Skip to content

How it works

From last night’s lightning to this morning’s search plan.

AfterStrike runs once a storm has passed. It scores every place lightning struck and issues a ranked watch list by 9am, with the reasons for every flag, so the first flight goes where a fire is most likely to be hiding.

01 · The pipeline

Nine things the model reads for every struck cell.

  1. 01

    Lightning

    Strokes from the lightning network, summed per 50 km cell for the storm day, the day before and the last three days. Stroke power where the network reports it.

    WWLLN (via WGLC) today; GPATS or Weatherzone strike data in production

  2. 02

    The storm's own rain

    Rain from 9am the day before to 9am the morning after, in the cell and its wettest neighbour. Dry storms start fires; wet storms hide them for days.

    SILO daily grids (Queensland Government) or NASA POWER

  3. 03

    Fuel dryness before the storm

    Keetch–Byram drought index, Griffith drought factor, days since rain and 7/30/90-day rainfall, all measured two days before the lightning.

    Derived from daily rain and temperature

  4. 04

    Fire weather

    Max temperature, humidity and vapour pressure deficit at the hottest hour, wind, and McArthur forest fire danger.

    SILO / NASA POWER

  5. 05

    The landscape

    Vegetation type, terrain ruggedness, how often the cell burns, years since it last burned, and how many lightning fires it has had before.

    MODIS land cover, AWS Terrain Tiles, NASA FIRMS, state fire records

  6. 06

    Fuel and ground

    Live fuel moisture and flammability, soil moisture, wind, slope and which way the land faces, and how green the vegetation was last month.

    ANU AFMS, NASA POWER, AWS Terrain Tiles, MODIS

  7. 07

    The bigger picture

    How close people are (so how fast a fire is likely to be noticed), and the climate drivers: El Niño, the Indian Ocean Dipole and the Southern Annular Mode.

    NASA SEDAC population, VIIRS night lights, NOAA

  8. 08

    The storm around it

    Lightning, storm rain and fuel dryness across the surrounding 250 km, because a single 50 km cell doesn't see the whole storm.

    Derived from the same feeds

  9. 09

    What satellites saw overnight

    New fire hotspots seen between the storm and 9am, from the overnight satellite passes. A fire lit in the afternoon can already be showing by morning, and the model weighs it against burn-offs and old fires.

    NASA FIRMS (VIIRS and MODIS, near real time)

02 · The morning watch list

Ranked, explained, and in the duty officer’s inbox by nine.

Each flagged cell comes with its three strongest reasons in plain words, such as “drought factor 10 · ~25 strokes · 1.5 mm storm rain”, so the call can be checked against local knowledge, not taken on faith.

The same scores feed a map layer, an SMS digest and an API for aircraft and drone tasking.

03 · Data sources

Built on open data today, live strike data to run it in real time.

Open today

  • WWLLN Global Lightning Climatology (daily, 0.5°)
  • SILO daily climate grids, Queensland Government
  • NASA POWER global daily weather
  • NASA FIRMS active fire archive
  • Fire history with cause: NSW (RFS/NPWS), Tasmania (theLIST), Victoria (DEECA), Western Australia (DFES/DBCA, via Geoscience Australia)
  • MODIS land cover (NASA GIBS) and AWS Terrain Tiles elevation
  • US Fire Program Analysis fire-occurrence database; Canadian National Fire Database

With partners

  • Strike-level lightning (GPATS, Weatherzone, Vaisala)
  • State ignition and cause records
  • Fuel type and fuel load maps
  • Live fuel moisture (ANU AFMS)
  • Radar rainfall at each strike

04 · Coverage

Where it’s proven, and where it’s next.

See the evidence