CAT II and CAT III Approaches: What Changes on the Flight Deck?

A commercial airliner on final approach in low visibility, illuminated by runway lights, with the cockpit instruments displaying radio altimeter readings for a CAT IIIb landing.

Ask ten airline pilots what “CAT IIIb, no DH” means, and the answers range from “you land no matter what” to “this is not allowed in Europe”. None of these answers is true. The category scale is an accounting device, not a scale of bravery: it records the lowest altitude at which a visual decision is still required, and the poorest visibility in which the operation has been demonstrated as functional. Each of these figures has an owner, and on any given day, only one of the four owners actually limits you.

The scale describes a consequence, not a level of competence.

A precision approach category is a property of an operation, never of an aircraft, a runway, or a crew alone: it labels the combination that the three allow simultaneously. Annex 6 of ICAO defines each step with two figures and nothing more: a decision height (DH) and a runway visual range (RVR) — an alternative visibility surviving only in CAT I. Everything else — ground equipment, automatic landing architecture, training, currency — exists to make these two figures defensible.

Two structural differences separate CAT I from everything below. A CAT I decision altitude is flown on the barometric altimeter, while a CAT II or CAT III decision height is a radio height: it is read on the radio altimeter, which measures the height above the terrain immediately below the aircraft, so the published figure for a runway is corrected for the approach terrain profile rather than assuming a flat terrain behind the threshold. And from CAT II downward, the operation ceases to be generic: it requires a specific approval attached to the operator, the type, and each individual pilot.

The precision approach scale as framed by Annex 6 of ICAO, with the values commonly applied in European approvals.
Category. Decision height. RVR. What the decision is based on.
CAT I. Not below 200 feet, flown as a barometric decision altitude DA/H. Not below 550 meters, or visibility not below 800 meters. Visual reference acquired at DA/H.
CAT II. Below 200 feet, not below 100 feet, radio height. Not below 300 meters. A defined visual segment — classically three consecutive lights of the approach lights, the touchdown zone, the runway centreline or the edge lights.
CAT IIIa. Below 100 feet, or none. ICAO floor 175 m; European approvals commonly cite 200 m. Sufficient visual reference to monitor the landing, or system status alone when no DH applies.
CAT IIIb. Below 50 feet, or none. ICAO floor 50 m; European approvals typically reach around 75 m. System status, with or without a visual window depending on the approval.
CAT IIIc. None. None. Long on paper, authorized nowhere in practice — nothing guides the aircraft off the runway afterward.

CAT IIIc sums up the entire system in one line. It is not held back because the aircraft cannot handle the landing — a fail-operational automatic landing never looks outside — but because afterward, there is no approved means of taxiing, clearing, or accessing the aircraft with emergency services. The bottom of the scale is set by the airport’s ability to absorb the aircraft, not by the aircraft’s ability to arrive.

A nuance traps readers of European texts: EASA’s rewritten all-weather operations rules have moved away from the a/b/c subdivision, distinguishing CAT III with DH from CAT III without DH. The charts and ATIS still say IIIa and IIIb, so both vocabularies coexist; when they diverge, the approval decides.

What “no DH” removes — and what it leaves intact.

It removes exactly one thing: the altitude at which the pilot must have acquired a visual reference to continue. Nothing else — not the minimum RVR, not the approach prohibition, and especially not the decision. The decision migrates: it ceases to be can I see enough? and becomes is the system still capable?, answered by the redundancy status, warnings, and alert height rather than by looking outside.

That’s why the same aircraft can be approved for CAT IIIb with a decision height of around 50 feet on one runway and none on another. Three locks must open, and the tightest one wins:

  • The aircraft. Its certification basis, in the flight manual, indicates whether the landing system is fail-passive or fail-operational, publishes the alert height if it exists, and sets the lowest demonstrated DH. A fail-passive aircraft cannot lose its decision height.
  • The aerodrome. Facility performance category, lighting, RVR evaluation points, protected zones, backup power. Without full CAT III infrastructure and low-visibility procedures in place, the no-DH option does not exist there.
  • The operator. The approval may be deliberately more restrictive than the aircraft’s capability: many operators retain a visual window on CAT IIIb even with fail-operational fleets, as a window makes the decision cleaner and more trainable.

The trap. “No DH” is often read as “no minima”. The RVR is the limiting constraint in a no-DH operation, and it stops the approach at the prohibition gate long before anything else. An aircraft certified without a decision height still cannot start an approach to a runway reporting an RVR below its approved figure.

A “no DH” approach means the pilot doesn’t need to see the runway at a specific height to land. Instead, the aircraft’s systems must be working perfectly. But the runway still needs enough visibility for the aircraft to taxi safely after landing. The decision to land is based on the system’s health, not what the pilot sees.

An Airbus A350 is approved for CAT IIIb with no DH at London Heathrow (EGLL). On a foggy morning, the RVR is 75 meters at touchdown, 100 meters mid-runway, and 120 meters at the stop end. The aircraft lands automatically, but the crew must still follow taxi instructions carefully because visibility is extremely low. If the RVR drops below 75 meters, the approach is prohibited.

Fail-passive vs. fail-operational: the architecture is the license.

The distinction has nothing to do with the quality of the autopilot. It concerns what the aircraft does the moment a part of the automatic system fails.

A fail-passive system disconnects without significant drift, trajectory deviation, or attitude change — but does not complete the landing. The pilot must, and a pilot needs time and height to take over. Hence, a fail-passive approval always carries a decision height and never reaches the lowest RVR figures.

A fail-operational system continues the approach, the flare and the landing on the remaining channels — and the rollout as well, when that function is also fail-operational. Nothing needs to be taken over. This single property makes the decision height optional and unlocks the bottom of the RVR range.

What the two architectures imply operationally.
Aspect. Passive failure. Operational failure.
Behavior in case of system failure. Clean disconnection, aircraft left in trim, landing not automatically completed. Landing completed by the remaining channels.
Decision height. Always required — typically around 50 feet in the deepest approvals. A low DH, or none, depending on the approval held.
Alert height. The concept does not apply. Published in the flight manual; often around 100 feet, higher on some types.
Lowest RVR range. Limited — stops well above the CAT IIIb floor. Reaches the bottom of the approved range, rollout guidance permitting.
Failure at or below DH. Go-around, unless the visual reference already acquired supports a manual landing. Below the alert height, a loss of redundancy does not stop the approach.

The alert height is a threshold in the failure logic, not in the visual logic.

The alert height (AH) is a radio height published for the type, above which a CAT III approach is interrupted if a required redundant element, on board or on the ground, fails — and below which it is continued, as the residual redundancy has been demonstrated capable of completing the job. It is a system threshold with no visual content, it exists only on fail-operational aircraft, and it is what makes a no-DH approval coherent: the aircraft is deliberately committed at a point where it was still demonstrated capable. The consequence surprises crews in the simulator — the same failure one second on either side of the alert height calls for opposite actions, and nothing outside has changed.

Half of the minima are in the aerodrome.

Half of a CAT III minimum is not in the aircraft at all. It is in the runway, and it evaporates as soon as an element becomes unserviceable.

An airplane on a foggy runway at night with landing lights on and ground crew nearby.
Landing lights and ground crew visibility are critical in low-visibility approaches.

ILS integrity. The localizer and glidepath beams bend when metal is in the wrong place, which is why the ground consequence is expressed in two zones. The critical zone is kept clear of vehicles and aircraft during all ILS operations, whatever the category. The sensitive zone is larger, and its occupation is controlled rather than prohibited — that’s where the category bites: as the operation deepens, the protection tightens to one aircraft at a time, holding positions are moved further back, stop bars are in service. This is the main reason why low-visibility procedures reduce the movement rate — not pilots slowing down, but the geometry of beam protection.

Lighting and power. Approach, axis, touchdown, edge, and taxiway lighting as well as stop bars carry serviceability requirements that tighten as the category deepens, supported by backup power with switch-over times kept within one second on the circuits that matter. A CAT III runway with a degraded central circuit is a CAT II runway that morning.

RVR in three points. On a CAT II or III runway, the RVR is measured at the touchdown zone, mid-runway, and stop end, as fog is not homogeneous and the aircraft needs different things at each stage: acquiring the visual segment, confirming centreline tracking, then staying on the runway during the rollout and finding the way off it.

The three RVR evaluation points and what each actually protects.
Evaluation point. Protects. Status.
Touchdown zone. Acquisition of the visual reference at or before the decision height, and the flare. Always constraining, at every category.
Mid-runway. Centreline tracking during derotation and the start of the rollout. Constraining for CAT III, with a separate minimum; what that minimum is depends on the approval and whether the aircraft has rollout guidance. Consult your operations manual, not a generic table.
Stop end. Directional control at low speed and ability to leave the runway. Indicative in most approvals; some make it limiting. Its status is written in the approval, never assumed.

Hence one point worth stating plainly: the chart is not today’s truth. The plate records what the runway offers when everything is working; what it offers that morning is in the NOTAM, the ATIS, and the controller’s degradation call.

Where the figure in your operations manual comes from.

The applicable minima are the result of an intersection, and the arithmetic never changes: take the most restrictive constraint. Nothing in the chain can be traded for something else.

Who sets what, and where the figure is recorded.
Owner. Sets. Where it is recorded.
Certification authority, through the manufacturer. Lowest demonstrated decision height, alert height, fail-passive or fail-operational status, wind and configuration limits for automatic landing. Aircraft flight manual, all-weather operations section.
Aerodrome operator and State. Facility performance category, lighting and power equipment, RVR evaluation points, protected zones, low-visibility procedures, published aerodrome operating minima. AIP, aerodrome manual, approach charts, NOTAM.
Operator. The approval actually held, minima published to crews, applicable increments, training and currency program. Specific approval and operations manual.
Crew. Individual qualification and currency for the category, role, and type. Training records — an omission silently increases the applicable minimum.

Two side doors exist. The first is the special CAT II authorization in FAA parlance, “other than standard” CAT II in European parlance: it buys CAT II minima on a runway lacking full infrastructure, in exchange for additional aircraft capability and training. The second is enhanced flight vision systems, whose operational credit sits at the top end of the scale and not in CAT III. Both are approvals, not techniques — neither is improvised on the day.

What changes in the cockpit on the day.

Less than folklore suggests, and in a specific direction: the workload shifts from flying to monitoring and verification.

  • Verify capability first. The approach is briefed against the announced capability of the aircraft, the minimum equipment status, the declared category of the runway, and confirmation that low-visibility procedures are in effect. An unconfirmed CAT III announcement is a CAT I approach with a CAT III briefing.
  • The limits of the automatic landing are not the limits of the aircraft. The demonstrated values of crosswind, tailwind, and headwind for an automatic landing are typically more restrictive than the aircraft’s own limits, sometimes markedly, and they often kill the approach on a windy, foggy day. The runway slope and non-standard glidepath angles may also fall outside the envelope.
  • The approach prohibition. European rules set the gate at 1,000 feet above aerodrome elevation: with an RVR reported below the applicable minimum, the approach is not continued beyond that point. If the RVR then drops, it may continue, the required visual reference at the DH doing the filtering — an asymmetry that protects a stabilized aircraft from a fluctuating reading.
  • The calls change function. They cease to describe the outside world and begin to describe the machine: modes, redundancy status, alert height, decision height, and the calls that mandate a go-around.
  • The landing is not the end. Leaving the runway at low RVR, finding the taxiway centreline, respecting stop bars: that’s where things go wrong even on a good day, and it’s the least often trained phase.

The right reflex. Brief the degradation before briefing the approach. Decide out loud, in advance, what you do if the runway goes to CAT II, if low-visibility procedures are lifted, if the capability announcement drops a level, or if the mid-runway RVR moves. A degradation managed at 800 feet in the clouds is a rushed decision; the same managed in cruise is a non-event.

What triggers a go-around, and why part of it is automatic.

The logic is in layers, and each layer answers a different question.

  • Above the alert height — loss of a required redundant element, on board or on the ground, means the approach is interrupted or degraded. The aircraft is not yet committed.
  • Below the alert height, fail-operational — the loss of a redundant element no longer stops the approach; the remaining channels complete the landing. That’s the whole point of the concept. A failure deep enough to leave the system incapable of landing is another matter, and always ends in a go-around.
  • On a fail-passive aircraft — a failure at or below the decision height means a go-around, unless the visual reference already acquired supports a manual landing.
  • Loss of ground signal integrity — a transmitter failure, or beam disruption caused by an intrusion into the sensitive zone.
  • Excessive deviation, or radio height disagreement — automatically monitored on most types and announced as a warning that mandates the go-around.
  • Absence of the expected mode — if the landing mode is not announced at the height the type requires, the approach ends there. The negative case is just as triggering as the positive one.

Most of these elements are detected by an automatic monitor, but the go-around itself remains a crew action on most types: the system tells you that the operation has ceased to be valid, it does not fly it for you. And this maneuver is the least practiced, most time-critical event of the sequence — near the ground, in the fog, possibly after touchdown, at a state of energy where hesitation is costly. A category approval says nothing about the quality with which a crew performs a go-around at 30 feet. The debate is a perennial one on the FL330 forums, and the answers that hold are anchored in a type’s flight manual rather than a general principle.

Qualification and currency: the constraint crews forget.

A CAT III approval is granted to an operator but exercised by two individually qualified pilots. The chain goes through ground training on failure logic, simulator work on failure cases around the alert height, and line experience under supervision — by type, by category, generally by role.

Currency then erodes it. The classic structure is a small number of approaches of the relevant category in a sliding window of a few months, entirely or partially in an approved simulator, with a requalification path defined once it expires. The exact number belongs to the operator’s program, not a universal figure — but the mechanism is universal, and the consequence is as well: a pilot out of currency silently increases the crew’s applicable minimum, whatever the aircraft and the runway can do.

The same logic drives the increment for the inexperienced captain: a recently released captain applies increased minima until a threshold of experience on the type is reached. This is not a judgment on the individual, but an acknowledgment that a decision at 50 feet in the fog is a pattern recognition task, and that the patterns are built by exposure. Operators rebuilding a low-visibility program will find these requirements structured in the FL330 training catalog.

The trap. The three chains — aircraft, aerodrome, crew — degrade independently and none announces the others. A fully capable aircraft on a fully capable runway is still held to CAT I minima if a pilot’s currency has silently expired. Check the crew’s currency during preparation, not at the start of the descent.

What to keep.

The category labels an operation, not a rank. Removing the decision height does not remove the decision: it moves it into the system, where the alert height and redundancy status answer it. A fail-passive system retains a DH because someone must take over; a fail-operational one can remove it because no one needs to. And the figure printed in the operations manual is always the most restrictive of four independent constraints — one of which is the crew’s own currency.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

FL330, c’est toute l’aviation professionnelle — explorez nos autres espaces.