire and flood events are watched across sources. The agency edition distributes by region and grade to workspaces and APIs; the public edition stays free. Every warning keeps its source and time, and issuance always stays with the client agency.

An emergency duty office with a large text-free weather contour chart on the wall, a duty officer seen from behind at a desk with a telephone
Figure 1 · The warning duty room: many sources land on one chart before anything is issued.
4hazard classes, cross-checked across sources
3distribution channels, each its own
4steps, one region first
1rule: issuance stays with the agency

Three problems on the emergency desk

Slow hazard information slows everything after it.

Source patrol done by eye

Earthquake, weather, fire and water data each sit in their own system. Staff switch screens all night, and night is when items slip.

True & false in the same stream

Disaster rumours travel faster than official information, and nobody can check a source quickly before forwarding.

Distribution that disagrees with itself

Internal teams, partner units and the public receive different versions, and the after-action review cannot reconcile them.

Network make-up

Detection, grading and distribution are accepted separately.

Multi-source detection

Public feeds for four hazard classes connect continuously. An event is graded after more than one source agrees.

  • Public earthquake data feeds
  • Weather and hurricane data
  • Wildfire and flood water data
  • One event, several sources agreeing before grading

Grading & regional subscription

Subscriptions follow hazard type, region and grade. Agency workspaces, APIs and the public page each use their own channel.

  • Subscriptions by hazard, region and grade
  • An agency workspace channel
  • An API channel
  • A free public page channel

Issuance & receipt trail

Who issued, who received and when it landed are recorded, so reviews have a ledger to work from.

  • Issuer, recipients and delivery times on record
  • Distribution versions numbered
  • After-action reviews reconciled against the ledger

How the platform connects

Warning is the first working use of the situational core.

Fed by the situational core

Events, places and timelines come straight from the core. Rumour flags are shared with the intelligence products.

Free public layer

The public warning page is openly reachable. Trust comes first; the agency edition adds permissions and receipts on the same engine.

ONE operations foundation

Feed health, missed-event review and rule changes run as an operation, with anomalies surfaced before the client spots them.

Delivery route

Rules are fixed first, one region proves them, and peer agencies can then copy the same pattern. Every step leaves something you can inspect.

1

Sources & rules fixed

The feed list, grading thresholds, issuance flow and receipt rules are fixed jointly with the client.
Deliverables: the rules pack: feed registry, grading threshold table, issuance flow and receipt rules.

2

Regional pilot

Connect one region, four event classes and the agency workspace. Run it through a hazard season.
Deliverables: the pilot connection pack: workspace accounts, subscription settings and one hazard season of run records.

3

Review & calibration

Misses and false alarms from the pilot are reviewed, and every rule change is logged before it is fixed.
Deliverables: the review report: missed-event and false-alarm lists, and the rule-change record.

4

Regional expansion

Subscriptions expand by region and agency, with API and public capacity growing alongside.
Deliverables: the expansion table: per-region subscription terms, API documentation and public-layer capacity records.

  • The agency edition is supplied by seat and region, with API access by volume.
  • The public layer stays free, with no paywall.
  • Acceptance checks delivery time, missed-event review and source-labelling completeness.

Issuance & data limits

Warning credibility comes from restraint.

Issuance stays with the agency

The system detects and recommends distribution. Formal issuance runs through the client agency or its authorised channel.

Sources always labelled

Every event carries a source and a generation time. Unsourced items never enter distribution.

Rumours flagged apart

Suspected false information enters a screening queue and is never mixed with confirmed events.

Data handled by jurisdiction

Regional data and subscriber lists reside and retain under the client jurisdiction.

Questions, answered

Does this replace official warning issuance?

Never. Issuance stays with the client agency. The system detects, grades and recommends distribution; formal issuance runs through the agency or its authorised channel. It supplements the official system, it does not stand in for it.

How is speed achieved?

Sources connect automatically, so the manual patrol gap closes. Public feeds for four hazard classes flow in continuously, and an event moves into grading and distribution once sources agree. Acceptance checks delivery time item by item, not by assertion.

How are false alarms controlled?

Cross-check first, grade second, calibrate after a season. One event needs more than one source agreeing before it is graded, thresholds are fixed jointly with the client, and after each major event the misses and false alarms are reviewed with rule changes on record.

Is the public layer really free?

Yes. The public layer stays free, with no paywall. The public warning page is openly reachable; the agency edition adds permissions, receipts and APIs on the same engine, supplied by seat and region.

Can we take the API only, without the workspace?

Yes. The channels contract separately. Workspace, API and public page each run on their own terms, so the event stream can feed your own systems directly.

Can past events be replayed for verification?

Yes. Events, sources and distribution records are all kept. A review can replay an event's sources, its cross-checking and its delivery receipts, and test the grading decision against the record.

Where to next

TRUST

Trust & compliance

Deployment models, data sovereignty, security operations, supply-chain screening and use limits, in one place.

Read the compliance position →

INTEL

Intelligence network

The other side of event detection is the awareness graph. See how public sources are organised.

See the network →

CITY

Smart city systems

Warning is a high-frequency IOC use case. See it inside the city architecture.

See city systems →

Connect one region first

Send the target region and the current information sources. The reply covers pilot scope and the feed list.

Send a warning brief →