On 21 May 2024, Singapore Airlines flight SQ321 hit convective turbulence above Myanmar that lasted 21 seconds. In a 4.6-second plunge, unbelted passengers struck the ceiling at 1.5G. One person died, 79 were seriously injured. Overhead panels shattered under the force of human bodies. The seatbelt sign had been off. SQ321 was extreme but not aberrant. Turbulence is the leading cause of injuries to passengers and crew in non-fatal accidents, according to the International Air Transport Association. The FAA records roughly a dozen serious turbulence injuries per year in the US alone. The top ten most turbulent routes globally all cross the Andes or Himalayas. But the geography of risk is shifting. The physics is direct: every degree of warming lets the atmosphere hold 7% more water vapour, the fuel of thunderstorms. A typical thunderstorm already contains more energy than a nuclear weapon, says Richard Woodward, a former RAAF test pilot and 32-year Qantas captain. More heat energy means stronger updrafts, more violent downdrafts, and turbulence extending thousands of feet above clouds and 20-plus miles laterally. In the same month as SQ321, a Qantas 737 hit severe turbulence approaching Brisbane at 11,400ft — three crew were injured, one flew subsequent flights with an undiagnosed concussion. The Australian Transport Safety Board excoriated Qantas's response. The more dangerous frontier is clear-air turbulence (CAT), which occurs at cruising altitude near jet-stream edges and is invisible to onboard radar. University of Reading scientists analysed four decades of atmospheric data (1979–2020) and found severe CAT increased 55%, moderate CAT rose 37%, and light CAT climbed 17%. These are not projections — they are measurements of what already happened. Under even the most optimistic warming scenario of 1.5°C above preindustrial levels, CAT will increase a further 27%. At 2.75°C of warming, the increase hits 56%. The detection gap is the structural vulnerability. Thunderstorm turbulence is at least partially visible on weather radar and can be forecast. CAT near jet streams can be modelled with reasonable resolution, says Ignacio Gallego-Marcos, founder of Turbli.com, a public turbulence forecasting site with 300,000 monthly users. But CAT generated near thunderstorm edges — exactly the type that downed SQ321's cabin — sits in a forecasting blind spot because convective cells are too small and dynamic for current weather models to resolve. Pilots still rely heavily on reports from other aircraft and weather advisories, which IATA's Stuart Fox acknowledges "may not be available at the right time and can vary in quality." IATA began collecting turbulence observations from about 30 airlines and more than 3,700 aircraft in 2018, building a data layer that didn't previously exist. Todd Lane, a professor of atmospheric science at the University of Melbourne who has studied CAT for 25 years, and Milton Speer at UTS have documented increasingly frequent thunderstorm microbursts in eastern Australia. Brisbane–Sydney consistently ranks as Australia's most turbulent route. The core policy question is adaptation speed. Aviation safety infrastructure — forecasting models, real-time detection, seatbelt compliance protocols, crew injury response — was built for a climate that no longer exists. The atmosphere is loading more energy into systems that were already powerful enough to kill. The detection gap for clear-air turbulence is not a technology problem waiting for a breakthrough; it is a systemic under-investment in sensor networks, model resolution, and operational protocols that must now race a physically accelerating threat.
