causalfactors

Metro pilots lost the big picture during a difficult approach.

BY MARK LACAGNINA

CFIT in (NOAA) National Oceanic & Atmospheric Administration U.S. 28 | flight safety foundation | AeroSafetyWorld | June 2007 causalfactors

n experienced pilot with a history of Accident Flight Route noncompliance with standard operating procedures (SOPs), an inexperienced and nonassertive copilot, excessive airspeeds Cape York Aand descent rates during a nonprecision approach Ť in bad weather, and the operator’s disregard of its own rules and training standards were found to have played roles in the May 7, 2005, crash of a Fairchild Metro 23 in Queensland, Australia. In its final report, the Australian Transport Wenlock River Safety Bureau (ATSB) said, “The accident was almost certainly the result of controlled flight into terrain [CFIT] — that is, an airworthy Lockhart River aircraft under the control of the flight crew was flown unintentionally into terrain, prob- ably with no prior awareness by the crew of the aircraft’s proximity to terrain.” Both pilots and all 13 passengers were killed in the accident, Princess which occurred near Lockhart River. Charlotte Bay The Metro 23 and eight other aircraft were op-

erated by Transair from its main base in Normandy River and ancillary bases in , Grafton and Inver- ell.1 The company employed 21 full-time pilots. The morning of the accident, the flight crew had flown the Metro from Cairns to Lockhart QUEENSLAND River and Bamaga. The accident occurred M itch ell R on the return trip to Cairns, on the leg from iver Bamaga to Lockhart River (Figure 1). Staaten River Exceeding the Limits Cairns Gilbert River The pilot-in-command (PIC), 40, held an airline transport pilot license and had 6,072 flight hours, including 3,249 flight hours in Metros. He was

employed by Transair as a line pilot in March Source: Australian Transport Safety Bureau 2001, promoted to supervisory pilot in September 2002 and to Cairns base manager in August 2003. Figure 1 The report said that there were no records indicating that the PIC had received training on conducted by the PIC. Several Transair copilots crew resource management (CRM), as required had expressed concern to a supervisory pilot that by the Transair Operations Manual. the PIC did not follow company procedures, in- The PIC had a history of noncompliance with cluding airspeed limits. One copilot said that the SOPs. A previous employer had placed him on PIC would slow down only if asked to do so by probation for not following company procedures. a copilot he respected. Another copilot said that Flight data recorder (FDR) data from the accident he had to be assertive to prevent the PIC from aircraft indicated that descent rates and airspeeds descending below the minimum sector altitude. had exceeded those specified by Transair’s SOPs “The chief pilot [of Transair] reported that CFIT in Queensland during two previous instrument approaches he could not recall ever receiving any specific www.flightsafety.org | AeroSafetyWorld | June 2007 | 29 causalfactors

indicated that he had passed aircraft ground Fairchild SA-227DC Metro 23 training despite earning a score of 77 percent on a test of aircraft systems and operating limita- tions; the company operations manual required a minimum score of 80 percent. The copilot also was not checked by a check pilot, as required by the manual, before he began line operations. “Pilots who flew with the copilot reported that he was keen to learn,” the report said. “The copilot’s flying ability and systems knowledge were generally reported as being consistent with his fly- ing experience.” The copilot also was described by colleagues as quiet, shy and nonassertive. The PIC and copilot previously had flown together on 10 days, completing 27 flight sec-

@ Craig Murray/Airliners.net tors. The copilot had told other Transair pilots that the PIC was difficult and authoritarian, and esigner Edward J. Swearingen’s Merlin corporate/business aircraft that he did not provide effective instruction and first flew in 1965 with Pratt & Whitney Canada PT6A‑20 engines. did not comply with SOPs. DAll subsequent versions of the Merlin and its longer-fuselage, 19- passenger regional airline derivative, the Metro, have had Garrett, now Bad Weather Honeywell, TPE331 engines. The original SA‑226TC Metro was introduced in 1969 and was Before departing from Bamaga at 1107 lo- replaced in 1974 by the Metro II, which has larger windows and im‑ cal time, the PIC told a ground agent that the proved systems. The SA‑227AC Metro III, introduced in 1981, has lon‑ weather was bad at Lockhart River and that they ger wings, a higher useful load and more powerful engines. Maximum might not be able to land there. takeoff weight was increased from 14,500 lb (6,577 kg) to 16,500 lb The forecast winds were from 130 degrees (7,484 kg) with the introduction of the more powerful SA-227DC Metro at 15 kt, gusting to 25 kt. The crew elected to 23 in 1990. The Merlin/Metro series was produced by Swearingen Aircraft Co., conduct the area navigation/global naviga- Fairchild Aircraft Corp. and Fairchild Dornier. Production was termi‑ tion satellite system (RNAV/GNSS) approach nated in 1999. to Runway 12, which had a minimum descent

Source: Jane’s All the World’s Aircraft altitude (MDA) of 1,040 ft — or 120 ft lower than the MDAs for the RNAV/GNSS approach to Runway 30 and the nondirectional beacon (NDB) complaints about the operational performance approach. of the PIC,” the report said. The chief pilot was The airport did not have a control tower. the managing director of Transair and also The automatic weather station at the airport served as training director and as one of the recorded only wind direction and velocity, tem- company’s two check pilots. perature and rainfall data. A meteorological ob- The copilot, 21, held a commercial pilot server performed observations three times a day license and had 655 flight hours, including 150 but did not have the capability to communicate flight hours in Metros. He had no experience in directly with pilots. The observation performed turbine aircraft or multi-pilot operations before at 1200 the day of the accident did not include being employed by Transair in March 2005. “A information on visibility or cloud bases. family member reported that the copilot was The report said that Australian Bureau of Me- given a training manual to study and was not pro- teorology estimates indicated that “the cloud base vided with any formal classroom training during was probably between 500 ft and 1,000 ft above his ground school,” the report said. His records mean sea level, and the terrain to the west of the

30 | flight safety foundation | AeroSafetyWorld | June 2007 causalfactors

aerodrome, beneath the Runway 12 RNAV/GNSS Unstabilized Approach approach, was probably obscured by cloud.” FDR data indicated that the aircraft accurately The PIC likely was the pilot flying because tracked the final approach course. However, recorded radio transmissions were made by the airspeeds and descent rates exceeded those copilot. There was no record of communication specified in the Transair Operations Manual and between the pilots because the cockpit voice those appropriate for a stabilized approach, the recorder (CVR) had malfunctioned and provided report said. The company operations manual no usable data for the last 30 minutes of the flight. did not provide specific guidance for conducting The copilot had an endorsement on his a stabilized approach. instrument rating to conduct NDB approaches, The report cited the elements of a stabi- but he was not endorsed for RNAV/GNSS lized approach recommended by Flight Safety approaches. There was no record that he had Foundation that include a maximum speed of received company-required training on the use VREF, landing reference speed, plus 20 kt and a of global positioning system (GPS) equipment maximum descent rate of 1,000 fpm.2 as the sole source of navigation information. An appropriate approach airspeed for the “The crew commenced the … RNAV/GNSS Metro under the existing conditions would approach, even though they were aware that the have been about 130 kt. FDR data indicated copilot did not have the appropriate endorse- that airspeed was about 226 kt when the aircraft ment and had limited experience to conduct this crossed the initial approach fix and about 176 kt type of instrument approach,” the report said. as it crossed the intermediate fix. The aircraft then descended from 3,500 ft to Complex Procedure 3,000 ft and remained at that altitude momen- The approach procedure was relatively complex, tarily (Figure 3, p. 32). “During this level flight, and the crew’s workload during the approach likely was very high (ASW, 2/07, p. 46). The Approach Track aircraft was not equipped with an autopilot. “There was a significant potential for [CRM] problems within the crew in high-workload LHRWG situations, given that there was a steep trans- 1139:56 cockpit authority gradient and neither pilot had 1086 LHRWE previously demonstrated a high level of CRM 124° 1140:28 5NM skills,” the report said. “A steep gradient between 5NM 1140:33 1509 6° E a dominant PIC and a submissive copilot may LHRWI 1625 result in the PIC not listening to the concerns of 124° 5NM the copilot and/or the copilot being less willing to 5NM 1142:19 1047 communicate important information to the PIC.” 1142:51 LHRWD LHRWF The report also said that the copilot’s lack of 5NM

340° training and experience in conducting RNAV/ 3.6NM GNSS approaches might have made it difficult for 124° NOT FOR NAVIGATION1787 7NM him to detect deviations during the approach. 1463 At 1139, the copilot announced on the LHRWM 3NM airport’s common traffic advisory frequency 538

(CTAF) that the Metro was over “Whiskey Golf” 0 5NM LHRWH — the LHRWG waypoint, an initial approach fix — and was inbound to “Whiskey India” Source: Australian Transport Safety Bureau — LHRWI, the intermediate fix, which was 12.5 nm from the runway threshold (Figure 2). Figure 2 www.flightsafety.org | AeroSafetyWorld | June 2007 | 31 causalfactors

the aircraft’s speed reduced to the maxi- The report said, however, that there was ­respond to the GPWS alert and warn- mum half-flap extension speed (180 no indication that the aircraft encoun- ings that were probably annunciated kt) and the flaps were extended [to half tered wind shear. during the final five seconds prior to of their travel],” the report said. “The impact,” the report said. aircraft did not descend below the seg- Two GPWS Alerts The accident likely would not ment minimum safe altitude (2,200 ft) The crew likely received two ground- have occurred if the aircraft had been during this initial descent and leveling.” proximity warning system (GPWS) equipped with a terrain awareness and Soon after the landing gear was “TERRAIN, TERRAIN” alerts. Postac- warning system (TAWS), which pro- extended, about 1.4 nm from the final cident simulations of the aircraft’s flight vides predictive terrain-hazard warn- approach fix, the aircraft began to path indicated that the first alert would ings, the report said. descend at 1,000 fpm. Airspeed was have occurred about 25 seconds before At 1143, the aircraft struck trees at about 177 kt when the aircraft crossed impact. The second alert would have 1,210 ft — about 90 ft below the crest the final approach fix. Power then was been followed by continuous “PULL of the northwest slope of South Pap, a reduced, and the descent rate increased. UP” warnings for the final five seconds heavily timbered ridge in the Iron Range Airspeed remained about 175 kt and of the flight. FDR data indicate that the National Park — about 11 km (6 nm) the average descent rate was 1,700 fpm crew did not respond to either alert. northwest of the airport. This high ter- during the last 48 seconds of the flight. However, the simulations also rain was not depicted on the approach The aircraft descended below 2,060 ft, indicated that a GPWS “TERRAIN, chart (see sidebar, p. 33). Initial impact the published minimum altitude for the TERRAIN” alert could result during occurred 850 ft below the published approach segment, soon after crossing a normal descent on final approach minimum altitude for the approach seg- the final approach fix. in aircraft with flaps in the approach ment. “The aircraft was destroyed by the “The higher-than-specified speeds configuration, even if the aircraft was impact forces and an intense, fuel-fed, and rates of descent reduced the amount established on the constant descent post-impact fire,” the report said. of time available to the crew to configure angle and/or above the segment mini- Investigators found no indication in the aircraft for the approach, accomplish mum safety altitude. The report said that the FDR data that a flight control or pow- the approach procedures and maintain GPWS alerts that occur during normal er plant problem occurred before impact. their awareness of their position on the operations increase the chances that pi- “There were no radio transmissions made approach,” the report said. lots will ignore them in other situations. by the crew on the air traffic services Turbulence was encountered during The second GPWS alert came too frequencies or the Lockhart River CTAF the last 25 seconds of the flight, which late. “There would have been insuf- indicating that there was a problem with further increased the crew’s workload. ficient time for the crew to effectively the aircraft or crew,” the report said.

Approach Profile

IF FAF MAP IAF LHRWI LHRWF LHRWM 3.49° 3500 124° 340° LHRWD 2860 Forecast LHRWE 124° cloud top LHRWG 2500 2115 2200 2060 1600 1600 MDA MAP Forecast cloud base THR 12 ELEV 46 NM FM 17 12 7 5 3.6 0 3 MAP 0.45

ELEV = elevation FAF = final approach fix FM = from IAF = initial approach fix IF = intermediate fix MAP = missed approach point MDA = minimum descent altitude NM = nautical miles THR = runway threshold

Source: Australian Transport Safety Bureau

Figure 3

32 | flight safety foundation | AeroSafetyWorld | June 2007 causalfactors

Chart Safety Factors

afety factors” related to the and approved by the Australian Civil aligned vertically with LHRWF in the design and charting of area Aviation Safety Authority (CASA) based profile view. The report said that this “Snavigation/global navigation on ICAO criteria. can cause a pilot to become confused satellite system (RNAV/GNSS) ap‑ The five-letter waypoints for these when scanning the chart. proach procedures were identified by procedures are shown on charts in up‑ Another factor specific to the the Australian Transport Safety Bureau percase. The first three letters identify Jeppesen chart is the use of the same (ATSB) in its final report on the Transair the airport, the fourth indicates the typeface and type size for waypoints Metro 23 accident.1 The report cited general direction from which the air‑ and the stepdown fixes — 5.0 NM and the importance of communicating craft travels on final approach, and the 3.6 NM — on the final approach seg‑ these factors, even though they were fifth is a standard letter that identifies ment. The report said that this results in not found to have contributed to the the purpose of the fix — for example, similar appearance of the letter “M” in accident. “I” for intermediate fix, “F” for final ap‑ the stepdown fixes and in LHRWM, and Among the cited safety factors was proach fix and “M” for missed approach could lead to misidentification of the the unique method used by Airservices point (ASW, 2/07, p. 47). Thus, the only MAP in high-workload situations. Australia to name waypoints. The variation in the waypoint names for a In addition, the stepdown fixes on report said that the similar, unpro‑ specific approach is the last letter. the Jeppesen chart are the only specific nounceable five-letter names cause “Research has (IAF) chart clutter and make it difficult for pi‑ shown that people LHRWG lots to distinguish waypoints shown on can automatically 3500 charts or displayed by on-board global (that is, instantly) (IAF) LHRWE 3500 positioning system (GPS) equipment. identify a number 1 “There was also no regulatory among letters, but (IF) LHRWI requirement for instrument ap‑ when identifying a 124° proach charts … to include colored letter among other (IAF) contours to depict terrain, as re‑ letters, identification LHRWD LHRWF 3500 5.0 NM quired by International Civil Aviation is slower,” the report to LHRWM 3.6 NM Organization (ICAO) Annex 4, to which said. “Research also 2 to LHRWM Australia had not notified a difference,” has shown that when NOT FOR NAVIGATION the report said. searching for a letter LHRWM The Transair flight crew likely in three-letter or five- used Jeppesen charts, rather than letter sequences, the Airservices Australia charts, during the time taken to detect accident flight. The report said that the the letter increases LHRWI Jeppesen chart for the RNAV/GNSS ap‑ the further its posi‑ 1.7 NM to LHRWF LHRWF 3 5.0 NM proach to Runway 12 at Lockhart River tion is moved from 3500' 124° 2860' to LHRWM [3.49°] 3.6 NM had several design aspects that “could the first letter.” 2115' to LHRWM LHRWM 2200' 1600' lead to pilot confusion or a reduction The information 2060' 1600' M in situational awareness.” Examples in‑ alignment factor MDA 4 5.0 2.0 1.4 3.6 cluded limited information on distance on the Jeppesen 12.5 7.5 0.5 0 to the missed approach point (MAP), chart resulted from 1. Similarity of waypoint names may make them more difficult to nonalignment of information on the the absence in the differentiate. plan view and profile view, the typog‑ profile view of the 2. The vertical alignment of two different waypoints may cause confusion. raphy used for waypoint names and initial approach fixes 3. Labels end with the same letter and use the same typeface and minimum segment altitudes, and the — LHRWD, LHRWE size for different information. absence of information on the offset, and LHRWG. The first 4. The lower set of numbers shows the distances to the runway, in degrees, between the final approach waypoint in the pro‑ but not to the missed approach point. The upper set of course and the runway centerline. file view is the inter‑ numbers shows the distance for each segment to the missed approach point, but not the distances from a waypoint to the As of 2005, more than 350 RNAV/ mediate fix, LHRWI. missed approach point. GNSS approach procedures had been This caused LHRWI in Source: Australian Transport Safety Bureau designed by Airservices Australia the plan view to be

www.flightsafety.org | AeroSafetyWorld | June 2007 | 33 causalfactors

indications of distance to the MAP. The than the optimal 5 nm. The stepdown designing and approving RNAV/GNSS scale at the bottom of the profile view fixes for the final approach also re‑ approaches. “There are limited options shows distances to the runway thresh‑ sulted from terrain considerations. The available to overcome these design old. The scale below the profile view on report said that these factors add to pi‑ problems,” the report said. “However, the Airservices Australia chart, on the lot workload and increase the chances the overall influence that these varia‑ other hand, shows distances to the MAP. for position confusion. tions can have needs to be considered Because of terrain northwest of Neither the Airservices Australia by CASA when evaluating and deciding the airport, the approach procedure chart nor the Jeppesen chart depicts whether to accept the approach.” is relatively complex. The final ap‑ terrain with color contours, as required — ML proach course is offset five degrees by ICAO under specific conditions, such Note from the runway centerline because of as when the final approach gradient is 1. The Australian Transport Safety a mountain northwest of the airport steeper than 3 degrees. The report said Bureau defines safety factor as “an — the 1,787-ft obstacle spot elevation that the charts provide no indication of event or condition that increases depicted on the plan view (Figure 2, the existence of high terrain under the safety risk” and one that, if repeated, p. 31). The constant descent angle is approach path, such as the ridge struck “would increase the likelihood of an 3.49 degrees, rather than the optimal by the Metro. occurrence [accident or incident] and/or the severity of the adverse 3 degrees. The distance from the final ATSB recommended that these consequences associated with an approach fix to the MAP is 7 nm, rather safety factors be considered when occurrence.”

Deficiencies Uncovered certificate and ceased operations in is inadequate” and recommended “fur- The report said that factors contribut- December 2006. ther work to address this safety issue.” ing to the accident included limitations ATSB did, however, recommend The report noted that CASA was in Transair’s safety policies and proce- improvements to government surveil- taking action to address other recom- dures, and deficiencies in regulatory lance of regular public transport opera- mendations, including the implementa- oversight of the company. tors. The report said that the Australian tion of regulations requiring regular “In particular, [Transair’s] flight Civil Aviation Safety Authority (CASA) public transport operators to provide crew training program had signifi- “did not provide sufficient guidance CRM training and to have a safety cant limitations, such as superficial or to its inspectors to enable them to effec- management system. ●

incomplete ground-based instruction tively and consistently evaluate several This article is based on Australian Transport during endorsement training, no formal key aspects of [Transair’s] management Safety Bureau Transport Safety Investigation training for new pilots in the operation- systems. These aspects included evalu- Report 200501977, “Collision With Terrain, 11 al use of [GPS] equipment, no struc- ating organizational structure and staff km NW Lockhart River Aerodrome, 7 May 2005, tured training on minimizing the risk resources, evaluating the suitability of VH‑TFU, SA227-DC (Metro 23).” The 532-page report contains illustrations and appendixes. of CFIT and no structured training in key personnel, evaluating organization- CRM (or human factors management) al change and evaluating risk manage- Notes and operating effectively in a multi- ment processes.” crew environment,” the report said. In November 2006, CASA told ATSB 1. “Transair” was the trading name for The company’s SOPs lacked clear that it was recruiting personnel with Lessbrook Proprietary Limited, which operated the accident aircraft under its air guidance on approach speeds, aircraft management and safety management operator certificate. configuration, elements of a stabilized expertise to improve its surveillance of approach and standard phraseology operators. In March 2007, CASA said 2. Flight Safety Foundation (FSF). “Killers for challenging another crewmember’s that it “has [provided] and continues in Aviation: FSF Task Force Presents Facts About Approach-and-Landing and decisions and actions. to provide substantial guidance mate- Controlled-Flight-Into-Terrain Accidents.” ATSB made no recommenda- rial in all aspects of surveillance.” ATSB Flight Safety Digest Volume 17 (November– tions regarding Transair, because the responded that it still believed that the December 1998) and Volume 18 (January– company surrendered its air operator guidance provided to inspectors “was and February 1999).

34 | flight safety foundation | AeroSafetyWorld | June 2007