National Transportation Safety Board
Washington,
D.C.
20594
Safety Recommendation
Date:
July
27,
1987
In
reply
refer
to:
A-87,-96
through
-98
Mr.
T.
Allan
McArtor
Administrator
Federal Aviation Administration
Washington,
D.C.
20591
?he
National Transportation Safety Board
has
completed
its
investigation and report
of
the
midair collision
between
Aeronaves de Mexico (Aeromexico) flight 498, a DC-9-32,
and
a
general aviation Piper PA 28-181, which occurred
over
Cerritos, California,
on
August 31, 1986. Aeromexico flight 498
was
operating under instrument flight
rules
(WR)
destined for the
Los
Angeles International Airport on
a
scheduled passenger flight from
Tijuana, Mexico. The PA-28
was
climbing under visual flight rules
(VFR)
after
departure
from Torrance, California.
The
collision occurred about
8
miles beyond
the
point
at
which
flight 498 crossed
the
boundary of
the
Los
Angeles
Terminal Control Area (TCA) near
6,500
feet
mean
sea
level. Both airplanes
fell
to
the ground within
the
city limits of
Cerritos. Five houses
were
destroyed and seven other houses were damaged by airplane
wreckage and postimpact fire. Fifty-eight passengers and
6
crewmembers on the DC-9,
the pilot
and
2 passengers on the PA-28, and 15 people on the ground
were
killed in the
accident.
-
1/
The Safety Board's investigation determined that
the
Los
Angeles Approach Arrival
Radar
Controller did not perceive the presence
of
the PA-28 and thus did not issue
a
traffic advisory to flight 498. The investigation
also
determined that the pilot
of
the
PA-28 entered
the
TCA without the required communication
or
clearance
from air traffic
control (ATC), and
that
his airplane
was
not properly equipped for
flight
in the
Los
Angeles TCA inasmuch
as
the
transponder did not have an altitude reporting (mode C)
feature. Examination of recorded data from
the
Los
Angeles
Automated Radar Terminal
System (ARTS) computer disclosed that the transponder replies from
both
airplanes were
processed by the
system
and that the appropriate symbology representing the airplanes
would have
been
displayed on the controller's scope; flight 498 would have
been
displayed
as
an analog beacon signal and
an
alphanumeric symbol with
a
full data block. Because
the
equipment setup used by
the
Los
Angeles
terminal
radar controllers intent anally
inhibited beacon signals
from
VFR aircraft, the analog beacon signal
for
the PA-26 would
not have been presented and
the
airplane would have been discernible
to
the controller
as
a
single alphanumeric symbol. It could not be established whether the symbol would have
been supplemented by
a
primary radar return.
._l--l___-
1/
-
For more information, read Aircraft Accident Report--"Aeronaves de Mexico, S.A.,
McDonnell Douglas DC-9-32, XA-JED, and Piper PA-28-181, N4891F Cerritos, California,
August 31, 1986'' (NTSB/AAR-87/07).
4479B/850-8
-2-
The
Safety Board also is continuing its investigations of three other midair collision
accidents that have occurred since the accident
at
Cerritos and that exemplify some
concerns common to those raised by the Cerritos accident.
On
January
15, 1987,
SLY
West
flight
834,
a
Swearingen Metro op
scheduled passenger commuter, collided with a privately operated Moo
Kearns, Utah.
Two
crewrnembers and
six
passengers on the Metro and two pilots aboard
the Mooney
were
killed.
The
Safety Board's investigation determined that flight
834
was
operating under IFR and was being vectored by the Salt Lake City
controller for a landing at the Salt Lake City International Airport. Flight
834
was
withi
the airspace defined by the Airport Radar Service Area (ARSA) when the colliiio
occurred.
The
pilots aboard the Mooney were involved in training and had been operating
from
an
uncontrolled airport south of Salt
Lake City. Three circumstances in this
accident were similar to the Cerritos accident:
the
pilot of the Mooney had not
communicated with the approach controller before entering the ARSA, the controller was
not aware of the presence of the Mooney when the flightpaths were converging even
though an examination of the ARTS-recorded data showed that the airplane's beacon
signal
was
processed
for
display, and the Mooney's transponder was not equipped
for
altitude reporting.
Missouri.
A
U.S. Army
U-21
airplane (Beechcraft King Air),
en
route to F
Leavenworth, Kansas,
was
operating under IFR while
level
at
7,000
feet and was be
controlled by a Kansas City approach controller.
The
U-21
collided with a Piper PA
airplane that was climbing en route under VFR after departure from Downtown Airport in
Kansas City.
The
pilot of the PAS1 had not established communication with ATC after
departure.
Three
persons were aboard each airplane
and
all were killed
i
The
second midair collision occurred on January
20,
1987,
near Independence
Unlike
the
Cerritos and Kearns accidents, this collision occurred outside
o
protected airspace where the intermix
of
VFR and
IFR
traffic is permitted and expected.
Also,
both aircraft were equipped with altitude reporting transponders and their beacon
signals had been processed by the Kansas City ARTS computer.
As
in the other accidents,
the controller did not detect the converging targets and consequently did not issue a
safety alert
or
traffic advisory.
The
other midair collision still under investigation occurred on May
1,
1987,
withi
the 20-mile outer area of the
ARSA
serving Orlando, Florida.
?he
airplanes involve
were a North American AT-6, which was returniw to Executive Airport after completin
a sky-writing session, and a Cessna
340
inbound to the Orlando Interna
Although the AT-6
was
operating under VFR, the pilot, in accordance with
procedures, had established radio contact with an Orlando approach
subsequently assigned a discrete transponder code and identified the
target.
The
AT-6 did not have mode C altitude reporting equipment.
The
Cessna
340
operating under IFR and
was
also
being controlled by the Orlando Approach Control.
pilot,
the
sole occupant of the AT-6, and
all
three occupants of the
fatally
injured.
-3-
The
Safety Board’s investigation determined that the Cessna was level
at
3,000
feet
and that the AT-6 had been cleared by the controller to descend from
10,000
to
1,500
feet.
An
analysis of the recorded ARTS data indicates
thst,
for
about
1
1/2
minutes
before the collision,
the
AT-6
was
almost directly overhead of, and following the same
ground track as, the Cessna
340.
The
examination of impact damage to the airplanes
corroborated other evidence that the collision occurred when the AT-6 descended onto the
Cessna
340.
The
preliminary analysis of ARTS data shows that the transponder beacon
replies from the AT-6 were intermittent and that
the
proximity of both airplanes caused
mutual
transponder reply interference, which affected the ARTS data processing and the
controller’s display.
The
controller did not discern the convergence of the two airplanes
and did not effect traffic separation.
Exsmination of the circumstances of the four midair collision accidents described
herein shows that:
All
of the accidents occurred in daylight visual meteorological conditions
in which the pilots and other flight crewmembers were expected to
remain vigilant and to
see
and avoid other airplanes.
All
of the accidents involved at least one airplane being operated under
IFR
by a pilot who was directly communicating with an ATC terminal
radar control (TRACON) fscility.
In
all
of the accidents, except the
Orlando collision, the conflicting airplane was being operated under VFR
and its pilot was not communicating with the ATC facility.
The
Cerritos accident occurred within the confines of
E
TCA; the Rearns
and Orlando accidents occurred within the confines of
an
ARSA; the
Independence accident was in the vicinity of,
but
outside the boundary
of, a TCA.
In both the
Cerritos and the Rearns accidents, the pilots of the VFR
rules and pilot/equipment requirements
are
imposed (TCA and ARSA
5
aircraft had entered the designated airspace wherein special operatin
without having established
the
requisite ATC communication.
All
of the airplanes involved in these accidents were equipped with
operating transponders. However, in three of the accidents, the airplane
being operated under VFR did not have mode
C
altitude reporting
capability.
In
a’l of the accidents, the transponder replies from both airplanes had
beer, processed
by
the ARTS computer and symbology representing the
aiplanes would have been displayed on the controller’s scope. However,
in the Orlando accident,
the
display was intermittent and the
convergence of flightpaths was probably not discernible.
In
the other
three accidents, flightpath convergence would have been discernible.
In
all of the accidents, the radar controllers were not aware of the
traffic conflict and no safety alerts
or
traffic advisories were issued.
Based on the above comparisons among the accidents, the primary issues of concern
exemplified
by
these
accidents are:
The
limitations of the
see
and avoid concept of collision avoidance.
-4-
The
effectiveness of ATC Terminal Radar Controllers to detect and
prevent conflicts between IFR and VFR airplanes near airports having
TCAs and ARSAs.
Future developments to prevent midair collision accidents.
Limitations of
See
And Avoid
I
Clearly, the
see
and avoid concept of collision avoidance was not in itsel
to prevent these four accidents.
In
two of the accidents (Kearns and Independence), the
convergence of the airplanes was nearly head on. Although not yet completed,
the
Safet
Board believes that these investigations might show that the pilots had margin
opportunity to see and avoid each other because of the relatively high closure
rate
of the
airplanes,
and
the small area presented by the airplanes when viewed head on, combined
with
the
human physical limitations for visual detection, recognition, and response.
In
the
Orlando accident, an analysis of the geometry of
the
converging flightpaths, combined
with the visibility envelope from the cockpit of the AT-6 and the cabin of
the
Cessna
340,
indicates that the visual detection of the other airplane by either pilot would have been
unlikely and perhaps impossible.
The
limitations in the
see
and
avoid concept of collision avoidance have long been
recognized and acknowledged by
the
Safety Board and other aviation safety advocates. It
has
also
been recognized that
the
risk of midair collisions is minimized when
all
airplanes
within
a
given volume of airspace are provided separation by an ATC radar facility. While
less effective than ATC-provided separation, midair collision avoidance is significantly
improved
when
the pilots of airplanes operating in a
see
and avoid environment are
alerted to the presence and location
of
potentially conflicting traffic.
Thus,
the issuance
of safety alerts and traffic advisories to those airplanes communicating with an ATC
radar controller and the intelligence that will be provided
by
the onboard Traffic Alert
Collision Avoidance System (TCAS) are key elements in the future prevention of midair
collision accidents.
Effectiveness of Air Traffic Control
The
Safety Board has recognized that positive control of all aircraft
in
all
airspace
is not presently practical.
In
fact, the Safety Board has recently expressed its concern
that the safety inherent in the ATC system may be derogated by
continual increases in
traffic operations within the
system.
Nonetheless,
the
circumstances of past midair
collision accidents and the analyses of near midair collision reports conclusively show that
the highest midair collision risk is associated with fliiht under VFR.
On this basis, the Safety Board repeatedly has advocated that the National Airspace
System be designed
so
that air carrier aircraft operate in an environment wherein
all
aircraft are provided ATC separation to the maximum practical extent from takeoff to
landing.
In
the
late
1960s and early
1970s,
the
Safety Board recommended establishment
of designated climb and descent corridors to join major airports with
a
stratum
of
level airspace wherein
all
aircraft would be subject to ATC. This concept was not
t
accepted but
was
partially satisfied by the establishment of the Positive Contr
(PCA), TCAs, and
ARSAs.
The
pilots of
all
aircraft in those areas must communicate
with
the
appropriate air traffic controller. Furthermore, airplanes operating in the PCA
and
in
9
of
the
23
present TCAs must be equipped with altitude reporting transponders.
However, a gap remains between the top of the TCA (which is typically at
7,000
above ground level) and
the
floor of the PCA (at
18,000
feet mean
sea
level), thr
which the air carrier aircraft
must
climb and descend. During visual meteorological
conditions, this airspace can be freely transited by aircraft that are operating under VFR
and thus are not participating in the ATC system. It is in such airspace, where a mix of
IFR and VFR aircraft exists, that the pilots of all airplanes
must
still rely heavily on the
see and avoid concept.
The segregation of air carrier aircraft operating under IFR in the PCA and TCAs has
proven to be effective in collision prevention
in
the high altitude en route environment
and in the immediate area of the major airports.
In
fact, the collision between
flight
498
and the PA-28 over Cerritos is
the
only collision of which
we
are aware that has occurred
in the protected airspace of a TCA; we are unaware of
any
collisions in the PCA.
The prevention of midair collision accidents in the PCA,
TCAs,
and
ARSAs
can be
largely attributed to the positive separation,
safety
alerts, and traffic advisories provided
by controllers to aircraft operating in these designated airspaces. 'Ihe ability of
controllers to provide these services is contingent upon their awareness of all t5e
airplanes in the contained volume of airspace as effected by the required pilot-to-
controller communication. The probability that a controller will
be
aware of those
aircraft that intrude into TCA
or
ARSA airspace is undoubtedly reduced when such
communication does not exist.
In
fact, the Safety Board believes that
a
controller might
be more inclined to overlook a potential conflict involving an unidentified VFR target
without a displayed altitude within the bounds of a TCA
or
ARSA than he
or
she would if
the conflicting targets were in airspace where a VFRAFR mix is expected. Unreported
VFR targets without a displayed altitude within the depicted horizontal boundary of a
TCA
or
ARSA are common and are normally associated with aircraft above
or
below the
altitude bounds, which thus do not threaten controlled airplanes within the TCA
or
ARSA.
The Safety Board believes that such logic might apply to the Cerritos and I<earns
accidents.
If
the theory is valid that controllers "overlook" VFR targets whose altitudes
arc unknown, the probability
of
awareness would likely increase if
all
aircraft in the
vicinity of the TCA
or
ARSA were equipped with mode C altitude reporting transponders.
'Ihe Safety Board believes that the FAA must
act
to reduce the number of
unauthorized VFR aircraft intruding into TCAs and ARSAs.
Thus,
the Safety bard
was
pleased when the FAA established a special tas!c group in September
1986
as a result of
the Cerritos accident.
L?
reviewing the design and procedures pertaining to TCAs, the
task group examined such factors as size, shape, traffic count, complexity, number and
type of
flight
infractions, procedures, past enforcement action, and general TCA
performance.
As
a result of the review, the task group made
40
recommendations, which
were subsequently consolidated into 39 action items approved by the FAA Administrator.
The sction items included the initiation
of
changes to simplify design of the TCA
bo indaries, to expand the airspace, to improve pilot education programs, to strengthen
enforcement actions against violators, and to apply more stringent requirements to pilots
and aircraft entering TCAs. Several of these actions are included in the Notice of
Proposed Rulemaking (NPRM) issued on June
11,
1987, "Terminal Control Area (TCA)
Classification and TCA Pilot and Equipment Requirements." The NPRM describes
rulemaking to simplify TCA design, to require altitude reporting transponders in aircraft
operating in all TCAs, and to enlarge the applicable volume of airspace around major
airports.
The Safety Board generally endorses these actions and recognizes that the
The Safety Board will continue to
optimization of TCA design is a complex problem.
monitor the FAA actions to achieve
an
appropriate final rule.
-6-
The
FAA actions appear, however, to address the problem
as
it
relates to TCAs
only.
The
Safety Board believes that the Kearns accident shows that the hazard of
unauthorized intrusion equally applies to ARSAs.
In
fact, the airports within ARSAs may
accommodate a greater
mix
of air carrier and general aviation aircraft than those
airports within TCAs. The Safety Board acknowledges
that
many of the action
it
pertaining to TCAs will have an inherent effect on the knowledge of pilots flying in
around ARSAs
as
well. The Safety Board believes that the FAA should impleme
actions to track, identify, and
take
appropriate enforcement action agains
intrude without the required ATC communication into ARSAs
as
well
as
Furthermore, the Safety Board believes that airplanes operating in the
A
equipped with transponders with
an
altitude reporting capability. The knowled
airplane's altitude will significantly enhance the ability of the radar eontro
and
resolve flightpath conflicts.
While acknowledging that the excellent record of midair collision prevention wit
the PCA and TCAs is a tribute to controller performance, the Saf
recognizes that controllers occasionally need help.
Thus,
the prevent
between two aircraft operating under IFR in airspace controlled
by
en route and terminal
radar control facilities is partly attributable to the enhancements in
during the
last
12
years, specifically the implementation of conflict alert
route and many terminal facility computer
systems.
Even when controllers have communicated with pilots and established the
identification of airplanes under their direct control, they occasionally miss a developing
flightpath conflict, perhaps because of distraction caused by other cont
workload. Such occurrences sometimes result in a compromise of stand
controller operational
error.
Undoubtedly, many operational
errors
compromises of separation between IFR airplanes have been prev
conflict alert feature at the control station alerted the controller to the developi
situation.
Until recently, the conflict alert logic in both en route and terminal
facilities was effective only between aircraft of which the controller was aware,
the controller had accepted a hand-off of the aircraft from an adjac
the
controller had taken specific action to enter the aircraft into the ATC traclcin
environment. The conflict alert logic is based upon
the
system'
aircraft's position, altitude,
and
predicted track, data that usually
are
determ
aircraft's transponder replies and the ATC computer.
In
the en route envir
controller can enter the aircraft's reported altitude into the
system.
Unless
the
enters the aircraft's reported altitude in the
en
route environment, and in
all
terminal environment,
the
conflict alert feature will only include airplan
transponders and the associated mode
C
altitude reporting equipment.
The
Safety Board is aware that the FAA is presently evaluating an enhan
the en route radar control computer system
in
the Houston Air Route
Traff
Center (ARTCC).
This
enhancement
incorporates th
:
additional compute
requirement and software logic changes to permit conflict alert betw
equipped with transponders having an altitude reporting function, irrespective of the
controller's previous awareness
of
the aircraft. Thus,
the
conflict alert signal wil
activate to bring
a
controller's attention to a convergence of any mode C-transponder
equipped VFR aircraft with an aircraft under his
or
her control,
in
that way permitting
positive separation
or
issuance of safetv alerts and traffic advisories.
The
Safetv Board is
-8-
to
the
ultimate installation of the AAS. Nevertheless, the Safety Board b
risk
of midair collisions in the terminal area will increase with projected increa
traffic and that such interim
measures
must
be taken promptly if catastr.ophic ace
are
to
be prevented during the
next
10
to
12
years.
upon
an
airplane's operating mode C altitude reporting transpo
believes that transponders with mode C should be required for
all
with those air carrier aircraft
for
which TCAS will be required, p
installed aboard air carrier aircraft. This could be largely acc
requirement for this equipment
as
a requisite for operations near airports served b
air carriers,
in
particular around those airports protected by TCAs and ARSAs.
Finally, the Safety Board believes that the circumstances
are probably
rare,
that
is, the occurrence of mutual interference of transpond
period of time long enough to mask a developing collision conflict.
accident illustrates the serious consequences of this phenomenon.
believes that this problem will
be
solved
when
the Discrete
transponder is introduced. The Safety Board realizes
that
the FAA
is
moving forward
with the implementation of mode
S
and will continue to monitor the progress of that
program.
activities, the Safety Board reiterates
the
following recommendations
to
the FAA:
Since airborne TCAS and ground based conflict alert
systems
are
b
Therefore,
as a result of these accidents and
a
review of the FAA's
A-85-64
Expedite the development, operational evaluation, and final certificati
of the Traffic Alert and Collision Avoidance System (TCAS) for
installation and
use
in certificated air carrier aircraft. (A-85-64)
A-85-65
Amend
14
CFR Parts
121
and 135 to require the installati
Traffic Alert and Collision Avoidance System (TCAS) equipment
i
certificated air carrier aircraft when it becomes available fo
operational
use.
In
addition, the Safety Board recommends that the FAA:
Implement procedures to track, identify, and take
enforcement action against pilots who intrude into Airport Radar Servic
Areas
(ARSAs)
without the required Air 'Raffic
communications.
(CWs
II,
Priority Action) (A-87-96)
Require transponder equibment with mode C altitud
operations around
all
Terminal Control
Areas
(TCAs)
a
Radar Service Areas (ARSAs) after a specified
d
implementation of Traffic Alert and Collision Avoidan
requirements
for
air carrier aircraft. (ClassIII, Lon
(A-87-97)
-9-
Take
e
edited action to add visual flight rules conflict alert (mode C
intruder
xe
logic to Automated Radar Terminal System (ARTS) computers
as
an
interim measure to the ultimate implementation of the Advanced
Automation System (AAS). (Class
III,
Longer Term Action) (A-87-98)
BURNETT, Chairman, GQLDMAN, Vice Chairman, and LAIJBER, NALL,
and
KOLSTAD,
Members, concurred in these recommendations.
V