Ignition Intelligence
01The problem
Lightning starts the fires nobody sees coming.
A dry storm can put tens of thousands of strikes into the landscape in an afternoon. Most do nothing. A few lodge in a dead stump or a peat bank and smoulder, out of sight, for days, until the wind gets up.
By the time there’s smoke, a crew is chasing a fire instead of putting out a spark. Today the call on where to look is made by experienced people with a rule of thumb: lots of lightning, little rain, dry fuel. AfterStrike makes that call with every storm, every cell and every fire on record behind it.
- >90%
- of the area burned in Black Summer was from fires started by lightning (Parliamentary Library).
- 1–5 days
- a strike can smoulder before anyone sees it. Wetter storms tend to cause the fires that smoulder longest before they’re found.
- ~1 in 20
- struck 50 km cells has a lightning fire in the next five days. The job is ranking them, not watching them all.
02How it works
From last night’s storm to this morning’s flight plan.
- 1
Detect
Every strike from the lightning network, placed on a 50 km analysis grid and summed over the last three days.
- 2
Score
Each struck cell is read against fuel dryness (KBDI, drought factor), the storm's own rain, heat, humidity, wind and the landscape's fire history, by two models: one weighs the cell's own readings, the other reads the storm map around it.
- 3
Send
By 9am: a ranked watch list for spotter planes, drones and towers, with the reasons for every flag in plain words.
Watch list · Tasmania
Issued 09:00, morning after the storm of 13 Jan 2016
- 142.75°S 146.25°E0.38
~7 strokes · 8.3 mm storm rain · drought factor 10
1 lightning fire followed
- 242.25°S 146.25°E0.23
~17 strokes · 5 mm storm rain · drought factor 9.7
1 lightning fire followed
- 342.25°S 145.75°E0.18
~23 strokes · 5.5 mm storm rain · drought factor 9.6
2 lightning fires followed
- 441.75°S 146.25°E0.16
~20 strokes · 3.3 mm storm rain · drought factor 9.7
1 lightning fire followed
- 541.75°S 145.75°E0.11
~17 strokes · 4.4 mm storm rain · drought factor 9.2
3 lightning fires followed
- 643.25°S 146.25°E0.10
~4 strokes · 4.4 mm storm rain · drought factor 8.5
1 lightning fire followed
| # | Cell | Risk | Why it was flagged | What happened |
|---|---|---|---|---|
| 1 | 42.75°S 146.25°E | 0.38 | ~7 strokes · 8.3 mm storm rain · drought factor 10 | 1 lightning fire |
| 2 | 42.25°S 146.25°E | 0.23 | ~17 strokes · 5 mm storm rain · drought factor 9.7 | 1 lightning fire |
| 3 | 42.25°S 145.75°E | 0.18 | ~23 strokes · 5.5 mm storm rain · drought factor 9.6 | 2 lightning fires |
| 4 | 41.75°S 146.25°E | 0.16 | ~20 strokes · 3.3 mm storm rain · drought factor 9.7 | 1 lightning fire |
| 5 | 41.75°S 145.75°E | 0.11 | ~17 strokes · 4.4 mm storm rain · drought factor 9.2 | 3 lightning fires |
| 6 | 43.25°S 146.25°E | 0.10 | ~4 strokes · 4.4 mm storm rain · drought factor 8.5 | 1 lightning fire |
03Proof
Tested on storms it had never seen.
We replayed every fire season we have records for, scoring each one with a model trained only on the seasons before it. Then we checked against 2,398 fires that agencies recorded as lightning-caused, each after lightning was detected in its cell: 1,426 in New South Wales (fire seasons 2015–16 to 2022–23), 819 in Western Australia (2015–16 to 2022–23) and 153 in Tasmania (2015–16 to 2024–25).
Flagging a tenth of struck cells catches 72% of Tasmania’s lightning fires, 68% of NSW’s and 68% of Western Australia’s. The rule of thumb catches 54%, 46% and 29%.
Tasmania · 153 lightning fires · fire seasons 2015–16 to 2024–25
View as table
| Method | Top 10% of cells | Top 20% of cells |
|---|---|---|
| Lightning density only | 29% | 54% |
| Dry-lightning rule (today) | 54% | 73% |
| AfterStrike | 72% | 89% |
New South Wales · 1,426 lightning fires · fire seasons 2015–16 to 2022–23
View as table
| Method | Top 10% of cells | Top 20% of cells |
|---|---|---|
| Lightning density only | 33% | 50% |
| Dry-lightning rule (today) | 46% | 67% |
| AfterStrike | 68% | 82% |
Western Australia · 819 lightning fires · fire seasons 2015–16 to 2022–23
View as table
| Method | Top 10% of cells | Top 20% of cells |
|---|---|---|
| Lightning density only | 22% | 40% |
| Dry-lightning rule (today) | 29% | 44% |
| AfterStrike | 68% | 79% |
04Physics, not places
It learns why fires start, not where they started before.
The model is never told a latitude or a longitude. It sees only what a fire behaviour analyst would: fuel, dryness, the storm and the landscape’s history. The same method, trained on each country’s own fire records, finds 62% of North American lightning fires in its top tenth of flags, and every flag can explain itself.
Fuel and vegetation
Wet forest, buttongrass and dry woodland ignite differently. Climate wetness stands in for fuel type until we add fuel maps.
Pre-storm dryness
Keetch–Byram drought index, the drought factor and fire danger, measured two days before the storm.
The storm itself
Strokes over three days, and how much rain fell with them. Dry storms start fires; wet ones hold them over.
The landscape's memory
Years since the cell last burned and how often it burns.
05Who it’s for
Built for the people who have to find it first.
- Fire and parks agenciesA ranked first-light search plan after every dry storm, with reasons a duty officer can defend.
- Plantation forestryKnow which compartments to check before a smouldering strike becomes a lost rotation.
- Power utilitiesLightning risk along easements and assets, ahead of fire-weather operating decisions.
- Insurers and reinsurersThe ignition piece of the wildfire model: which storms turn into fires, and where.
Data partnership
Your fire records make it sharper for your country.
The model already works where it has local records. Share your ignition and cause history, and we’ll tune AfterStrike to your landscape and run it beside your team through a fire season.