Methodology
How Contrail Research evaluates flight and atmospheric context
The map is an investigative aid, not a remote chemical sensor or operational aviation forecast. It compares public flight, weather, satellite, radar, and observation records to display research hypotheses.
The research question
For a defined place and time, does the available record support atmospheric conditions in which an aircraft trail could form and persist as ice cloud? This is different from claiming that one aircraft caused a particular visible cloud. Attribution requires compatible timing, geometry, transport, and independent observation.
Input sources
| Input | Role | Key limitation |
|---|---|---|
| Public flight feeds | Aircraft position, altitude, speed, and direction where available. | Coverage, latency, filtering, and provider availability vary. |
| Numerical weather products | Upper-air temperature, humidity, cloud, and wind context. | Gridded values do not resolve every thin atmospheric layer. |
| NOAA satellite and radar | Independent cloud and precipitation context. | Imagery does not identify a source aircraft. |
| Documented sky observations | Time- and place-bound visual comparison. | Camera direction, visibility, timing, and selection bias limit attribution. |
Provider names describe provenance, not endorsement. Availability and terms can change.
Scientific basis
Contrail formation is commonly evaluated through the Schmidt–Appleman framework: aircraft exhaust mixes with sufficiently cold ambient air, water droplets form, and those droplets freeze. Cold air alone is not enough. Persistence depends on whether the surrounding upper air can sustain or grow ice crystals. Wind, mixing, sedimentation, and surrounding cloud conditions influence later appearance and spread.
Research workflow
Align records
Flight, weather, imagery, and observation records are compared by time and location.
Check atmospheric plausibility
The model evaluates whether the available atmospheric record supports formation or persistence.
Keep layers distinct
Modeled hypotheses remain visually separate from NOAA observations and public sky reports.
Compare alternatives
Natural cirrus, timing errors, incomplete flight coverage, and weather uncertainty remain part of the interpretation.
How to interpret the map
- Model indicator
- A relative research signal derived from the available inputs. It is not a measured probability.
- Modeled trail or cloud
- A visual hypothesis, not a satellite-detected feature or chemical measurement.
- Independent observation
- Satellite, radar, or documented field evidence used for comparison.
Validation
Evaluation requires positive and negative examples that were not used during model development. Useful reporting includes detection and false-alarm rates, performance across regions and seasons, and clear separation between model indicators and calibrated probabilities.
Known limitations
- Upper-air humidity and thin ice-supersaturated layers are difficult to resolve.
- Weather, flight, satellite, radar, and camera records have different spatial and temporal resolution.
- Public flight feeds may omit aircraft or contain delayed and incomplete positions.
- Tracks that overlap on a two-dimensional map may be separated by altitude or time.
- Aged contrails and natural cirrus can become visually difficult to distinguish.
- Public observations may overrepresent unusual skies.
Sources and further reading
- Schumann, U. (1996). On conditions for contrail formation from aircraft exhausts.
- Federal Aviation Administration: Contrails.
- Gierens, K. et al. (2022). Meteorological Conditions That Promote Persistent Contrails.
- NOAA GOES-R Advanced Baseline Imager band guides.
- OpenSky Network FAQ and coverage information.