Build weather-aware operations alerts

Weather decides whether a crane lifts, a pilot boards a vessel, or a helicopter flies. This recipe assembles an operational alerting check for one fixed site from Sugra Earth: a point forecast for the work window, a lightning and multi-hazard sweep around the coordinates, and sea-state and aerodrome add-ons for marine and flight-dependent work. It ends with a go/no-go rule you can run on a schedule. The example site is a berth near the Port of Rotterdam (51.95 N, 4.14 E); substitute your own coordinates throughout.

What you build

A scheduled check (every 10 to 30 minutes) that calls four or five Sugra Earth endpoints for one site

Forecast inputs (hourly gusts, precipitation probability) plus hazard inputs (lightning, multi-hazard alerts) in a box around the site, with optional wave-state and METAR add-ons

A single go/no-go decision rule combining threshold checks on those fields

1. Pull the 48-hour site forecast

GET /api/v2/weather/forecast returns hourly and daily rows for any coordinate worldwide. Request the window you plan against; wind_speed_unit=ms keeps gusts in the unit most equipment limits use.

Response (truncated to the fields the rule reads):

{
  "data": {
    "latitude": 51.934, "longitude": 4.147, "timezone": "Europe/Amsterdam",
    "units": {"hourly": {"wind_gusts_10m": "m/s", "precipitation_probability": "%", "visibility": "m"}},
    "hourly": [{"time": "2026-07-19T12:00", "temperature_2m": 19.5, "precipitation_probability": 4,
                "wind_speed_10m": 6.4, "wind_gusts_10m": 11.4, "visibility": 31800.0, "cape": 20.0}],
    "daily": [{"date": "2026-07-19", "precipitation_probability_max": 75, "wind_gusts_10m_max": 11.5}]
  },
  "meta": {"endpoint": "/api/v2/weather/forecast", "source": "openmeteo", "cached": false}
}

Timestamps are in site-local time (timezone is resolved automatically). The midday row is calm, but precipitation_probability_max for the same day is 75 percent: scan every hourly row inside your work window rather than sampling one, and take the window maximum of wind_gusts_10m and precipitation_probability.

2. Check for lightning near the site

GET /api/v2/hazards/lightning serves flash detections from the EUMETSAT Meteosat Third Generation Lightning Imager, filtered by bounding box. Each response carries the newest complete 10-minute granule, so polling once per cycle is enough.

Response (truncated):

{
  "data": {
    "query": {"min_lat": 51.0, "max_lat": 53.0, "min_lon": 3.0, "max_lon": 5.5, "limit": 3},
    "total_matched": 0, "count": 0, "events": [],
    "upstream": {"coverage_start": "2026-07-19T08:50:02Z", "coverage_end": "2026-07-19T09:00:02Z", "flash_count": 3062}
  },
  "meta": {"endpoint": "/api/v2/hazards/lightning", "source": "eumetsat_mtg_li"}
}

total_matched: 0 is a clean negative, not an outage: the full-disc granule held 3,062 flashes in the same 10 minutes, none inside the box. upstream.coverage_start and coverage_end date the granule, so the rule can also verify freshness. Coverage is the Meteosat 0-degree field of view (Europe, Africa, the Atlantic and eastern South America); outside it, use the forecast's cape field as a convective proxy instead.

3. Sweep the multi-hazard overlay

GET /api/v2/hazards/alerts returns GDACS multi-hazard alerts (earthquake, tropical cyclone, flood, drought, volcano, wildfire) with a Green/Orange/Red level. Use a wider box here: regional events can affect site logistics.

Response (truncated):

{
  "data": {
    "total_matched": 1, "count": 1,
    "events": [{"event_id": "gdacs-WF-1029489-3", "event_type": "wildfire",
                "geometry": {"lat": 48.375, "lon": 2.5494}, "place": "Forest fires in France",
                "metrics": {"alert_level": "Orange", "country": "France", "is_current": false},
                "source": {"institution": "GDACS (European Commission JRC + UN OCHA/UNOSAT)", "tier": "I"}}]
  },
  "meta": {"endpoint": "/api/v2/hazards/alerts", "source": "gdacs"}
}

One Orange wildfire matched the box, but metrics.is_current is false, so it does not trigger. Gate on alert_level in Orange or Red combined with is_current: true. The overlay is automated and not human-reviewed; treat a hit as a signal to escalate to official channels, never as the official warning itself.

4. Add sea state for marine work

GET /api/v2/transport/marine samples the NOAA GFS-Wave global model at the nearest ocean grid node. Query a point in open water off the site; a node over land or inland water returns nulls with a coverage_note.

Response (truncated):

{
  "data": {
    "location": {"requested": {"lat": 52.0, "lon": 3.9}, "node": {"lat": 52.0, "lon": 4.0}},
    "valid_time": "2026-07-19T00:00:00Z",
    "values": {"wave_height": 1.39, "wave_period": 6.64, "wave_direction": 343.5},
    "units": {"wave_height": "m", "wave_period": "s", "wave_direction": "deg"},
    "provenance": {"institution": "noaa", "model": "gfs-wave", "run_kind": "analysis"}
  },
  "meta": {"endpoint": "/api/v2/transport/marine", "source": "noaa_gfs_wave"}
}

Significant wave height at the harbor approach is 1.39 m, inside a typical 2.5 m pilot-boarding limit. This is modelled guidance, not a measurement and not of navigational class; use it to schedule, not to navigate.

5. Add aerodrome observations for flight-dependent work

GET /api/v2/transport/aviation/metar returns decoded current aerodrome observations worldwide from the NOAA Aviation Weather Center. Query the ICAO ids nearest your site; Rotterdam is EHRD.

Response (truncated to the first station):

{
  "data": {
    "total_matched": 2, "count": 2,
    "stations": [{"station": "EHRD", "obs_time": "2026-07-19T08:55:00Z", "wind_dir_deg": 340,
                  "wind_speed_kt": 10, "visibility_mi": "6+", "flight_category": "VFR",
                  "raw": "METAR EHRD 190855Z AUTO 34010KT 270V040 9999 SCT034 19/10 Q1020 NOSIG"}]
  },
  "meta": {"endpoint": "/api/v2/transport/aviation/metar", "source": "awc_aviation"}
}

flight_category compresses visibility and ceiling into one gate word (VFR, MVFR, IFR, LIFR), and the raw report is preserved. Consultative only; it does not replace an official preflight briefing.

The go/no-go rule

Combine the verified fields into one deterministic rule, evaluated on every polling cycle:

NO-GO if any of:
  lightning total_matched > 0                    flash inside the box in the newest granule
  max(hourly wind_gusts_10m) >= 15 m/s           over the next work window
  max(hourly precipitation_probability) >= 60    for precipitation-sensitive tasks
  marine wave_height >= 2.5 m                    pilot boarding, small-craft transfer
  metar flight_category != VFR                   helicopter operations
ESCALATE TO A HUMAN if:
  any alerts event with alert_level in (Orange, Red) and is_current = true
otherwise GO, and re-evaluate on the next cycle.

The thresholds are illustrative; calibrate them to your equipment limits and your regulator's requirements. All of these feeds are consultative overlays with named sovereign and intergovernmental upstreams; wire the rule to escalate to official warning channels, not to replace them.

Going further

Sugra Earth direction - the full weather, hazards, energy and transport catalog, including forecast history and tides.

Rate limits and errors - size your polling interval against your plan's daily budget.

Quickstart - the response envelope, authentication and rate-limit headers used by every request above.

Build weather-aware operations alerts

What you build#

1. Pull the 48-hour site forecast#

2. Check for lightning near the site#

3. Sweep the multi-hazard overlay#

4. Add sea state for marine work#

5. Add aerodrome observations for flight-dependent work#

The go/no-go rule#

Going further#

What you build

1. Pull the 48-hour site forecast

2. Check for lightning near the site

3. Sweep the multi-hazard overlay

4. Add sea state for marine work

5. Add aerodrome observations for flight-dependent work

The go/no-go rule

Going further