"'Catastrophic' airplane turbulence is only getting worse", ran a New York Post headline on 12 October, in an article that quoted University of Reading meteorologist Paul Williams1. The article opened with Avianca flight 42 from Medellín to New York, which diverted to Miami on 9 October after what officials described as severe turbulence2. Avianca said three crew members and six passengers were affected, and Miami-Dade Fire Rescue took eight people to hospital2. The headline figure in the article, 55% more turbulence over the North Atlantic, comes from a peer-reviewed study published in 202313. For pilots the useful questions are what that study actually measured, how certain it is, and what it changes on the flight deck.

What the study measured

The study, by Mark Prosser, Paul Williams, Graeme Marlton and Giles Harrison, appeared in Geophysical Research Letters in June 20233. It used the ERA5 reanalysis for 1979 to 2020 and calculated 21 clear-air turbulence (CAT) diagnostics on the 197 hPa pressure level, roughly FL3903. At an average point over the North Atlantic, the annual duration of severe-or-greater CAT on the fitted trend rose by 55%, from 17.7 hours in 1979 to 27.4 hours in 20203. Moderate-or-greater CAT rose by 37%, from 70.0 to 96.1 hours, and light-or-greater by 17%, from 466.5 to 546.8 hours3. Over the continental United States, the relative increase in moderate-or-greater CAT was 29%3. For moderate-or-greater CAT over the North Atlantic, 17 of the 21 diagnostics showed a significant upward trend and none a significant downward trend3. The authors say the increases are consistent with climate change, as warmer air from CO2 emissions increases wind shear in the jet streams4.

The limits matter. These are diagnosed values from reanalysis data, not turbulence measured by aircraft: the authors write that the record of automated eddy dissipation rate (EDR) measurements is too short for such a trend and that pilot reports are not quantitative3. They also note that improvements in forecasting should reduce the probability of encountering turbulence, even if the amount in the atmosphere is increasing3. A 2025 follow-up from Reading, by Joana Medeiros and Paul Williams, used 26 climate models and found that wind shear at cruising altitudes will increase by 16 to 27% and the atmosphere will become 10 to 20% less stable between 2015 and 21005. Those are projections, not observations5.

Not every severe encounter is clear-air turbulence

The best-known recent accident was convective. In its final report dated 18 May 2026, Singapore's Transport Safety Investigation Bureau (TSIB) found that Singapore Airlines Boeing 777-300ER 9V-SWM, operating flight SQ321 from London to Singapore, "had encountered severe convective induced turbulence" over southwest Myanmar on 21 May 20246. The aircraft was cruising at FL370 when vertical acceleration went from +1.35G to -1.5G within 0.6 seconds and then to +1.5G within the next four seconds6. One passenger died and 56 occupants were seriously injured6. The crew reported no cloud on the immediate flight path and no returns on their navigation displays6. The bureau could not rule out that the weather radar was under-painting or not painting, although the radar manufacturer found no evidence that it failed to detect the weather6.

How turbulence is forecast and reported

The FAA describes CAT as high-altitude turbulence outside convective cloud that is often met "unexpectedly and frequently without visual clues"7. It is most common near the jet streams; the FAA gives 110 kt as the usual threshold core speed, and says moderate CAT is likely when vertical wind shear reaches 5 kt per 1,000 ft or horizontal shear 40 kt per 150 miles7. High-level significant weather (SIGWX) charts highlight areas of moderate or severe turbulence, and SIGMETs and AIRMETs warn of severe and moderate turbulence respectively8. Graphical Turbulence Guidance (GTG), developed by the National Center for Atmospheric Research and produced by the Aviation Weather Center, runs more than 10 turbulence algorithms and weights them against pilot reports and automated aircraft data7. The Aviation Weather Center expresses GTG in EDR to the one-third power9. EDR is the ICAO standard for automated turbulence reporting and describes the atmosphere rather than the response of one aircraft type7. IATA's Turbulence Aware platform receives existing turbulence data from airlines, de-identifies it and provides it back to them as real-time turbulence reports10. An operator article republished on EASA's Community site in May 2024 describes a Shear Rate value on its operational flight plans, derived from the wind 2,000 ft above and below the planned level11. Pilot reports still matter: the FAA notes that the success of GTG is proportional to the number of quality PIREPs and automated reports available7.

Procedures on the flight deck

The same operator guidance recommends briefing cabin crew on the type, severity, duration and location of expected turbulence, and switching the seat belt sign on when turbulence is expected11. Where turbulence is expected in the descent, the article suggests preparing the cabin for landing early enough that it is secure when the turbulence starts11. Convective cells are best avoided laterally rather than vertically, because buffet and performance margins shrink at high altitude and turbulence can extend above the visible top of a cumulonimbus11. For CAT, it says a vertical escape can be considered, keeping in mind that the buffet margin reduces with altitude11. Aircraft manuals also specify a turbulence penetration speed: on the Singapore Airlines flight the captain reduced from Mach 0.84 to 0.82, the speed the TSIB says the flight crew operating manual requires to avoid exceeding structural design limits, and switched on the seat belt sign6. The FAA warns that a seat belt sign left on for long periods for other reasons can lose its effectiveness, and says flight attendants should feel free to ask for it to be switched on12.

The cabin is where people get hurt. Between 2009 and 2018 turbulence was involved in 37.6% of accidents in US Part 121 operations, 111 of 295, and most of them resulted in serious injuries but no aircraft damage8. Convective turbulence was present in 57.7% of those accidents and CAT in 28.8%8. About half happened during descent or approach, 60% of those below 20,000 ft, and flight attendants were the most commonly injured8. Of 207 serious turbulence injuries on Part 121 flights from 2009 to 2024 listed by the FAA, 166 were crew13. Seat belts work: from 1980 to 2003, the FAA counted only four people seriously injured in turbulence while seated with belts fastened12.

What this means for pilots in training

Williams has said pilots may need to keep seat belt signs on longer and suspend cabin service more often, and that airlines will also need new technology to spot turbulence before it hits5. EASA said in 2023 that climate change is likely to increase the frequency and severity of CAT in some dense traffic regions, such as the North Atlantic, South-East Asia and the North Pacific, and called severe CAT one of the main causes of passenger and cabin crew injuries worldwide14. For trainees that puts weight on familiar skills: reading SIGWX charts and jet stream positions, interpreting GTG and EDR, filing accurate PIREPs, knowing the limits of airborne weather radar and briefing the cabin early. The Singapore Airlines accident also shows how little time there can be: according to the TSIB, the crew did not have time to make a passenger announcement6.

Frequently asked questions

Can weather radar detect clear-air turbulence?

No. Airborne weather radar mainly detects water droplets, so even convective cloud can show only a weak return11. Clear-air turbulence cannot be seen on radar, the Reading researchers note5.