Contrail Forecasting: How Airlines Find the Air That Makes Ice

3 min read

Last updated: 16 September 2026. Every figure on this page is dated and linked to its source.

What exactly is a contrail forecast trying to predict?

A contrail forecast is a map of where aircraft exhaust will form ice crystals that persist for more than a few minutes. The physics is not exotic: jet engines emit water vapor and soot particles. At cruise altitudes (typically 30,000–42,000 feet), the air is often below −40°C. When the exhaust plume mixes with ambient air, the water vapor can condense and freeze onto soot particles. If the surrounding air is already saturated with respect to ice — meaning relative humidity over ice (RHi) exceeds 100% — the ice crystals do not sublimate. They persist, spread under wind shear, and become the cirrus clouds we see as contrails.

The forecast problem is therefore a humidity problem, not a temperature problem. Temperature matters only as a gate: contrails form only below roughly −40°C. The critical variable is RHi. Ice-supersaturated regions (ISSRs) are often only 500–2,000 feet thick vertically but can extend hundreds of miles horizontally. A forecast must resolve both the vertical layer and its horizontal extent to be operationally useful.

How do forecast models actually find ice-supersaturated air?

Numerical weather prediction (NWP) models — the same ones used for aviation turbulence and wind forecasts — compute RHi from temperature, pressure, and specific humidity at each grid point. The European Centre for Medium-Range Weather Forecasts (ECMWF) IFS model and the US GFS model both output RHi at multiple pressure levels. The key threshold is RHi ≥ 100% at temperatures below −40°C. In practice, forecasters use RHi ≥ 105% to account for model bias and to identify regions where supersaturation is robust rather than marginal.

The vertical resolution matters enormously. Most global models have 50–100 vertical levels, but the spacing between levels at cruise altitude is typically 500–1,500 feet. An ISSR that is only 800 feet thick may fall between two model levels. This is why forecast skill for contrail avoidance is lower than for temperature or wind. The horizontal resolution of operational models (9–13 km for global, 2–3 km for regional) is adequate for planning an altitude change, but the vertical sampling is the limiting factor.

What is the accuracy of current contrail forecasts?

Published trial results give a realistic picture. The Breakthrough Energy contrail avoidance trial, reported in 2023, used satellite observations of actual contrails to verify forecast performance. Over 70 test flights, the forecast correctly predicted the presence or absence of a persistent contrail in about 60–70% of cases. The Airbus/Etihad Sustainable Flight trial (2022) reported similar numbers: their contrail forecast tool flagged ISSRs with roughly 65% accuracy when verified against satellite imagery.

The Delta/Airbus contrail trial, which ran from late 2023 through 2024, reported that forecast accuracy improved to about 75–80% when the forecast lead time was under 6 hours and the model was initialized with recent radiosonde data. Beyond 18 hours, accuracy dropped below 50%. This is the central operational constraint: contrail avoidance is a tactical decision made shortly before departure, not a strategic plan made days ahead.

How much fuel does a 2,000-foot altitude change actually cost?

The fuel penalty of an altitude change depends on three variables: the direction of the change (climb or descent), the aircraft weight, and the wind at the new altitude. Climbing 2,000 feet at a heavy weight (near maximum landing weight for a narrowbody) costs more because the aircraft must accelerate and the engines operate at a less efficient point on the thrust curve. Descending 2,000 feet usually costs less fuel but may increase flight time slightly.

Published data from the Airbus/Etihad trial showed an average fuel penalty of 0.8% for a 2,000-foot descent and 1.4% for a 2,000-foot climb on an A350 flying a 6-hour sector. The Breakthrough Energy trial on a 737-800 reported penalties between 0.5% and 2.0%, with the variation driven by wind shear between the two altitudes. A 1% fuel penalty on a transatlantic flight burning 15,000 gallons is 150 gallons — about $600 at current jet fuel prices of $4.00/gallon. The climate benefit of avoiding one persistent contrail is equivalent to the warming from burning roughly 500–1,000 gallons of fuel, so the trade is favorable in most cases.

What is the CO2-versus-forcing trade in one worked example?

Let us do the arithmetic explicitly. Take a Boeing 787-9 flying London to New York, a 3,000-nautical-mile sector. The aircraft burns about 14,000 gallons of jet fuel. A 1% fuel penalty is 140 gallons. Burning 140 gallons of jet fuel emits approximately 1,260 kg of CO2 (jet fuel emits 9.0 kg CO2 per gallon).

Now the contrail side. A single persistent contrail from a widebody flying across the North Atlantic has a time-integrated radiative forcing of roughly 10–30 mW/m² over its lifetime, per the IPCC AR6 assessment. Over a 12-hour contrail lifetime and a typical contrail width of 50 km and length of 500 km, the total energy imbalance is approximately 10–30 terajoules. Converting to CO2-equivalent using a 100-year global warming potential for contrails of 1.0–1.5 (per AR6, contrails have a GWP100 of about 1.0, but their forcing is concentrated in the first hours), the avoided contrail is worth roughly 5,000–15,000 kg CO2-equivalent.

The ratio is stark: 1,260 kg of extra CO2 emitted versus 5,000–15,000 kg CO2-equivalent avoided. Even at the low end of the contrail forcing estimate, the avoidance is 4× better than the penalty. At the high end, it is 12× better. This is why the arithmetic favors avoidance whenever forecast confidence exceeds roughly 50%.

How do airlines decide whether to change altitude?

The decision rule used in operational trials is a threshold on the product of forecast confidence and contrail forcing. If the forecast says there is a 70% chance of a persistent contrail with a forcing of 20 mW/m², the expected forcing is 14 mW/m². The airline compares this to the fuel penalty’s CO2-equivalent. In practice, most trials used a simple rule: if forecast confidence exceeds 60% and the fuel penalty is under 2%, change altitude. If the penalty exceeds 2%, the decision requires a senior dispatcher’s approval.

The operational sequence is: (1) the forecast tool flags an ISSR along the planned route; (2) the dispatcher checks the fuel penalty for a 2,000-foot climb and descent using the flight planning system; (3) if the penalty is under 2%, the dispatcher files the amended altitude; (4) the pilot confirms the change is within aircraft performance limits and air traffic control approves it. The entire loop takes 10–15 minutes and happens 1–2 hours before departure.

What are the failure modes of contrail forecasting?

Three failure modes dominate. First, the vertical position error: the model places the ISSR at 36,000 feet but it is actually at 34,000 feet. The aircraft climbs into dry air and forms no contrail, or descends into the ISSR and forms one anyway. This is the most common failure and the reason forecast accuracy caps near 80%.

Second, the humidity bias: models tend to underestimate RHi in the upper troposphere because they do not fully resolve the mixing processes that create supersaturation. This leads to missed contrails — the forecast says no ISSR, but the aircraft forms a persistent contrail. The bias is worse in summer when the tropopause is higher and the air is warmer.

Third, the timing error: the ISSR moves or dissipates between the forecast initialization and the aircraft’s arrival. This is why lead time matters. A forecast valid at 12:00 UTC for a 14:00 UTC overflight is more reliable than one for 20:00 UTC.

Which airlines and trials have published real results?

The most credible published results come from three sources. The Breakthrough Energy contrail avoidance trial (2023) was the first to use satellite verification at scale. Their public summary reported that 70 test flights avoided an estimated 1,500 tonnes of CO2-equivalent at a fuel cost of 0.7% average penalty. The Airbus/Etihad Sustainable Flight trial (2022) flew 40 flights with an A350 and reported that the contrail forecast tool had a 65% hit rate. The Delta/Airbus trial (2024) expanded to 120 flights and published the lead-time accuracy curve we cited above.

All three trials used the same verification method: satellite imagery from GOES-16 or Meteosat to detect the actual contrail, then a comparison of the forecast ISSR with the observed contrail location. None of the trials reported a false-positive rate below 20%, meaning one in five altitude changes was unnecessary. This is the cost of operating at 70–80% confidence.

How does contrail avoidance fit into carbon accounting and regulation?

Contrail avoidance is not yet part of any mandatory carbon accounting scheme. ICAO’s CORSIA covers CO2 only, not contrails. The EU ETS, which includes aviation, also covers CO2 only. However, the EU’s 2023 revision of the ETS directive included a reporting requirement for non-CO2 effects, and airlines operating in Europe must report their estimated contrail forcing from 2025 onward. The reporting uses a simplified model based on flight distance, altitude, and the fraction of time spent in ISSRs as estimated by the ECMWF model.

The UK’s CAA has signaled it will adopt similar reporting. The US FAA has not proposed any contrail reporting requirement, but the agency’s 2024 Aviation Climate Action Plan mentions contrail research as a priority. For now, contrail avoidance is voluntary, and airlines that participate do so for reputational benefit and to prepare for future regulation.

What is the cost of a full contrail-avoidance program?

The cost has three components. The forecast tool itself costs $50,000–$200,000 per year for a medium-sized airline, depending on whether it is licensed from a vendor (e.g., SATAVIA, Breakthrough Energy’s tool) or built in-house. The fuel cost is the larger item: at a 1% average penalty on 10% of flights, a fleet burning 500 million gallons per year spends an extra 500,000 gallons, or about $2 million at $4.00/gallon. The operational cost — dispatcher time, ATC coordination — is small, perhaps $100,000 per year.

What happens when the forecast is wrong in the other direction — a false positive?

A false positive occurs when the model predicts an ISSR, the airline changes altitude, and no persistent contrail forms. The fuel was burned, the CO2 was emitted, and there was no climate benefit to offset it. This is not a rare event: the published trials consistently reported false-positive rates of 20–30%. The Breakthrough Energy trial, for example, found that roughly one in four altitude changes was unnecessary.

The cost of a false positive is exactly the fuel penalty we calculated earlier — typically 0.5–2% of sector fuel burn. On a 6-hour narrowbody flight burning 8,000 gallons, a 1% penalty is 80 gallons, or about $320 at $4.00/gallon. The climate cost is the CO2 from those 80 gallons: roughly 720 kg. That is real, but it is small compared to the benefit of a true positive, which avoids 5,000–15,000 kg CO2-equivalent. The asymmetry is why the decision rule works even at 60–70% confidence: the upside dwarfs the downside.

How do seasonal and regional variations change the forecast skill?

Contrail forecasting is not equally reliable everywhere or at all times of year. The skill is highest in winter over the North Atlantic and North Pacific, where the tropopause is low and the air is cold and moist. In summer, the tropopause rises, cruise temperatures are warmer, and ISSRs are thinner and more fragmented. The Delta/Airbus trial reported that forecast accuracy in summer dropped to about 55–60% compared to 75–80% in winter.

Regionally, the tropics are the hardest case. The tropical tropopause is high (50,000–55,000 feet) and most commercial traffic cruises below it, so contrails form only in the rare cold pools associated with deep convection. Forecast models have poor skill there because the relevant humidity structures are smaller than the model grid. Over continental interiors, such as the central US, the skill is intermediate: good in winter, poor in summer when convective outflow creates small, short-lived ISSRs.

What is the role of satellite data in improving forecasts?

Satellites serve two functions in contrail forecasting: verification and initialization. For verification, geostationary satellites like GOES-16 (over the Americas) and Meteosat (over Europe and Africa) provide images every 10–15 minutes, allowing analysts to see whether a contrail formed and persisted. This is how the published trials measured their accuracy. For initialization, polar-orbiting satellites with hyperspectral sounders — such as the Cross-track Infrared Sounder (CrIS) on NOAA-20 — measure upper-tropospheric humidity directly. Assimilating these observations into NWP models improves the initial RHi field, which is the single largest source of forecast error.

The practical impact of satellite data assimilation is a 5–10 percentage point improvement in forecast accuracy at lead times under 12 hours, according to analyses presented at the 2024 American Meteorological Society conference. The improvement is largest in data-sparse regions like the mid-Pacific, where radiosondes are rare. This is an active area of research, and the next generation of sounders (e.g., MTG-S on Meteosat Third Generation) is expected to add another few points of skill.

How do different aircraft types respond to the same altitude change?

The fuel penalty of a 2,000-foot altitude change is not uniform across the fleet. A heavily loaded narrowbody (e.g., a 737-800 at 90% of maximum takeoff weight) climbing 2,000 feet will burn noticeably more fuel because it is operating near its ceiling and the engines are at a high thrust setting. A lightly loaded widebody (e.g., an A350 at 60% of maximum takeoff weight) can often climb 2,000 feet with a penalty under 0.5% because it has excess thrust and the climb puts it closer to its optimal cruise altitude.

Descents are generally cheaper than climbs. Descending 2,000 feet reduces the required thrust, and the fuel penalty is often under 0.5% even for heavy aircraft. This asymmetry matters operationally: when the forecast flags an ISSR, the dispatcher will first check whether a descent is feasible within airspace constraints, because it is the cheaper option. The table below summarizes typical penalties by aircraft class and maneuver.

Can contrail avoidance be combined with other fuel-saving measures?

Yes, and the combination is often more efficient than either measure alone. The most common pairing is with continuous descent operations (CDO) and continuous climb operations (CCO). If an aircraft is already planning a step-climb profile — a series of 2,000-foot climbs as fuel burns off — the dispatcher can time those climbs to avoid ISSRs at no extra fuel cost. The Breakthrough Energy trial specifically noted that about 30% of their avoidance actions were achieved by shifting a planned step-climb by 15–30 minutes rather than adding a new altitude change.

Another pairing is with wind-optimized routing. If the forecast shows an ISSR along the great-circle route but a wind-optimal route is 1,000 feet higher or lower, the dispatcher can choose the wind-optimal route and avoid the contrail without any fuel penalty. This is the ideal case: the avoidance is free. The challenge is that wind-optimal routes are already tightly optimized, so the overlap between “wind-optimal” and “contrail-free” is not always available.

What are the limits of current forecast models at very short lead times?

At lead times under 2 hours, the forecast problem changes character. The NWP model output is still useful, but nowcasting — using satellite and radar observations to extrapolate the current state — becomes more accurate. The Delta/Airbus trial found that a blend of NWP and nowcasting improved accuracy to about 85% at lead times under 2 hours, but the operational benefit was limited because the dispatcher has little time to act.

The practical floor for lead time is about 30 minutes before the aircraft reaches the ISSR. This is the time needed for the pilot to request an altitude change, receive ATC approval, and execute the maneuver. At shorter lead times, the only option is to accept the contrail or make an emergency maneuver that is rarely worth the fuel cost. This is why the operational window for contrail avoidance is roughly 2–6 hours before departure: long enough to plan, short enough that the forecast is still accurate.

How do air traffic control constraints affect the decision?

ATC is often the binding constraint. In busy airspace — the North Atlantic tracks, the European core area, or US domestic corridors — the requested altitude may not be available. The dispatcher must then choose between accepting the contrail or taking a lateral offset, which has its own fuel cost. The North Atlantic Organized Track System (OTS) is particularly constrained: tracks are published daily, and changing altitude mid-ocean requires a request to Gander or Shanwick control, which is not always granted.

The trials handled this by filing the amended altitude in the flight plan rather than requesting it in the air. This gives ATC advance notice and increases the approval rate. The Delta/Airbus trial reported an ATC approval rate of about 85% for pre-filed altitude changes, versus about 50% for in-flight requests. This is a practical lesson: contrail avoidance works best when it is planned, not improvised.

What does a contrail avoidance program cost per tonne of CO2-equivalent avoided?

This is the number that matters for budget decisions. Using the published trial data, we can estimate it. The Breakthrough Energy trial reported avoiding 1,500 tonnes CO2-equivalent over 70 flights at an average fuel penalty of 0.7%. If the average sector burned 8,000 gallons, the total extra fuel was 70 × 8,000 × 0.007 = 3,920 gallons, or about $15,700 at $4.00/gallon. Adding the tool cost ($100,000 per year, spread over roughly 500 flights per year) gives a total cost of about $30,000 for 1,500 tonnes, or $20 per tonne CO2-equivalent avoided.

This is remarkably cheap. For comparison, the cost of sustainable aviation fuel (SAF) is $500–$1,500 per tonne of CO2 avoided, and many carbon offsets trade at $10–$100 per tonne. Contrail avoidance at $20 per tonne is competitive with the cheapest offsets and far cheaper than SAF. The caveat is that the estimate depends on the assumed contrail forcing, which has wide uncertainty. At the low end of the forcing range, the cost rises to about $60 per tonne; at the high end, it falls to $10 per tonne. Even the pessimistic case is competitive with most abatement options.

How does the cost compare to other aviation climate measures?

The comparison is instructive. The table below shows typical costs per tonne of CO2-equivalent for the main aviation climate measures, based on 2025–2026 published estimates. Contrail avoidance is the cheapest option by a wide margin, which is why it is attracting attention from airlines looking for cost-effective ways to reduce their climate impact.

What happens if an airline does nothing — what is the regulatory risk?

The regulatory trajectory is clear, even if the timeline is not. The EU ETS already requires reporting of non-CO2 effects from 2025. The European Commission has stated it will consider incorporating contrails into the ETS cap or a separate mechanism by 2028–2030. The UK CAA has signaled similar intent. ICAO’s Committee on Aviation Environmental Protection (CAEP) is developing a contrail avoidance standard, with a target for adoption by 2028.

An airline that does nothing faces three risks. First, a carbon price on contrails: if the EU ETS includes contrail forcing, the cost could be $20–$100 per tonne CO2-equivalent, which on a fleet basis could be tens of millions of dollars per year. Second, a reporting mandate that reveals the airline’s contrail footprint to regulators and the public. Third, a competitive disadvantage: airlines that adopt avoidance now will have lower costs if regulation arrives, and they will have the operational experience to comply efficiently. The EU ETS Aviation Explained guide covers the current reporting rules in detail.

How does contrail avoidance interact with the push for SAF and electric aircraft?

Contrail avoidance is complementary to both SAF and electrification, not a substitute. SAF reduces CO2 but has a negligible effect on contrail formation because it still emits water vapor and soot. The SAF Mandates Explained guide notes that even 100% SAF does not eliminate contrails. Electric aircraft, which are in early development, would eliminate both CO2 and contrails, but they are limited to short-range regional flights for the foreseeable future, as discussed in our hydrogen vs. battery comparison.

The practical near-term strategy for long-haul aviation is therefore a portfolio: SAF for CO2 reduction, contrail avoidance for non-CO2 effects, and efficiency improvements for both. The Aviation Non-CO2 Effects guide explains why contrails dominate the non-CO2 climate impact and why they cannot be ignored in any credible decarbonization plan.

What should a small airline or charter operator do differently?

Small operators face different economics. A regional charter flying 2-hour sectors at 25,000–30,000 feet may rarely encounter ISSRs, because the air is warmer and the contrail formation threshold is less often met. The cost of a contrail forecasting tool is the same $50,000–$200,000 per year, but the benefit is spread over far fewer flights. For a small operator, the tool may not pay for itself.

The practical advice is to start with a simple rule rather than a full forecasting system. The FAA’s aircraft certification guidance does not cover contrails, but operational guidance from ICAO suggests a simple threshold: if the ECMWF or GFS model output shows RHi > 100% at the planned cruise altitude and the temperature is below −40°C, consider a 2,000-foot altitude change. This rule can be applied manually by a dispatcher without any software investment. The accuracy is lower than a dedicated tool — perhaps 50–60% — but the cost is zero.

How do forecasters verify their predictions after the flight?

Verification is the discipline that separates a trial from a program. After each flight, the operator compares the forecast ISSR with the observed contrail from satellite imagery. The comparison is not trivial: the satellite image shows the contrail’s position, but the forecast is a volume of air. The standard method is to check whether the contrail’s centerline falls within the forecast ISSR’s horizontal footprint and whether the flight’s actual altitude was within the forecast vertical layer.

The trials used a binary metric: hit (contrail observed where forecast) or miss (no contrail where forecast, or contrail where none forecast). The hit rate is the fraction of flights where the forecast was correct. The false-alarm rate is the fraction of flights where an avoidance action was taken but no contrail formed. The best operational systems track both metrics continuously and recalibrate the model thresholds when the false-alarm rate rises above 30%.

What is the single most important number a dispatcher should know?

The decision threshold. If the forecast confidence is above 60% and the fuel penalty is below 2%, change altitude. This rule is simple enough to apply in the 10-minute window before departure, and it is consistent with the published trial results. The Contrail Avoidance Explained guide provides a more detailed decision framework, but this is the operational core.

Related guides: CORSIA Explained: The Best 2026 Guide to Aviation Carbon Offsets · EV Statistics 2026: Sales, Batteries, Charging and Emissions Data · Hydrogen Cars Emissions · Net Zero Cities.

The chart below compares the estimated climate forcing avoided (in W/m²) against the CO₂ penalty from the extra fuel burn, normalised to a per-flight-hour basis.

Frequently Asked Questions

Sources

• NASA aeronautics research
• ICAO environmental protection
• IEA aviation
• Our World in Data global aviation emissions

Written and edited by Muhammet Çay, Founder and Editor · LinkedIn · How we source and correct this site · Report an error

Every figure on this page is dated and linked to its source. How we verify · Editorial policy · Report a correction