DH8A, Saulte Ste. Marie ON Canada, 2015

DH8A, Saulte Ste. Marie ON Canada, 2015


On 24 February 2015, the crew of a Bombardier DHC8-100 continued an already unstable approach towards a landing despite losing sight of the runway as visibility deteriorated in blowing snow. The aircraft touched down approximately 140 metres before the start of the paved surface. The continued unstable approach was attributed by the Investigation to plan continuation bias compounded by confirmation bias. It was also found that although the aircraft operator had had an approved SMS in place for almost six years, it had not detected that approaches made by the aircraft type involved were routinely unstable.

Event Details
Event Type
Flight Conditions
Flight Details
Type of Flight
Public Transport (Passenger)
Take-off Commenced
Flight Airborne
Flight Completed
Phase of Flight
Location - Airport
Approach not stabilised, Inadequate Aircraft Operator Procedures, Non Precision Approach
Into terrain, Into obstruction, No Visual Reference, Vertical navigation error, IFR flight plan, Undershoot on Landing
Plan Continuation Bias, Procedural non compliance, Violation, Ineffective Monitoring - PIC as PF
Precipitation-limited IFV
Damage or injury
Aircraft damage
Non-aircraft damage
Non-occupant Casualties
Number of Non-occupant Fatalities
Number of Occupant Fatalities
Off Airport Landing
Causal Factor Group(s)
Aircraft Operation
Safety Recommendation(s)
None Made
Investigation Type


On 24 February 2015, a Bombardier DHC8-100 (C-GTAI) being operated by Jazz Aviation (Air Canada Express) on a scheduled domestic passenger flight from Toronto to Saulte Ste. Marie as JZA7795 touched down approximately 140 metres short of the intended landing runway at destination in day IMC before continuing onto it and stopping after travelling 460 metres along the 1,830 metre-long runway. After a requested inspection by the emergency services found no obvious damage to the aircraft, it was taxied to the terminal for normal passenger disembarkation. Although none of the 18 occupants were injured, the aircraft was subsequently found to have sustained "significant" damage and damage to the approach lighting was also found.


An Investigation was carried out by the Canadian TSB. Relevant data was successfully recovered from the FDR but as the crew failed to trip the CB protecting the 30 minute CVR after the accident, relevant data was lost. Data stored on the NVM from the aircraft TAWS was also recovered.

It was found that the Captain, who was PF for the investigated flight, had been employed by Jazz Aviation for 17 years and had accumulated over 12,000 total flying hours including 9,000 hours on the Company's DHC-8 aircraft. The First Officer had been employed by Jazz Aviation for 2 years and had accumulated 6,630 flying hours including over 1,300 hours on the Company's DHC-8 aircraft.

It was established that the flight crew had been aware of the likelihood of marginal weather conditions at their destination when the Captain conducted the approach briefing shortly before the top of descent and he had included the possibility of a go around on that account. The applicable Vref was determined as 96 knots and the corresponding Vapp as 101 knots. When ATC issued descent clearance to 5,000 feet QNH, they noted the then-prevailing weather conditions and requested that the crew report when they had the destination aerodrome in sight for a visual approach. However when ice crystals subsequently reduced in-flight visibility, the crew requested a VOR/DME approach to runway 30 rather than a visual approach and were so cleared. Once below 3,000 feet QNH, the aircraft cleared the area of ice crystals and the runway could be seen. However, a significant snow shower was also visible approaching the aerodrome from the west and, having reported this to ATC, the flight was cleared to deviate as necessary from the VOR approach maintaining visual reference.

Descent was continued visually on a 3° vertical profile and 5nm from touchdown, ATC advised of a "line of weather currently rolling across the runway" which had led to RVR (recorded near the western edge of the runway) decreasing to 335 metres. At 1,000 feet and just inside 3nm from touchdown, landing clearance was given with the recorded RVR now 300 metres. With airspeed at 148 KCAS, the flaps were selected to 15° and engine torque, which had been at or near flight idle since 1,500 feet, was increased to 25%. The aircraft had remained on an appropriate vertical profile with adjustments to engine torque in response to a gusting headwind and by 500 feet, airspeed had reduced to 122 KCAS - noted as still 21 knots above Vapp.

At 200 feet, the Captain "began to reduce toque to idle and, as a result, the airspeed began to reduce rapidly". Although nose-up pitch "was gradually increased […] the vertical path steepened due to the decreasing airspeed and resultant ground speed reduction" and the aircraft descended below the 3° vertical path. The crew reported that "at some point below 200 feet" they had "lost visual reference to the ground due to the approaching weather system of blowing snow" but had nevertheless continued the approach. When the TAWS auto callout at 20 feet occurred "torque was increased towards 30%" and two seconds later, the aircraft "contacted the ground" approximately 140 metres prior to the runway threshold at a speed of 94 KCAS. This ground was covered in approximately 20 to 30 cm of snow. As the aircraft contacted the ground, FDR data showed that it had been in a level pitch attitude and that contact had involved a peak vertical acceleration of 2.32g.

Following this premature touchdown, the NLG struck and damaged an approach light situated 90 metres prior to the runway threshold. The pilots reported having "heard a thump, but had not seen the light and were unsure what had caused the noise". The aircraft came to a stop on the runway and "the flight crew assessed the ground visibility as very poor, due to the blowing snow". Since they were uncertain as to the status of the aircraft, TWR were asked to organise an external inspection before taxiing was attempted and also a bus to take the passengers to the terminal. However, "when the emergency vehicles arrived and assessed the condition of the aircraft, no significant damage was noticed [and] the flight crew elected to taxi the aircraft to the gate" to disembark the passengers and this was accomplished without further event. Damage to the aircraft was subsequently found which required replacement of the NLG assembly and its bay doors and since both MLG assemblies had exceeded load limits, they also had to be replaced.

It was noted that the most recent recurrent training received by both pilots had been three months earlier and had included "practising rejected landing and missed approach procedures following loss of visual cues at 100 feet" which they had completed "without recorded difficulty although neither could remember the specifics of the rejected landing training event".

It was found that the Jazz Aviation AOM stated that once on the inbound track of a non-precision approach, the landing gear should be extended, the flaps set to 15°, the landing checks completed and airspeed reduced to 120 knots. A different section of the AOM stated that "under normal conditions in visual meteorological conditions, the aircraft shall be in a stabilised approach by 500 feet height above aerodrome" and specified that such an approach requires that both forward and vertical speed be stable and that airspeed should be Vref + 5 knots to 500 feet and then reduce gradually to achieve Vref at touchdown". The AOM also required "a missed approach if the runway environment is lost to view below minimums" and indicated that "failure to achieve or maintain a stabilised condition is the basis for a missed approach".

It was further found that "the flight crew’s understanding of the appropriate airspeeds for normal operations" at the time of the investigated approach was 150 knots in descent to 500 feet then 120 knots from 500 feet to 200 feet and then Vref + 5 knots to touchdown, to be "achieved by power reduction as necessary".

In order to to determine if the speeds flown during the investigated approach were an exception to those of normal operations, "the remaining 285 flights recorded on the occurrence aircraft’s FDR were examined for similar SOP deviances". It was found that "84% of recorded flights exceeded the 10-knot allowable tolerance over the required Vref + 5 knots below 500 feet" as measured at 400 feet with the average speed at that point being Vref + 22 knots. This data confirmed that a constantly decelerating approach was habitual. It also confirmed that although "the majority of recorded flights were outside of the criteria for a stable approach" none of them resulted in a go around.

The Investigation found that soon after the event under investigation, although not as a result of it, Jazz Aviation had modified AOM references to "stabilised approach factors” to include the comments that stabilised airspeed is normally equal to the bugged approach speed and that deviations of +10/-5 knots are acceptable if airspeed is trending towards VApp. The Investigation examined the data from the second fleet aircraft FDR which recorded flights made subsequent to promulgation of these changes but found that "in terms of speed exceedance and approach deceleration, the results were similar to those from the occurrence FDR". These findings led to "the TSB's determination that there appeared to be a systemic deviance from stabilised approach SOPs, particularly the recorded airspeeds at 400 feet" and this was communicated to the operator. The response was that "it was the pilot's responsibility to understand that the target speed of 120 knots […] would be an inappropriate speed to maintain below 500 feet with a briefed bug speed of 96 knots" and that "as long as it is not excessive, Jazz considers a speed reduction from an SOP target speed (of 120 knots to 500 feet) to the minimum manoeuvring speed (Vref + 5 knots) to be stable".

It was noted that Jazz Aviation had had a Transport Canada-approved SMS since June 2009. However, they had not taken any steps to establish an OFDM programme for their DHC8-100 fleet (nor for their DHC-300 fleet) due to uncertainty about their future nor had they substituted any other process for monitoring the effectiveness of or compliance with their stabilised approach procedures. It was also noted that as the result of an earlier Investigation into a 2011 accident, the Board had already made a Safety Recommendation (A14-01) to Transport Canada to "require operators to monitor and reduce the incidence of unstable approaches that are continued to a landing". Action by Transport Canada in response to this Recommendation to date was noted and it was further noted that the Board looks forward to gaining a better understanding of "what measures airlines have implemented and assess whether they are effective in addressing the underlying safety deficiency associated with Recommendation A14-01".

The Investigation considered the likely motivation for the violation which had occurred when the final approach was continued after loss of visual reference and came to the conclusion that plan continuation bias compounded by confirmation bias was probable.

The formally stated Findings as to Causes and Contributing Factors were as follows:

  1. The company standard operating procedures require an approach speed of Vref + 5 knots; however, this is being interpreted by flight crews as a target to which they should decelerate, from 120 knots, once the aircraft is below 500 feet. As a result, the majority of examined approaches, including the occurrence approach, were unstable, due to this deceleration.
  2. Due to ambiguity in the guidance and uncertainty as to the required speeds during the approach, the crew did not recognise that the approach was unstable, and continued.
  3. On the approach, the pilot flying reduced power to idle to reduce the approach speed from 122 knots toward 101 knots at 200 feet above ground level. This steepened the aircraft’s vertical path.
  4. The rapidly decreasing visibility resulted in the airport environment and the precision approach path indicator lights becoming obscured; as a result, the steepened vertical profile went unnoticed and uncorrected.
  5. Although the loss of visual reference required a go-around, the crew continued the approach to land as a result of plan continuation bias.
  6. The terrain awareness and warning system did not alert the crew to the aircraft’s proximity to the ground once the aircraft was below 50 feet, possibly due to the rapid rate of closure. This lack of warning contributed to the crew not being aware of the aircraft’s height above ground.
  7. Due to the uncorrected steepened vertical profile, loss of visual reference, and lack of normal terrain warning, the aircraft contacted the surface approximately 450 feet prior to the runway threshold.

The formally stated Findings as to Risk were as follows:

  1. If guidance provided to flight crews allows for large tolerance windows, and crews are not trained to recognise an unstable condition, then there is a continued risk that flights that are unstable will be continued to a landing.
  2. If approaches that require excessive deceleration below established stabilisation heights are routinely flown, then there is a continued risk of an approach or landing accident.
  3. If crews do not report unstable approaches and operators do not conduct flight data monitoring but rely only on safety management system reports to determine the frequency of unstable approaches, there is a risk that these issues will persist and contribute to an accident.

Safety Action eventually taken by Jazz Aviation "as a direct result" of the investigated event has included significant modifications to the stabilised approach criteria specified in the AOM. By 1000 feet aal, airspeed must now be trending towards the Target Speed of 120 KIAS or the bugged VApp - whichever is higher - and by 500 feet aal, speed must be stabilised at the bugged VApp with deviations up to +10/-5 knots only acceptable if the airspeed is trending toward the bugged VApp.

The Final Report of the Investigation was authorised for release on 4 January 2017 and it was officially released on 9 March 2017. No Safety Recommendations were made.

Related Articles

SKYbrary Partners:

Safety knowledge contributed by: