1. Vacuum pump failure during flight resulted in the loss of suction to the air-driven gyroscopic instruments.
2. Erroneous attitude and heading indications from these inoperative flight instruments.
3. Dark sky conditions with no discernible horizon.
4. The pilot had limited recent experience flying in conditions of reduced external visual reference.
5. Inability of the pilot to control the aircraft by sole reference to the remaining flight instruments.

The circumstances of this accident were consistent with a loss of control by the pilot at night, resulting from inoperative attitude and directional indicators. The combination of a dark night, high level cloud, and limited ground lights would have provided the pilot with few external visual cues. This would have required him to quickly modify his instrument scan to allow him to control the aircraft by sole reference to the remaining flight instruments. One of those instruments, the electric turn coordinator, was probably the 'electric backups' that the pilot referred to in his call to FS.

To achieve a desired flight performance, the aircraft is placed in a particular attitude, together with an appropriate power setting. Precise attitude information can either be gained by reference to the natural horizon, or to a gyro-stabilised attitude indicator, when external indications are either not available or are unreliable. The altimeter, air speed indicator and vertical speed indicator can, in combination, also provide limited attitude information.

The loss of vacuum to the air-driven gyroscopic flight instruments would have resulted in those instruments providing erroneous aircraft attitude and heading indications to the pilot. It is possible that the pilot did not mask the failed attitude and directional indicators. Consequently, the pilot may have inadvertently continued to respond, however briefly, to the erroneous indications from the failed instruments. His instrument scan proficiency, with the attitude indicator as the primary focus, would have been developed over several years. Such a scan could not easily have been modified to ignore the very powerful stimuli from erroneous attitude indications.

The pilot was dependent on alternative instruments for aircraft attitude information and it is likely that while attempting to control the aircraft as well as calculate flight time and distance to Mildura, the pilot became spatially disorientated and lost control.

The following factors were determined to have contributed to the accident:

1. The glare of the sun shining through the aircraft's windscreen may have partly obscured the tree from the pilot's field of vision.
2. The use of the Satloc light bar indicator may have resulted in the pilot focussing his eyes, for at least some of the time, on a close object causing distant objects, such as the tree, to be out of focus and thus more difficult to see.
3. The failure of the cockpit truss to remain in position when the aircraft struck the ground inverted made the accident unsurvivable.

Manoeuvring the aircraft to commence the next swath run was a high workload period for the pilot, requiring precise flying to position the aircraft accurately in both lateral and vertical planes. As the pilot was lining the aircraft up to conduct a swath run in a westerly direction in the late afternoon, the sunlight coming through the windscreen would have created glare in the pilot's field of vision, partly obscuring the tree. When using the Satloc system's light bar indicator to align the aircraft with the line of the next swath run, the pilot's vision would have been focussed, for at least some of the time, on the light bar. When the pilot's eyes were focussed on the light bar, more distant objects, such as the tree, would not have been in focus and would therefore have been harder to see. Although the pilot must have been aware of the tree's presence, the high workload, the glare from the setting sun and the pilot focussing on the light bar indicator all contributed to the pilot temporarily being unaware that the tree infringed his intended flightpath.

The damage to the left wing when it struck the tree would have resulted in a loss of lift and an increase in drag from this wing. This would have caused the aircraft to roll rapidly to the left and, as the ailerons had jammed, the pilot would have been unable to stop the roll. Thus, after impact with the tree, the aircraft was most likely uncontrollable.

When the cockpit truss was tom from the aircraft, the space remaining between the ground and the pilot's seat was too small for the accident to have been survivable.

The witness description of the aircraft's behaviour, along with the evidence at the impact site, indicated that the aircraft probably stalled aerodynamically as the pilot reversed the direction of the procedure turn. It is likely that the aircraft was in a nose-high attitude at the time. This would explain the apparent low forward speed of the aircraft at impact. Assuming that the pilot was conducting the turns at typical altitudes, it is unlikely that there would have been sufficient height available for the pilot to recover the aircraft to normal flight. The pilot's low experience level on turbine powered Air Tractor aircraft compared with piston engine powered models may have contributed to the loss of control.

Details obtained during the investigation did not allow any conclusions to be drawn concerning the security of the pilot's safety harness at impact.

After arriving at the property strip, the pilot was briefed on the spray area by the property owner. In the meantime, the aircraft was loaded with 1,500 litres of spray solution. The pilot subsequently boarded the aircraft and was observed to secure his safety harness, and don gloves and a helmet, before departing for the task.

The spray area was about 800 m from the strip. Witnesses at the strip observed the aircraft as it began spraying and noted that the procedure turns at the end of each run were conducted at a consistent height. The turns were flown initially to the left to offset the aircraft from the previous run, and then reversed to align the aircraft for the next run. About 30 minutes after spraying had commenced, one of the witnesses observed the aircraft rolling right during a turn reversal in a procedure turn. The aircraft continued rolling until it was in an inverted attitude, and then descended into the ground where it immediately caught fire.

1. A significant number of pilots were operating on the flight service frequency, which made it difficult for crews to report their positions or exchange information for traffic avoidance.
2. The crew of the DHC-8 did not report their intention to change flight level to the flight service officer.
3. The crew of the C130 did not report the Mount Sandon position to the flight service officer.
4. The crew of the C130 was unable to contact the crew of the DHC-8 to arrange separation between the aircraft, due to frequency congestion.
5. The crew of the C130 did not report that the aircraft was maintaining FL130 to the radar controller after previously reporting leaving that level.
6. The flight service officer did not provide directed traffic information in accordance with MATS.
7. The flight service officer assumed that the DHC-8 had entered controlled airspace and subsequently provided unverified information to the crew of the C130.
8. The crew of the C130 received incorrect traffic information on the DHC-8.

The investigation revealed that a combination of traffic disposition and the use of retransmission on the FS3 frequencies created a congested radio environment for the FS3 officer. As a consequence, the provision of reports by pilots and the provision of traffic information by the officer were constrained.

Following unsuccessful attempts to obtain a clearance to enter controlled airspace from the radar controller, the crew of the DHC-8 were also unable to report to FS3 that the aircraft was maintaining FL125 due to the congestion on the frequency. Similarly, the crew of the C130 was probably unable to report the Mount Sandon position to FS3 after changing to that frequency. Provision of a position report by the crew of the C130 or notification of the change in level by the crew of the DHC-8 would have assisted the FS3 officer to better appreciate the disposition of the aircraft.

The radar controller had advised the FS3 coordinator that the DHC-8 was in non-controlled airspace at FL125 and was traffic for the C130. However, this information was provided at a time when the FS3 operator was busy with transmissions from other pilots. Despite the coordinator's efforts, he was unable to advise the FS3 officer before the latter advised the crew of the C130 that the DHC-8 was not significant traffic. The FS3 operator should have verified the situation with the radar controller or alternatively advised the crew that the position of the DHC-8 was unknown.

The crew of the C130 were aware of the oncoming DHC-8 and had remained at FL130 for a further 3 minutes while attempting to report to FS and to contact the crew of the DHC-8 to arrange separation. Once the FS3 officer told them that they could disregard the DHC-8, the crew of the C130 descended the aircraft and unknowingly placed it in conflict.

While still at FL130 in controlled airspace, the crew of the C130 was having difficulties communicating with FS3 and the crew of the DHC-8. Had the crew of the C130 advised the radar controller of their intention to maintain FL130, it is likely that the controller would have been prompted to provide greater assistance to resolve the conflict.

1. The ADSO did not send a hard copy departure message on the B737's flight.
2. The ADSO was a trainee and was not adequately supervised.
3. The Sector 1 handover/takeover involved poor inter-controller communication and coordination.
4. The oncoming Sector 1 controller did not pass coordination of the B737's TAMOD position report to the Perth Outer controller.
5. There was a lack of appropriate procedures in relation to the departure processing task and the handover/takeover task.

Although there was no breakdown in separation, this incident involved a serious breakdown in coordination that resulted in a B737 not being under effective air traffic control from 0858 EST until after the radar return was identified at 0947. Had the Perth controller not detected the problem, there would have been a breakdown of separation. Although the two aircraft would have both been to the east of TAMOD at the time of passing, and therefore on different routes, this situation was not planned.

The incident directly resulted from a combination of active failures. The two most significant errors were: (a) the omission of a hard copy departure message via the AFTN; and (b) the omission of any coordination on the aircraft from the Melbourne Sector 1 controller.

Processing the Hard Copy Departure Message

Had the trainee ADSO sent the departure message via the AFTN, Perth control would have activated the flight progress strips and, therefore, been better prepared to detect the missing coordination on an aircraft from Melbourne Sector 1. This error consequently removed an important safety defence.

Factors that contributed to the missing AFTN message included:

1. the aircraft movement advice form being incomplete, and therefore not providing an appropriate cue; high workload;
2. insufficiently detailed task procedures; and
3. a lack of direct supervision of the trainee ADSO.

It is likely that the supervising ADSO became too involved in the operation of departure processing tasks to be able to effectively monitor the trainee's performance.

Passing Coordination of the TAPAX Position Report

The omission of the oncoming Sector 1 controller to pass coordination on the B737 to the Perth outer controller meant that the Perth controller had no forewarning of the arrival of that aircraft into his airspace. Therefore, this error increased the likelihood of a breakdown of separation standards. Although the Perth controller did detect the problem in sufficient time to avoid a breakdown in separation, it is likely that such a detection would not usually occur.

Factors that contributed to the omission included the absence of clear communication between the departing Sector 1 controller and the oncoming controller as to who would pass the coordination, and the inability of the departing controller to remain available to ensure the oncoming controller was fully briefed. More specific handover/takeover procedures could have reduced the likelihood of these factors. The development of appropriate teamwork and team resource management programs would also help reduce the likelihood of such inter-controller coordination and communication failures.

The coordination omission was also partly a result of the inherent nature of the task. It was probably an action slip, or an error where the controller's actions did not proceed as planned. More specifically, the error appeared to be a substitution of one highly automatic task (asking the B737's flight crew to transfer frequency at TAMOD) for another (passing coordination on the B737 crew's TAPAX report to the Perth Outer controller). Such substitution slips are usually associated with a period of inattention or distraction. In this case, the controller may have been distracted by the call from another flight crew immediately before she performed the task. As the controller had only been working on the sector for a few minutes, she could also have been distracted by her scanning and familiarisation activities.

An increase in the likelihood of skill-based errors is often correlated with fatigue. In this case, the controller had a less than normal amount of sleep in the 2 nights prior to the incident. However, the investigation was unable to determine if fatigue influenced the controller's performance at the time of the occurrence.

There were cues available to alert the Sector 1 controller to the error, such as the times on the TAMOD strip and the absence of a tick in the last box of the TAPAX strip. However, having developed the intention to conduct a frequency-transfer task, it is unlikely that the controller would have considered checking times on a strip. In addition, research has shown that the absence of a cue (such as a tick) is often not detected. Shortly after the error occurred, the controller disposed of the flight strips and effectively removed any remaining cues she could use to detect the error.

With the introduction of The Australian Advanced Air Traffic System (TAAATS) throughout Australia in 1998 and 1999, the likelihood of this specific type of error should be reduced. Transfers of aircraft between sectors will be conducted automatically. In addition, for procedural control sectors such as Sector 1, the spatial positions of aircraft will be pictorially displayed. It is reasonable to expect that these changes will reduce the likelihood of a controller incorrectly perceiving the position of an aircraft.

Flight Levels

The westbound A320, VH-HYA, was maintaining a non-standard flight level (FL370) which resulted in an eastbound A320, VH-HYR, being given a non-standard level (FL350) for the initial level clearance. This consequential action had the effect of placing VH-HYR in direct conflict with the B737. Had standard levels been applied on the two-way route system that was under procedural control, a safety net would have been put in place. This net would have become prominent had the Perth controller not observed the radar paint of the B737.

At 0627 EST, a Boeing 737 (B737) departed Melbourne for Perth. Melbourne Sector Inner West advised Melbourne centre of the departure. An airways data systems operator (ADSO) received this information and notified Adelaide centre of the departure by intercom. He then activated the flight strips for Melbourne Sector 1 and delivered these strips to this sector. A hard copy departure message should also have been sent via the Aeronautical Fixed Telecommunications Network (AFTN) to all units affected by the flight, but this task was not completed. As a result, Perth Air Traffic Control did not receive any notification that the aircraft had departed.

For much of its flight to Perth, the B737 was on route L513 and was under the control of Melbourne Sector 1, a procedural control sector. This sector was combined with Sector 5 until approximately 0850 when it was decided to split the two sectors. The departing controller proceeded to hand over control of Sector 1 to the oncoming controller, but kept control of Sector 5.

At 0853 EST, during the Sector 1 handover/takeover, the crew of the B737 reported having passed TAPAX (a position reporting point 523 NM east of Perth) at 0852. They also reported that they were maintaining FL350, and estimating TAMOD (a position reporting point 153 NM east of Perth) at 0951. The departing controller took the position report and appropriately marked the flight strips. The report was required to be passed to the Perth Outer controller, as the sector boundary was 10 NM east of TAMOD. This coordination task was not done.

The departing Sector 1 controller later reported that he did not perform the coordination task as he assumed that the oncoming controller would do it. He also reported that he did not specifically point out the need for the oncoming controller to perform this task, but assumed that she had heard the position report and understood that it needed to be done. The oncoming controller later reported that she could not remember hearing the position report. Neither controller could recall whether the relevant flight strip had been cocked on the flight progress board to indicate that there was an outstanding task to be performed.

The oncoming controller took control of Sector 1 at 0856. Soon after taking over, Perth control contacted her to advise that an eastbound Airbus 320 (A320), VH-HYJ, was about to transfer to her frequency and that the crew would soon contact her. This crew contacted her at 0857. At 0858, immediately after taking this crew's report, the controller contacted the westbound B737's crew and advised them to contact Perth at TAMOD. This frequency-transfer task was normally performed when the aircraft was 5 minutes from TAMOD, or 0946 in this case. Shortly after performing this task, the controller removed the B737's flight strips from the flight progress board. This action was normal practice for filing the flight strips after an aircraft had left the Sector 1 area of responsibility.

At 0929 EST, another eastbound A320, VH-HYR, departed from Perth for Melbourne. The aircraft was planned on route L513 until TAMOD, before turning onto the one-way route Y53. The estimate for TAMOD was 0953. The Perth controller informed Sector 1 of the departure. He advised that the aircraft had planned FL370 but that he recommended restricting the aircraft to non-standard FL350 due to a westbound A320, VH-HYA, at non-standard FL370 and estimating TAMOD at 1006. The Sector 1 controller agreed with this restriction.

Although having the air traffic under his jurisdiction procedurally separated, the Perth Outer controller checked the radar at 0941 and observed a return approximately 220 NM east of Perth (67 NM east of TAMOD). As the displayed information was based on extreme range returns, it was considered unreliable. However, it indicated a westbound jet aircraft at FL350. The controller checked his coordinated and pending traffic, but was unable to identify the aircraft.

At 0943, Perth Outer contacted Melbourne Sector 1 to report the return and ask whether she knew its identity. The Sector 1 controller was not aware of any aircraft in that area other than HYA, the westbound A320 that was 15 minutes behind the position of the unknown aircraft. Perth suggested restricting HYR to FL330 until the problem could be rectified. Another Perth controller advised HYR's crew of the problem. At 0946, Perth Outer directed the crew to divert to the left of track, as the observed radar return's altitude was still unverified.

With the assistance of other controllers, the Sector 1 controller retrieved the used flight strips and identified the return as the B737. The relevant information was reported to Perth Outer at 0947. The aircraft was then transferred to the Perth controller.

As the A320 had been restricted to FL330 and diverted left of track, there was no breakdown in separation. The two aircraft passed each other at 0950:01. HYR passed TAMOD at 0949, and the B737 reached TAMOD at 0951. Analysis of the radar tape indicated that HYR would have reached FL350 between 0949:20 and 0950:20. The B737 was fitted with a Traffic Alert and Collision Avoidance System.

Departure Message Processing

After receiving the B737's departure notification from Melbourne Sector Inner West, the ADSO annotated the callsign, departure point, destination and departure time on an aircraft movement advice form before advising Adelaide of the departure. However, this information was not entered in the appropriate boxes on the form and the SSR code was not entered. In addition, various unnecessary items were written on the form, including the numbers of four AFTN messages that had been sent around the time that the B737's departure was being processed.

The ADSO who took the departure notification was a trainee. He was required to be directly supervised during his task performance, but this supervision was not present at the time the departure was being processed. It was later reported that workload was high at the time of the departure, and that the trainee and the supervising ADSO were both performing departure processing tasks.

Prior to the incident, the tasks required to process a departure were specified in written procedures, but they were not all specified in the same procedural documentation. The order in which the tasks should be performed was not specified in any procedural documentation.

Sector 1 Handover/Takeover

Sector 5 was in the process of being transferred from the Bight Group, which also had Sectors 1 and 4, to the Desert Group. A console for Sector 5 had been set up in the Desert Group's area, and the controllers in that group were currently being trained in Sector 5 operations. Until the transfer was completed, Sector 5 could also be operated from Sector 1's console.

Immediately after the oncoming controller took over Sector 1, the departing controller took the flight strips for Sector 5 and arranged them on the relevant flight progress board in the Desert Group. He then proceeded to conduct a training session on Sector 5 for another controller.

It is desirable for a controller to perform all outstanding tasks prior to handing over to another controller, but this is not always practical. In this case, several tasks had to be conducted during the handover/takeover and the workload level was significant. Both the departing and the oncoming controller reported that the handover of Sector 1 appeared to be well conducted. Prior to the incident, there was no written procedure that detailed all of the required tasks to be performed during a handover/takeover.

Sector Boundary

The boundary between Melbourne Sector 1 and Perth Outer was approximately 10 NM east of the reporting point TAMOD (163 NM east of Perth). The range of the relevant radar was typically 220 NM to the east of Perth between FL330 and FL370. The Perth controller was therefore able to see to approximately 50 or 60 NM east of TAMOD at high flight levels. The air route structure had been redesigned with the intention of the Perth radar being used to its full potential, but the sector boundary had not been changed.

Personnel Information

The oncoming Sector 1 controller commenced duty at 0700 on the morning of the incident. She finished duty on her previous shift at 2030 the previous night. As she lived 90 minutes from her place of work, she had only slept 5 hours during the night before the incident. Due to other factors, she had slept even less during the previous night.

In addition to not passing coordination on the B737, the oncoming controller made three minor errors during the period after she took over Sector 1 until the Perth Outer controller detected the incident. These errors were an attempt to call HYA on the wrong frequency, and two occasions of contacting the wrong Adelaide controller (as she forgot that the relevant Adelaide sectors were combined). There were no other problems noted with any aspect of her performance or behaviour during this period.

Neither of the Sector 1 controllers had received any training in the use of teamwork or team resource management skills.

1. The pilot used incorrect fuel planning data.
2. The pilot did not fully understand the effects on fuel consumption of altitude and engine power settings.
3. The engine failed because of fuel exhaustion.

Air Traffic Control services

Sector 3 was responsible for the provision of air traffic control in that part of the Bangkok FIR to the south of Bangkok, and for associated portions of the Phnom Penh and Ho Chi Minh FIR'S. The Sector 3 console had three positions: a radar control position with a radar display located on the left, a procedural control position in the centre, and a radar assistant position on the right. The radar controller was responsible for separating all aircraft in the sector. This included aircraft outside radar coverage. The Sector 3 procedural controller was responsible for issuing clearances using procedural control, and assisting the radar controller with flight progress strip (FPS) marking and coordination. The procedural controller was required to notify the radar controller of any changes that would affect procedural separation.

The radar controller was the only position with facilities to transmit and receive on the control frequency of 135.5 MHz. The procedural controller could monitor the frequency, but required the radar controller to transmit control instructions to aircraft operating under procedural control.

The console had been originally designed for operations utilising a flight data processor for aircraft flight information, without the need for flight progress strips. As a result, the radar operator and assistant radar positions were not fitted with flight progress strip bays. The flight data processor had subsequently proved to be less effective than planned and the sector had reverted to using flight progress strips. This required the radar and the radar assistant consoles to be fitted with temporary strip holders.

A high level of interaction and cooperation was required between the radar and procedural controllers to effectively manage the sector's airspace. The flight progress strips for each aircraft were required to be retained in the procedural display until the crew reported at the next position. This was to enable controllers to observe that an aircraft was in transit between the previous and next positions. However, due to limited space to display the strips, the Sector controllers had developed a habit of removing flight progress strips at the earliest opportunity to make space for new strips.

The strip for QFI6, annotated for the portion of the route from ALGOR to KABAS, did not include an estimate for KATKI. There were two strips for KAL362: one annotated for the portion of the route before KANTO, and one for KANTO to SINMA. The KANTO strip for KAL362 was located in the same bay as the KABAS strip for QF16. The radar controller removed the KAL362 KANTO strip from the bay after the crew reported at that position, prior to the SINMA position.

When the crew of KAL362 requested climb from FL270 to FL290, the radar controller scanned the flight progress strips at the procedural position. There was a SINMA strip for KAL362, but no other strip to indicate a possible conflict. SINMA was located east of the route of QF16. The procedural controller was carrying out coordination duties at the time, and did not hear the climb request. The radar controller then issued a clearance for the crew of KAL362 to climb to FL290, without consulting with the procedural controller.

Crew awareness

The crew of QFI6 were using the same control frequency as KAL362 when the crew of that aircraft reported at KANTO, and were subsequently approved to climb to FL290. There were reports of radio interference on that frequency, including reports of interference from aircraft operating at lower levels. The crew of QF16 did not recall hearing the crew of KAL362 request climb to FL290, nor the amended clearance and readback.

1. The design of the Sector 3 console did not allow for all relevant flight progress strips to be displayed.
2. The radar controller removed the KANTO flight progress strip after the KAL362 crew reported at that position.
3. The procedural controller did not hear the request for climb to FL290 by the crew of KAL362.
4. There was no requirement for the display of KATKI on flight progress strips for all aircraft using the intersecting routes.
5. The radar controller did not consult with the procedural controller prior to instructing the crew of KAL362 to climb to FL290.
6. The radar controller did not ensure that vertical separation was maintained between KAL362 and QF16 while the aircraft were in an area of conflict
7. The crew of QF16 did not hear the request and subsequent approval for KAL362 to climb to FL290.

The procedural controller was responsible for issuing clearances to aircraft under procedural control, as was the case in this event. The role of the radar controller was to pass on the clearance to the aircraft. By not consulting with the procedural controller, the radar controller bypassed the established system of control, leading to a breakdown in safety.

The configuration of the Sector 3 console provided insufficient space to adequately display all relevant flight progress strips. As a result, controllers had developed the habit of removing strips at the earliest opportunity, thereby creating the potential for vital information to be missed.

The KANTO flight progress strip for KAL362 should have been retained on the procedural board until the crew reported at SINMA, the next position. The removal of the KANTO strip by the radar controller removed the only reminder available to all controllers that the intended tracks of KAL362 and QFI6 would cross.

Inclusion of the KATKI position on all flight progress strips for aircraft using the intersecting routes would have enabled controllers to more readily assess separation requirements in the procedural airspace. If the strips had required the KATKI position it is probable that the details for QFI6 and KAL362 would have been displayed under the same designator on the board, allowing controllers to recognise the potential conflict.

The inability to monitor the control frequency while conducting coordination reduced the likelihood of the procedural controller maintaining a complete appreciation of the disposition of traffic.

The reason why the crew of QFI6 did not hear the transmissions regarding KAL362 climbing to the same level, while operating on the same frequency within direct line of sight, could not be determined.

QFI6, a Boeing 747, had departed Bangkok for Melbourne and was tracking southbound on airway G463 at flight level (FL) 290. The aircraft was in contact with Bangkok Area Control Centre (BKK ACC) Sector 3 on 135.5 MHz. Sector 3 was a combined radar and procedural control sector. At 0212:54 QF16 reported passing ALGOR at FL290, estimating KABAS, the flight information region (FIR) boundary, at 0221. Just prior to reaching KABAS, the aircraft would pass the intersection of G463 and B219 at KATKI. These positions were all located beyond radar coverage, over international waters, within the procedural control portion of BKK ACC Sector 3 airspace.

A Korean registered Boeing 747, KAL362, had departed Kuala Lumpur for Seoul, tracking via B219 at FL270. Approaching KANTO, located to the west of KATKI, the aircraft was transferred to the BKK ACC. The crew of KAL362 contacted Bangkok Sector 3 on 135.5 MHz and reported passing KANTO at FL270, estimating KATKI at 0219, and requesting climb to FL290. The next reporting position was SINMA, to the east of KATKI. At 0217:20 Bangkok Sector 3 cleared KAL362 to climb to FL290. KAL362 reported leaving FL270 for FL290. At 0220:21 the pilot in command of QF16 advised the Sector 3 controller of having received a traffic alert and collision avoidance system (TCAS) traffic advisory (TA), and that the aircraft had been climbed to FL300 to avoid a collision with KAL362, but was now descending to FL290.

The crew of QF16 had received a TCAS TA, followed by a resolution advisory (RA) commanding a climb to avoid climbing traffic some 800 ft below. Each crew sighted the other aircraft. The KAL362 crew also received a TCAS TA, followed by a TCAS RA commanding a descent. The crew of KAL362 did not report a traffic confliction.

An investigation carried out by Aeronautical Radio of Thailand Ltd, the Thai air traffic control organisation, indicated that KAL362 was incorrectly given a clearance to climb to FL290 by the Bangkok Sector 3 controller, and that the crews of both QF16 and KAL362 were acting in accordance with the clearances issued to them. The minimum required distance between the aircraft was 1,000 ft vertical separation, or a lateral separation of not less than 15 minutes between their estimated times of arrival at KATKI, the intersection of their intended flight routes.

In accordance with ICAO Aircraft Accident and Incident Investigation Annex 13, paragraph 5.3, the circumstances of this occurrence provide for the State of Registry to institute and conduct any necessary investigation. After consultation with the Korean authorities it was agreed that BASI would take responsibility for the ongoing investigation.

Witness reports indicate that the pilot had established the aircraft below the cloud layer some distance from Jerilderie. The most likely reason for this was in order to make a visual approach to Jerilderie. However, in the area of the cold front, instrument meteorological conditions extended from the base of the cloud layer to the ground. With visibility rapidly reducing, the pilot had the option of continuing straight ahead and climbing in instrument meteorological conditions to the lower safe altitude of 3,300 ft or turning back to remain in visual conditions.

The pilot initiated a left turn and entered instrument meteorological conditions close to the ground. The turn would have required the pilot to fly with sole reference to flight instruments. The transition from visual flight to flight by sole reference to instruments may take several seconds, and an aircraft in a turn will generally loose altitude until the pilot takes corrective action. However, as the aircraft was at low altitude when it began the turn, it probably struck the ground before the pilot had completed his transition to instrument flight.

Although the pilot held an instrument rating, he had minimal experience of flight in actual instrument meteorological conditions. He may not have experienced the conditions that confronted him on this occasion. The combination of rapidly deteriorating weather, gusting winds and low altitude in conjunction with low experience and recency with actual flight in instrument meteorological conditions, would have required a high level of skill and experience.

On the morning of the accident, the pilot reported for duty between 0630 and 0645 in order to have his aircraft prepared by 0700 for the day's task. The aircraft had been refuelled the night before, as was the company standard practice. After loading the freight onto the aircraft, he departed Wagga for Jerilderie at 0804 with an estimate for Jerilderie of 0845. The aircraft had full fuel and 115 kg of freight. The flight was a standard company "Run 46". The flight plan indicated that the flight was to be conducted in the IFR category at 4,000 ft with a minimum safe altitude for the route to be flown of 3,300 ft.

At 0838 the pilot contacted Melbourne Flight Service (FS) and advised that he was 15 miles east of Jerilderie and would commence descent in one minute. FS responded that there was no reported IFR traffic. There was nothing further heard from the pilot after he acknowledged this transmission.

An eyewitness working near the maintenance shed on the property Bungoona, under the flight path and north of the accident site, observed the aircraft at an estimated altitude of 100 m (328 ft). The witness described the weather conditions as misty rain, like a very thick fog or low cloud. This observation is consistent with other witnesses in the vicinity of the accident site who heard but did not see the aircraft. It was observed to bank to the left and the witness lost sight of it as it passed behind the shed. He heard an explosion, ran to the other side of the shed and saw a ball of smoke. The time of the aircraft impact was approximately 0843.

Examination of the accident site and wreckage found that the aircraft had initially struck a small tree, before impacting the ground in a left turn with 13 degrees angle of bank and 52 degrees nose down. The aircraft broke up initiating a fire that consumed the fuel and much of the wreckage and freight. The centreline of the wreckage splay was 061 degrees magnetic, and the wreckage trail was approximately 60 m in length, starting from the impact crater.

Examination of the wreckage did not find any pre-existing defects or malfunctions that would have precluded other than normal operation. The maintenance records showed that the aircraft had completed periodic maintenance on 9 June 1997. There were no outstanding maintenance action items at the time of the accident. The aircraft was due to undergo scheduled maintenance later in the week of the accident. Impact marks on the propeller and the nature of the blade damage indicated that the engine was operating under power at impact.

The pilot was the holder of a Commercial Pilots Licence, and he was appropriately qualified for the flight. He was issued with an initial instrument rating on 15 December 1996 for single engine aircraft. He subsequently gained a Multi Engine Command Instrument Rating on 17 June 1997. He had accrued approximately 563 hours total flight time, with approximately 58 hours flight time using instruments as a sole reference. This information was obtained from the company computer "Duty Flight Times" log, as the pilot's logbook was destroyed in the accident. There were no entries for hours flown since 17 August 1997.

The specialist Bureau of Meteorology report, advised that conditions at the time of the accident were a cold front passing through the Deniliquin/Jerilderie district. The frontal change had passed through Jerilderie at 0843 hours EST and would have been approaching the accident site. The wind was 270 degrees magnetic, 15 gusting to 25 knots with some localised moderate turbulence associated with the passage of the change. The visibility was 4,000 m reducing to 2,000 m with moderate showers and light rain. The temperature was 8 degrees C and the barometric pressure was 1020 hPa.

The calculated aircraft weight at take-off, was 1477 kg with a maximum allowable weight of 1724 kg. The aircraft was within the weight and centre of gravity limits at the time of the accident.

Jerilderie aerodrome does not have a published instrument approach procedure. If the pilot was unable to make an approach in visual conditions, the company policy was for the pilot to proceed to Deniliquin.