Lightning prediction · Aviation safety engineering

Predicting lightning before it forms.

AirLume models the atmospheric physics behind lightning formation to forecast strike risk along a flight route, before crews would otherwise see it coming.

$2B
Spent annually on mandatory post-strike inspections
88%
Detection rate on 181 real NASA SWDI strikes
63%
Of strikes occur in conditions crews read as safe
88.2ms
Average processing time per 8-waypoint route
CYOW CYYZ peak risk · WP6
Sample route output - Ottawa to Toronto, FL300. Waypoint colour reflects modelled strike probability.

The problem

01
Lightning management in aviation today is reactive, broad, and expensive.

Detection happens after the strike

Aircraft are struck by lightning one to two times a year on average. Current systems identify a strike only once it has happened, triggering a mandatory inspection regardless of damage.

Avoidance is imprecise

Without route-specific risk data, crews avoid entire storm systems with detours of 100 or more miles, adding fuel burn and delay for risk that may be concentrated in a small stretch of the route.

A majority of strikes happen in clear air

63% of recorded strikes occur in conditions that flight crews did not associate with adverse weather, sometimes 25 miles from the nearest radar return. Visual and radar cues alone are not sufficient.

The approach

02

AirLume doesn't forecast weather, it forecasts lightning formation, using the same atmospheric data already available to flight planning, run through plasma physics models rather than pattern-matching on cloud cover.

Paschen's Law

Breakdown voltage

Calculates the voltage at which atmospheric breakdown occurs from pressure and gap distance at a given altitude.

Townsend Avalanche

Ionisation cascade

Models how free electrons multiply in an electric field to estimate charge buildup and discharge likelihood.

Electric field model

Pre-strike conditions

Computes field strength at cruise altitude from real-time atmospheric data, ahead of any visible discharge.

Multi-altitude profiling

FL200 / FL300 / FL400

Runs risk assessment at each cruise level so a route can be re-planned by altitude, not only by heading.

A route's risk is computed waypoint by waypoint, which is what makes a 10–30 mile correction possible in place of a 100-mile detour around an entire storm system.

Validation

03
Tested against the NASA Severe Weather Data Inventory and against published Lightning Launch Commit Criteria - no simulated data.
ComponentResultTargetStatus
C physics engine 88% on 181 historical strikes 80%+ Met
Real-time performance 88.2 ms average per route <100 ms Met
Ada safety monitor 100% alert classification, 20/20 DO-178C dissimilar redundancy Met
ML enhancement layer 55% recall, 5-fold cross-validation 80% Below target

The ML layer's shortfall is attributed to a 100-sample training set, too small for a Random Forest classifier to generalise. The physics engine and Ada safety monitor, which carry the core prediction and certification logic, both met their targets independently.

Architecture

04
Four languages, each chosen for what it's actually good at - the same logic behind safety-critical avionics stacks like the Boeing 787's.
Python

Weather integration

Pulls and validates atmospheric data from OpenWeatherMap on a 15–20 minute refresh cycle.

C

Physics engine

Runs Paschen's Law and Townsend Avalanche calculations across waypoints in under 100ms, roughly 60x faster than the equivalent in Python.

Ada

Safety monitor

Independently re-derives risk using a separate empirical model, flags any disagreement beyond 15%, and enforces ARINC 653 timing budgets, the dissimilar redundancy DO-178C requires.

Jakarta EE

Web interface

Orchestrates the pipeline, persists results, and renders the route as a colour-coded waypoint map.

Roadmap

05
Next 6 months
  • Expand validation to 500+ historical strikes
  • Replace file-based IPC with shared memory or sockets
  • First pilot partner engagement
  • Retrain ML layer on a larger labelled dataset
12–18 months
  • Transport Canada certification process
  • Live METAR / ARINC 618 data-link integration
  • First airline pilot program
  • ForeFlight / Garmin Pilot integration

Built, tested, and documented.

The full technical report covers design iterations, the physics model, and complete test results. Happy to walk through any part of it.