Live inputs: UTC clock · OpenSky aircraft · Open-Meteo weather · AviationWeather METAR · FAA NAS attemptModel outputs: SFI · Traffic · Weather · Upper-Air · directional outlook
System Status
Checking…
Expect: Acquiring live data…
Horizon--Driver--Trajectory--
Traffic
--
Weather
--
Upper-Air
--
Sky Friction Index (SFI)
67/100
Watch↗ Rising
What This Means Now
Acquiring live inputs…
Primary Driver
--
Checking live inputs…
Systemic Fragility
--
Rises when 2+ inputs compound above 65 simultaneously
Directional Outlook · +1h / +3h / +6h
Now 67 → 70 / +1h
Velocity-extrapolated · not a calibrated forecast
--
Worsening
--
Stable
--
Improving
Delay Pressure High71/100
· diagnostic only · excluded from SFI
Operator Read
Acquiring live inputs…
Trajectory
Checking…
Pressure Threshold
--
Horizon to SFI ≥70
Data Status
Live Inputs
⚠ Heuristic Model
SFI weights are not statistically fitted — they are expert estimates. This is a pressure monitor, not a forecast. It does not predict accidents or specific delays.
Sky Friction Index Modeled3-input SFI
67/100
24h avg 61 · 7d baseline 55
7d24hNow
Watch↑ Rising
What it is: A single 0–100 score combining the three inputs below. Above 70 = capacity degrading. 55–70 = watch. Below 55 = normal.
Traffic Derived42% of SFI
73/100
Live aircraft density proxy · corridor model
7d24hNow
High→ Steady
What it is: Live ADS-B aircraft count normalised for time-of-day. High when corridors are packed relative to baseline capacity.
Weather Analyzed36% of SFI
58/100
Live weather sampling · key hubs
7d24hNow
Watch↑ Rising
What it is: Thunderstorms, wind, visibility and precipitation severity at 12 major hubs — blended from Open-Meteo model + live METARs.
Upper-Air Derived22% of SFI
62/100
250 hPa pressure-level wind · cruise-altitude signal
7d24hNow
Elevated↑
What it is: Wind speed at 250 hPa (~35,000 ft cruise altitude). Strong jet stream = longer routes, more fuel burn, cascading delays.
Delay Pressure DiagnosticNot in SFI
71/100
FAA NAS live attempt · fallback model
7d24hNow
High↑
Why shown: Downstream indicator — useful context but lags the SFI inputs. Not in the model.
Flights Tracked ObservedContext
14,823
ADS-B aircraft states · global proxy
7d24hNow
Live→
Why shown: Raw input to the Traffic score. High count → high Traffic score. Not a score itself.
Space Weather NOAADiagnostic
--Kp
NOAA planetary K-index · polar/HF route pressure
049
Normal→
Why shown: Kp ≥5 disrupts HF radio on polar/Atlantic routes. Not in SFI but can amplify operational pressure independently.
Traffic (42%) — Aircraft count from OpenSky ADS-B, normalised for time of day. High count in peak hours = higher pressure than the same count overnight.
Weather (36%) — Thunderstorm, wind, visibility severity at 12 hubs. Open-Meteo model blended with live METARs where reachable.
Upper-Air (22%) — 250 hPa wind speed at cruise altitude. Strong jet stream compresses routes and propagates delays.
Heuristic weights: The 42/36/22 split is expert-estimated, not statistically fitted to historical data. SFI measures operational pressure — it is not an accident predictor or delay guarantee. Stale feeds fall back to fixed baselines, never random values.
Directional Outlook (SFI)
3–6h
-- Now
-- +1H
-- +3H
-- +6H
Peak 70 → easing to 61
Illustrative extrapolation only — not a trained forecast.
SFI may peak near +1 hour as North Atlantic corridor density and JFK/LHR spacing pressure overlap. By +6 hours, the model eases as transatlantic wave compression normalizes, but Southeast Asia weather remains a watch item.
Directional extrapolation from live inputs. Not a trained forecast — direction only.
Trend Analysis
Builds after ~30 min
Tracks how SFI is moving — rate, acceleration, and which input is leading. All values are local session history only and reset on page reload.
Rate of Change
--
Collecting — updates every 2 min
Acceleration
--
Needs 7+ cycles
vs Session Baseline
--
Needs local history
Traffic Momentum (RSI)
--
Needs 14+ data points
Which input is leading the move
TrafficCollecting
WeatherCollecting
Upper-AirCollecting
Rate of Change = SFI points per minute. Acceleration = is the rate itself speeding up? RSI > 70 = momentum overbought, correction-prone. All based on this session only.
Historical Pressure Events Validation log
Records each time SFI exceeded 70 then recovered. Compare against known disruptions to validate model signal. Stored locally — resets if browser data is cleared.
▸ Model & Sources
Model & Sources
Purpose: GAPS is a situational-awareness prototype. It combines three independent inputs into one operational-pressure score: OpenSky aircraft density, Open-Meteo/AviationWeather hub weather, and Open-Meteo current-hour 250 hPa upper-air winds.
SFI = 42% Traffic + 36% Weather + 22% Upper-Air Weights are heuristic and not fitted coefficients. Delay pressure and wind/vertical stress are diagnostics, not independent SFI inputs.
Limits: This is not an aviation safety system, accident predictor, or trained forecast. It shows when traffic, weather, and upper-air conditions are compounding at the same time.
About GAPS
GAPS is a public tool. It costs nothing to use and has no affiliation with any airline, government agency, or aviation authority.
What GAPS Does
GAPS watches the global commercial airline network and asks one question: is the system under unusual stress right now, and is that stress building? It does not track individual flights. It watches the whole network — the way you would watch traffic on a highway system from a helicopter, not from inside one car.
What We Measure
Traffic — We count every aircraft broadcasting its position via ADS-B right now. On a normal day there are roughly 14,000–16,000 aircraft airborne globally. If that number is unusually high for the time of day, the corridors are congested and the system has less room to absorb problems.
Weather — We check live weather reports from 12 major hub airports: JFK, LAX, London Heathrow, Chicago O'Hare, Singapore, Dubai, and others. Thunderstorms, heavy rain, low visibility, and strong winds all reduce how many flights an airport can handle per hour.
Upper-Air — At cruise altitude — roughly 35,000 feet — aircraft ride or fight the jet stream. When it is unusually strong, it compresses flight paths, forces reroutes, and adds time and fuel burn. We measure this using wind data at 250 hPa pressure level.
The Sky Friction Index
We combine those three readings into one number: the Sky Friction Index, or SFI, running from 0 to 100. Below 55 is normal. 55–70 is watch territory — pressure is building. Above 70 means the network is losing its ability to absorb shocks. The formula is transparent: SFI = 42% corridor congestion + 36% weather pressure + 22% upper-air jet stream.
What Makes It More Than a Snapshot
Velocity — is the SFI rising or falling, and how fast?
Acceleration — is the rate of rise itself speeding up? This is the early warning signal. A system rising faster and faster is heading somewhere bad.
Systemic Fragility — when two or three inputs are high simultaneously, risk multiplies rather than adds. A crowded sky on a clear day is manageable. A crowded sky during a major storm with a powerful jet stream is a different situation entirely.
Emergency Squawk Monitoring — we watch for geographic clusters of Squawk 7700 emergency transponder codes from the live ADS-B feed.
SIGMET Alerts — official aviation hazard advisories for severe turbulence, volcanic ash, and major storms over key routes, pulled directly from AviationWeather.gov.
Space Weather — geomagnetic storms disrupt the radio communications that oceanic flights depend on. We monitor the NOAA planetary K-index every 15 minutes. A reading above 5 means North Atlantic and polar routes are under pressure you will not find in a weather report.
What GAPS Cannot Do
GAPS cannot predict accidents. It cannot tell you a specific flight will be delayed. It cannot see inside airline operations, crew scheduling, or maintenance. It works entirely with public data. What it can do is identify when physical conditions are compounding in ways that historically precede widespread disruption — before those events appear in the news.
The Validation Log
Every time SFI exceeds 70 and then recovers, the system automatically records the event — when it peaked, how high, what was driving it, and how long it lasted. Over time that log can be compared against real disruption events to assess whether the signals are genuinely predictive. The model earns its credibility over time.
Global Airport Voronoi Diagram
Each colored region on this globe shows the area of Earth that is closer to one specific airport than to any other. This is called a Voronoi partition — a way of dividing space based on nearest-neighbor distance. The dense, tiny cells over Europe and the U.S. East Coast reflect the highest concentrations of commercial airports in the world. The vast, sparse regions over oceans and remote land show where the nearest airport can be hundreds or thousands of miles away. It is, essentially, a map of how the global airline network carves up the planet.
