Source of truth · Explainer
Three technologies are sold under the word “lightning”. One reports a strike after it has happened, one reads the electric field at a single mast, and one forecasts where the next strike is likely. Here is what each measures, how far ahead it can see, and where it belongs in a safety plan.
Detection network
AfterSensors triangulate a strike that has already happened. Output: a dot on the map, with a radius rule that decides when you stop.
Electrostatic field meter
DuringOne mast measures the electric field at one point. Output: a needle that rises as charge builds nearby, for any reason.
AI prediction · Flash
BeforeA model turns 30+ integrated data signals into a forecast for each 1×1 km cell, up to 60 minutes out. Output: a cell that changes colour before the first strike.
Definitions
After the strike
A national array of ground sensors listens for the electromagnetic pulse of a lightning return stroke and triangulates where it hit, usually within seconds. It is the reference record of what happened, and the kind of independent ground truth a forecast like Flash is scored against. A detection alert is precise about the past and silent about the next hour.
At one mast
A single mast measures the static electric field around one point and raises an alarm when the field climbs past a threshold. Because it reads charge rather than lightning, it can respond to a storm building overhead, and also to machinery, power lines and dust. It sees only its own surroundings, needs power and mounting at every site, and its warning time is whatever the local field allows.
Before the strike
Flash ingests 30+ integrated data signals and forecasts, for every 1×1 km cell, the likelihood of a cloud-to-ground strike in the next 60 minutes. It refreshes every 2 minutes and needs no on-site hardware. Its output is a forecast, so it is scored like one: hit or miss per cell on the one-hour window, against an independent ground truth, with the method published.
Side by side
| Criterion | |||
|---|---|---|---|
| What it measures | The electromagnetic pulse of a strike that has already occurred, located by triangulation across a national sensor network. | The static electric field at one mast, over the few hundred metres around it. | 30+ integrated data signals, scored per 1×1 km cell. |
| Lead time | None before the first strike. The alert follows the strike, within seconds. | Vendor-stated warnings of several minutes; validation methods vary by product. | Up to 60 minutes before a first strike, per 1×1 km cell, refreshed every 2 minutes. |
| Where it works | Anywhere the network covers; nothing at the site. | Only around the mast. Each site needs its own unit, power and connectivity. | Continental U.S.; any site you place on the map. |
| False-alarm behaviour | Very low for strikes that happened. Your radius rule decides how often you stop for strikes that never reach you. | Sensitive to local charge sources such as machinery and power lines: the basis of the “false alarm” caution in athletic-safety guidance. | Scored per cell on the one-hour window against an independent ground truth; the method is published and the validation dataset can be requested. |
| Cost & install | Subscription; sirens optional. | Hardware purchase or lease per site, plus mast, mounting and maintenance. | Software subscription, nothing to install; horn and strobe relays optional. See the pricing tiers. |
| Best for | Confirming a strike and running your policy's return-to-play clock. | Sites whose written policy names a field-meter alarm. | Clearing the field before the first strike, across one site or hundreds. |
Read this before you buy
Position statements from athletic-training and collegiate sports-medicine bodies caution against relying on a single-point electrostatic “lightning prediction” device as the sole trigger for clearing a field. The reasoning is sound: a local field reading can rise for reasons other than an approaching storm, and it cannot see a cell that is outside its sensing range. That caution is about an instrument class.
Flash is a different category. It is a forecast model, not a field meter: it scores every 1×1 km cell on the one-hour window against an independent record of cloud-to-ground strikes, and publishes how that score is computed, what it covers and what it does not.
We recommend Flash alongside your detection feed and your written policy, not instead of them. If your policy names a detection radius or a sideline device, keep it. Flash gives you the 60 minutes before that rule fires.
Use them together
Not one or the other. Each does one job well, and a written policy can name all three.
Before · The 60 minutes
Per 1×1 km cell, refreshed every 2 minutes, up to 60 minutes ahead. Clear the field on Advisory or Warning, whichever your policy sets.
On duty: Before the storm
Glossary ↓ 1×1 km cell, Advisory · Watch · Warning
During · Confirmation
Confirm each strike and start your return-to-play clock from the last one. Flash shows detected strikes on the same map.
On duty: During the storm
Glossary ↓ NLDN, Return-to-play
On the sideline · Policy
Where your state requires an on-site reading, take it with the device the policy names. Flash carries the heat-planning outlook for the day.
On duty: On the sideline, for policy
Glossary ↓ WBGT
= one complete safety plan
Skye, Flash Weather's AI weather agent, answers from the same cells and acts in your tools.
See SkyeGlossary
8 terms
FAQ
No system predicts the exact strike point. Flash forecasts the likelihood that a 1×1 km cell sees a cloud-to-ground strike in the next 60 minutes, and scores that forecast against an independent ground truth: up to 99.6% accuracy on the one-hour window.
No. A field meter reads the charge at one mast. Flash reads 30+ integrated data signals across the whole grid, refreshes every 2 minutes and publishes how its accuracy is scored.
Keep it. Detection confirms the strike and anchors your return-to-play clock; Flash gives you the time before it. Flash shows detected strikes on the same map so the two never disagree about what happened.
The accuracy method publishes the ground truth, the unit of measure and how each cell is scored on the one-hour window, so you can read exactly what up to 99.6% means, why it is “up to”, and how to request the validation dataset.
Book a demo · No sensors required for the forecast
We'll load your locations, replay last season's strikes against Flash's predictions, and show you the alert log your safety officer would have received.