Severe convective turbulence over Myanmar injured passengers and crew aboard a Boeing 777. The crew diverted to Bangkok. One passenger died and 56 people sustained serious injuries.
The encounter was about 30nm beyond PTN over Southwest Myanmar. The occurrence-area diagrams support an approximate marker. Airport markers provide geographic orientation. Basemap: Natural Earth, public domain.Read supporting evidence
At 37,000ft over Southwest Myanmar, rapidly changing vertical air movements caused severe acceleration changes. The crew stabilised the aircraft and diverted to Bangkok after learning of multiple cabin injuries.
Read the full incident Follow what happened, from the flight to the investigation findings.
What investigators concluded
TSIB found severe convectively induced turbulence. It considered intermittent missing or weak weather radar returns possible and identified limits in the prescribed maintenance test.
1,508 words / about 7 minutes to read. Based on the investigation records cited below.
A smooth flight toward Myanmar
Singapore Airlines 321 left London Heathrow at about 21:38 UTC on 20 May 2024 for Singapore. The Boeing B777-300ER, registered 9V-SWM, carried 211 passengers, three flight crew and 15 cabin crew. Before departure, the pilot-in-command briefed the in-flight manager that weather could be expected nearer Singapore. The forecast package included occasional cumulonimbus and isolated embedded cumulonimbus between southern Myanmar and Singapore. The flight remained smooth through departure, climb and cruise until the Yangon Flight Information Region.
At about 07:30 on 21 May, the aircraft was approaching Southwest Myanmar at 37,000ft. The pilot-in-command occupied the left seat and was flying; an augmenting captain in the right seat monitored the flight. The senior first officer was resting behind the cockpit. Passengers were being awakened and cabin crew were preparing breakfast. The pilots reported that the immediate flight path looked clear, with cirrus visible in the distance, and that their navigation displays showed no weather returns.
The aircraft had two selectable weather radar systems sharing an antenna. The right system was selected for the inbound flight, following the operator's policy. The captain used manual mode, selecting a weather slice at 33,000ft and maximum gain. The right-seat pilot used automatic mode with calibrated gain. The pilots reported no radar fault or failure indications. The equipment manufacturer recommended automatic mode and calibrated gain for normal use.
At 07:44:52, approaching waypoint PTN, the crew asked Yangon air traffic control for a direct route to EKAVO and cited weather. They later explained that they had no weather concern at that point and used that reason to improve the prospect of approval. ATC instead offered a direct route to DWI, which they accepted. Both pilots briefly expanded their display ranges for route planning. The right-seat pilot subsequently reduced his range through 320nm and 160nm to 40nm. He reported no weather returns during those adjustments; the voice recorder captured no discussion of weather between the two pilots.
At 07:49:23, while cruising at Mach 0.84, the aircraft began encountering turbulence. Over the following 17 seconds, recorded vertical acceleration varied between +0.44G and +1.57G. The augmenting pilot saw airspeed increasing toward the overspeed range, alerted the captain and deployed the speed brakes. The captain selected Mach 0.82, the prescribed turbulence penetration speed, switched on the fasten-seat-belt sign and called for belts to be fastened. There was insufficient time for a public-address announcement.
As altitude increased, the autopilot progressively commanded a nose-down attitude to regain the selected 37,000ft. The aircraft continued rising and reached 37,362ft. At 07:49:40, about 30nm beyond PTN, vertical acceleration fell from +1.35G to -1.5G in 0.6 seconds. It then rose from -1.5G to +1.5G over the next four seconds. Those changes could lift an unbelted occupant into the air and bring that person down again. The stick shaker activated momentarily. The crew made manual control inputs, disengaging the autopilot, and controlled the aircraft manually for 21 seconds before reengaging it at 07:50:05. TSIB assessed their response as understandable and appropriate.
Cabin crew informed the pilots of multiple injuries. The senior first officer, awakened by the turbulence, returned toward the cockpit and saw toppled service carts, broken service wares and an injured cabin crew member receiving help. At about 07:55, the crew reported severe turbulence to Yangon ATC and asked about reports from other aircraft at their altitude. ATC had received none from aircraft in the vicinity. At about 08:01 the pilots declared Mayday and requested a diversion to Bangkok Suvarnabhumi. Two minutes later they informed the airline by an aircraft communications message.
The senior first officer took over monitoring duties from the augmenting captain, who helped assess the cabin and coordinate preparations. One passenger was unconscious and receiving cardiopulmonary resuscitation. Cabin crew and passengers helped injured people back into seats where possible. Those with injuries that prevented movement were cushioned with blankets and pillows and held securely by people seated beside them for landing. The in-flight manager confirmed that crew capable of carrying out safety duties could man all cabin doors. The aircraft landed at about 08:45 and reached its parking bay at about 08:50, where medical personnel treated the injured and arranged hospital transport.
The final report's injury table records one passenger fatality, 56 serious injuries and 23 minor injuries. Five cabin crew and 51 passengers were seriously injured; one cabin crew member and 22 passengers had minor injuries. Another 149 people were uninjured. Injuries mainly involved the head, neck and spine, including fractures, compression injuries, lacerations and dislocations. Thai Police performed an autopsy, and their official certificate gave heart failure and lung edema as the passenger's cause of death.
The cabin had broken ceiling panels, dislodged passenger service units, punctured overhead bins, damaged seat-back reclining mechanisms, cracked sidewalls and broken lights. Inspections found no evidence of pre-existing defects in cockpit or cabin seat structures, and the seat belts were in working condition. A separate test in Bangkok found the fasten-seat-belt sign system operating normally. Myanmar delegated the investigation to Singapore's Transport Safety Investigation Bureau, which classified the occurrence as an accident.
Investigators compared satellite imagery, Yangon ground radar information and observations from four other flight crews in Myanmar airspace. Those crews saw widespread clouds and used their radar systems to divert around weather cells. They experienced light to moderate turbulence. Their aircraft flew different tracks and altitudes from 9V-SWM, so their observations described conditions in the surrounding region. TSIB could not determine why the occurrence crew had not seen the widespread clouds.
Satellite analysis identified a rapidly developing convective cloud at the encounter location. Its top rose from about 27,500ft at 07:40 to about 40,000ft at 07:50, growing through the aircraft's cruising altitude. TSIB considered the turbulence likely to have resulted from the strong vertical movements associated with this growth. The operator estimated an updraft of about 8,000 to 9,000ft/min; an independent Boeing calculation estimated maximum vertical winds of about 9,000ft/min. TSIB identified severe convectively induced turbulence and excluded clear air and mechanical turbulence as factors.
Maintenance records showed two reports of weak returns from the aircraft's right radar on 29 April and 1 May 2024. A further report on 15 May described no returns despite convective clouds being visible outside. On that flight, switching to the left radar brought weather returns onto the displays. No radar fault messages accompanied those three events. After each report, the system passed the prescribed maintenance test. The three reports fell outside the operator's trigger for additional attention to a defect recurring three times within ten days.
TSIB considered the occurrence pilots' description of blank weather displays credible, supported by the recorded absence of weather discussion and the captain's surprise after the encounter. Returns reappeared during the diversion after descent below 31,000ft. A ferry crew later observed weaker returns from the right radar than from the left. The equipment manufacturer disputed that comparison because photographs were taken one minute after switching systems, before the specified five-minute cycle had elapsed. The crew said the display remained similar after five minutes. TSIB considered the observations together with the maintenance history to support concern about intermittent radar performance.
Investigators sent the radar components for extensive examination. The equipment manufacturer assessed the reconstructed system's performance as within limits and found no hardware defect explaining the reported missing weather. TSIB questioned differences in test returns, including weaker returns directly ahead and stronger returns to the side. Cold-temperature testing produced erratic antenna scanning under some conditions, while testing across estimated occurrence-flight temperatures passed. Control-panel tests intermittently showed a variable-gain indication with the knob at its calibrated position; the panel subsequently passed another test without adjustment. Transceiver stress tests and visual inspection found no defective diodes.
The equipment manufacturer concluded that the tests provided no evidence of inaccurate weather detection or display during the occurrence flight. TSIB concluded that intermittent missing or weak returns remained possible. It found that the maintenance manual's ground test checked system integrity but could neither reproduce the in-flight environment nor identify the faults behind those display problems. Existing equipment also lacked a recording of the radar images shown to the pilots. Investigators therefore could not examine the displayed weather directly.
The final report, issued on 18 May 2026, recorded changes made by Singapore Airlines. These included turbulence information on crew tablets, additional forecasting information for rapidly developing cumulus areas and the SkyPath application. Weather radar refresher training ran from July to December 2024. The airline added turbulence learning materials and cabin crew training, strengthened seat-belt reminders and issued guidance on identifying and troubleshooting radar display problems. It also began sending reports of missing or weak returns to Boeing and the radar equipment manufacturer.
TSIB made four recommendations: Boeing should develop guidance for crews to detect and respond to missing or weak weather returns, and improve maintenance guidance for identifying those problems. The radar manufacturer should develop a way to record displayed weather images. ICAO should extend flight crew-machine interface recording requirements to older aircraft types in the specified large-aircraft category. TSIB also stressed keeping seat belts fastened while seated, even when the sign is off, because turbulence can arrive suddenly.
The acceleration changes could lift unbelted occupants and bring them down again. The injury table records 56 seriously injured and 23 with minor injuries.
TSIB found severe convectively induced turbulence associated with rapidly developing clouds over Southwest Myanmar. Clear air turbulence and mechanical turbulence were excluded as factors.
The captain switched it on after turbulence began and called for belts to be fastened. The crew had no time for a public-address announcement before the severe acceleration changes.
TSIB considered intermittent missing or weak weather returns possible. Earlier defect reports, the crew observations and the limits of maintenance tests supported that finding.
The tests did not identify a hardware fault that explained the occurrence. TSIB and the equipment manufacturer reached different assessments of the evidence for intermittent radar display problems.
The aircraft did not record the weather images shown on its navigation displays, limiting reconstruction of the radar information available to the crew.
Actions and recommendations are described at the dates of the cited records. A recommendation is not proof of implementation.
Airline training and weather information
The final report records radar refresher training from July to December 2024, turbulence learning materials and additional weather information tools for flight crews.
Singapore Airlines issued radar troubleshooting guidance and began forwarding reports of missing or weak weather returns to the aircraft and radar manufacturers.
TSIB RA-2026-004 called for recording weather display images. RA-2026-005 called for ICAO to extend interface recording requirements to older aircraft types in the specified large-aircraft category.
These comparisons connect specific research topics. Shared circumstances do not establish identical causes. Evidence from both investigations is provided below.
Both investigations examined injuries during sudden vertical acceleration changes and the protection provided by seat belts. Singapore Airlines 321 encountered severe convective turbulence; Qantas 72 involved nose-down flight-control commands following erroneous angle-of-attack data. The ATSB found substantially greater injury rates and severity among occupants who were standing or seated without belts.
Explore the records below for additional context. They open on the publisher's site and are not all included in our report search. Source reuse terms apply.
Kevin Murphy writes the incident accounts on AirlineIncidents.com, out of a long interest in aviation safety and the lessons each investigation leaves behind.
Prepared by
AirlineIncidents editorial desk
Report citation check
Independent expert review
Not completed
Citation checks compare supporting text with the indexed report pages. They do not constitute independent expert review. How these pages are prepared.
.jpg){kind=link}