The Contrail Fix Works. It Travels In Item 18, As Free Text.

GOTHENBURG, Sweden — Between March 2025 and April 2026, one airline asked one air navigation service provider to move 3,114 flights up or down to keep them out of ice-supersaturated air.

Of those, 1,775 were approved. The approved flights formed measurably fewer persistent contrails than comparable flights that stayed where they were, and the extra fuel came to 0.11 per cent of the airline's block fuel across the whole period.

The physics is not the problem. The form is.

What the aircraft is being asked to do

A contrail that lasts is not the same object as a contrail that vanishes. The short white line behind an aircraft on a dry day is ice that sublimates in seconds and warms nothing.

The ones that matter form in ice-supersaturated regions: air already holding more water vapour than ice at that temperature can keep, waiting for something to condense onto. An aircraft supplies the something. The trail then spreads, thins into cirrus, and traps outgoing heat for hours.

These layers are shallow. In this airspace over the fourteen months, the median supersaturated layer was about 1,900 feet deep. An aircraft cruising at flight level 350 that climbs or descends 2,000 feet is usually out of it.

That is the entire proposition. Not spraying anything, not adding anything, not removing anything. Moving a small number of aircraft a small distance vertically for part of a sector.

What the trial found

We asked both organisations for the request records rather than the results, and joined them to the satellite detections ourselves.

On the approved diversions, the detection product scored a persistent trail on 19 per cent of scored segments. On matched segments flown at the originally filed level, it scored one on 47 per cent.

The fuel: 1.4 per cent more on the diverted sector itself, 0.11 per cent across all of the airline's block fuel in the period, because most flights were never diverted at all.

Those are good numbers. They are the numbers the idea has needed for a decade.

Item 18

An ICAO flight plan has coded fields for almost everything. Aircraft type, wake category, departure aerodrome, cruising level, alternates — all of it in fixed characters that a machine reads without a person in the loop.

There is no coded field for "I would like to be 2,000 feet higher because the air where I am is going to make a cirrus sheet."

So it goes into Item 18, which is the free-text box. All 3,114 requests were filed there, as a string beginning CLIMATE/, with a median length of 141 characters, alongside remarks about catering uplift and unserviceable lavatories.

A controller may read it. A controller may not. Nothing in the system requires it, flags it, or counts it.

Of the 3,114, some 1,339 — 43 per cent — were refused for traffic. The dataset has a field for the refusal reason. It is filled in 61 times.

The two in three

Here is the part that is worse for the trial than anything the form does to it.

Of the 1,775 approved diversions, the satellite retrieval afterwards indicates that the originally filed level would not have produced a persistent trail on 1,171 of them. Roughly two diversions in three moved an aircraft out of air that would not have made a lasting trail anyway.

The trial's own meteorologist disputes the word we first used for that, at length, at the foot of this page, and her objection is the best argument in this story. We had written "unnecessary". She points out that the supersaturated layers covered 11 per cent of cruise segments, that the product was aiming at real air 34 per cent of the time, and that scoring an avoidance decision against what subsequently happened is a standard nobody applies to de-icing or to lightning.

She is right, and we have left our figure standing with her correction under it rather than quietly swapping the sentence.

How it stopped

No ruling ended this trial. No regulator, no objection, no finding.

Dr. Marit Ekelund-Byrne built the daily supersaturation product and ran it alone for fourteen months. She went on parental leave in May. The product stopped being issued, because nobody else at either organisation had been trained to issue it.

She told the airline in writing on 4 February, twenty-two weeks ahead, with a handover plan and three names. There was a budget line for the training. It was not spent.

The trial is described by both parties as suspended rather than concluded. The working group next meets in November.

What this desk takes from it

We have spent eleven years on this desk counting trails twice a day to establish that persistence is humidity. It is humidity. Somebody has now built a product that forecasts the humidity and an airline has flown against it and it works, at a fuel cost you would struggle to see in an annual account.

The thing standing between that and the sky is a text box, a form with no field for the question, and a training course that nobody booked.