🚢 IEC 60872 ARPA: The Radar System That Sees Collisions Before They Happen

At sea, a fully laden VLCC (Very Large Crude Carrier) traveling at 16 knots needs roughly 2.5 nautical miles and 15 minutes to come to a complete stop. In congested shipping lanes like the Singapore Strait, where vessels routinely pass within half a nautical mile of each other, this physics reality means one thing: collision avoidance decisions must be made early, and they must be right. ARPA — the Automatic Radar Plotting Aid standardized by IEC 60872 — is the electronic bridge between raw radar returns and informed navigational decisions. It automates what was once the single most mentally demanding task on a ship’s bridge: tracking every vessel on the radar screen, calculating where each is heading, and determining whether any of them are on a collision course with your ship.

📡 How ARPA Works: The Full Pipeline from Radar Echo to Collision Alert

ARPA is not a separate sensor. It is a processing layer that sits between the marine radar’s video output and the human operator, extracting motion information that the radar front-end detects but does not interpret. Understanding each stage of the ARPA processing pipeline is essential for knowing when to trust its output — and critically, when not to.

1.1 Target Acquisition: Finding Ships in the Noise

The first challenge ARPA faces is distinguishing real vessel echoes from sea clutter, rain clutter, radar interference, and sidelobe returns. This is done through:

• Manual Acquisition: The operator designates a specific radar echo using a joystick, trackball, or touchscreen cursor. The ARPA places an acquisition gate around that position and begins the tracking process. Manual acquisition gives the operator full control but requires active attention.
• Automatic Acquisition (Auto-ACQ): The operator defines one or two guard zones (annular rings or sector-shaped boundaries on the PPI display). Any echo entering these zones is automatically acquired and tracked. While reducing operator workload, auto-acquisition is a double-edged sword — it can flood the tracking processor with false targets from sea clutter near the coastline or in heavy weather.

IEC 60872 mandates a minimum tracking capacity of 20 targets simultaneously for an ARPA to comply. Modern ARPAs routinely handle 40-200 targets, but the standard’s baseline ensures that even entry-level compliant systems are functional in moderate traffic density.

1.2 Tracking and Filtering: The Kalman Engine

Once a target is acquired, the ARPA receives a new position measurement (range and bearing) with every antenna rotation — typically every 2 to 3 seconds. These raw measurements are noisy due to:

• Target scintillation (glint): The apparent radar center of a vessel shifts as different parts of its superstructure dominate the return.
• Quantization error: Range and azimuth discretization in the radar digitizer.
• Sea clutter modulation: Wave returns can momentarily mask or distort the echo.

ARPA smooths these measurements using a tracking filter — typically a Kalman filter or an alpha-beta filter in older implementations. The filter maintains a state vector for each target (range, bearing, range rate, bearing rate) and recursively updates it with each new observation, weighting the new measurement against the predicted position based on an estimate of measurement noise. The result is a smoothed track that converges to stable course and speed estimates over time.

The critical operational takeaway: track convergence takes time. IEC 60872 specifies accuracy requirements that apply “after 1 minute of steady-state tracking.” During those first 60 seconds — and for a similar period after the target or own ship maneuvers — the displayed vectors and CPA/TCPA values are still settling and must not be used for collision avoidance decisions.

1.3 CPA and TCPA: The Numbers That Matter

The Closest Point of Approach (CPA) and Time to CPA (TCPA) are the two derived numbers that ARPA continuously calculates for every tracked target. They are the quantitative distillation of “is this ship going to hit us, and if so, when?”

• CPA is the minimum distance at which the target will pass own ship, assuming both vessels maintain their current course and speed.
• TCPA is the time remaining until that closest point is reached.

The operator sets alarm thresholds — typical settings are CPA = 1.0 NM and TCPA = 12 minutes for open-sea conditions. When any tracked target’s CPA drops below the threshold AND its TCPA drops below the threshold, the ARPA triggers an audible and visual collision warning. The alarm is a logical AND: a vessel with CPA = 0.3 NM but TCPA = 45 minutes is a distant concern, not an immediate threat.

⚓ The Transition from Manual Plotting to Automated Tracking — and What We Lost

2.1 What the Grease Pencil Taught That ARPA Obscures

Manual radar plotting was slow, but it forced the officer to engage intimately with the raw radar picture. Every plotted position was actively observed and judged. The process itself — marking the echo, waiting three sweeps, marking again, drawing the relative motion line, constructing the velocity triangle — created a mental model of the traffic situation that no automatic system can fully replicate.

ARPA’s automation removes this forced engagement. An officer can glance at the vector display and the CPA/TCPA readout without ever examining the raw radar video underneath. This creates a risk known in human-factors circles as automation complacency: the tendency to trust the system’s processed output without verifying its inputs. The engineering response to this problem has been twofold:

1. Integrity monitoring: Modern ARPA systems display confidence indicators (track quality metrics, convergence status) alongside the processed data, making the system’s own uncertainty visible to the operator.
2. Mandatory training: STCW (Standards of Training, Certification and Watchkeeping) requires all navigating officers to demonstrate competence in both manual radar plotting and ARPA operation, specifically so they understand what the automation is doing — and what it might be getting wrong.

2.2 The Target Swap: ARPA’s Most Dangerous Failure Mode

When two tracked targets pass close to each other — say, two fishing vessels crossing within 0.2 NM — the ARPA’s data association algorithm faces a difficult decision: which echo from this sweep belongs to which track from the previous sweep? If the algorithm guesses wrong, the two track IDs swap their histories. Target A now displays Target B’s course and speed, and vice versa.

This is rarely self-evident on the ARPA display. The operator sees two perfectly reasonable-looking vectors and two plausible CPA/TCPA values — except they belong to the wrong ships. If an avoidance maneuver is based on swapped data, the consequences can be catastrophic. Modern ARPAs mitigate this risk through:

• Multi-hypothesis tracking (MHT) algorithms that maintain multiple data-association possibilities and defer commitment until ambiguity resolves.
• Visual and audible alerts when tracked targets come within a configurable proximity of each other.
• AIS fusion: if both vessels broadcast AIS, their MMSI identities anchor the track associations and prevent swaps.

🛠️ Engineering Lessons for Collision Avoidance System Design

3.1 The Sensor Quality Chain: Garbage In, Garbage Out

ARPA accuracy is bounded by the weakest sensor in the chain. This is a fundamental systems engineering principle that is often overlooked during both equipment procurement and bridge team operations.

Gyrocompass heading feeds the ARPA’s true-motion transformation. If the gyro drifts by 2 degrees over 6 hours (within spec for many commercial gyrocompasses), every target’s computed true course is offset by 2 degrees. On a CPA calculation at 6 NM range, a 2-degree angular error translates to a lateral position error of roughly 0.2 NM — enough to turn a safe 1.0 NM passing distance into a 0.8 NM close-quarters situation without the operator ever being aware of it.

Speed log accuracy matters enormously. A single-axis speed log measuring speed through water (STW) will report 14 knots when the vessel is making 14 knots through water — but if the ship is in a 3-knot cross-current, the speed over ground (SOG) is a different vector entirely. ARPA computes target true motion by vector addition of own ship’s motion and the target’s relative motion. Using STW instead of SOG when calculating target true vectors in areas with significant current produces systematically incorrect results. IEC 60872 requires ARPA to accept dual-axis speed inputs (both longitudinal and transverse components) from Doppler logs or GPS, which provide SOG rather than STW.

Radar azimuth quantization is an often-unappreciated error source. If the antenna azimuth encoder has a resolution of 0.1 degrees (3600 pulses per revolution), then the bearing of every echo — and every tracked target’s position — is quantized to 0.1-degree bins. At 12 NM range, 0.1 degrees subtends an arc of approximately 0.02 NM (37 meters), which is negligible. But older radars with 0.5-degree encoders (720 pulses per revolution) produce bearing quantization of 0.5 degrees, subtending 0.1 NM at the same range — enough to visibly degrade the smoothness of ARPA tracking.

3.2 Why ARPA + AIS Fusion Is Harder Than It Looks

The marriage of ARPA tracking and AIS target data seems straightforward: AIS provides a ship’s GPS position, course, speed, and identity; ARPA provides independent radar-based position and motion data. Fuse them, and you have the best of both worlds. In practice, the integration raises subtle engineering problems:

Temporal alignment: AIS position reports arrive asynchronously, at intervals ranging from 2 seconds (Class A, fast-moving) to 3 minutes (Class B, at anchor). The ARPA updates every 2-3 seconds with every antenna rotation. A fused display must interpolate or dead-reckon the AIS position between reports and somehow represent the growing uncertainty as the AIS data ages.

Spatial alignment: The radar measures range and bearing from the antenna position, which is offset from the ship’s center. AIS reports the GPS antenna position, which is also offset from the ship’s center — but by a different amount, in a different location. Both must be referenced to a common point (typically the conning position) before fusion.

Correlation ambiguity: When 15 AIS-equipped vessels are in radar range, matching each radar track to its AIS counterpart is a combinatorial assignment problem. A good correlation algorithm uses not just position but also course, speed, rate of turn, and vessel dimensions — and still must handle edge cases (vessels with similar courses and speeds at similar ranges).

3.3 The Trial Maneuver: An Underutilized Safety Mechanism

IEC 60872 mandates that every ARPA provide a trial maneuver function. The operator enters a hypothetical course and/or speed change. The ARPA simulates — without executing — how all tracked targets’ relative motions would evolve if own ship actually made that maneuver. The simulated vectors, CPA, and TCPA values are displayed on the screen, typically in a distinct color or with dashed lines to distinguish them from live data.

Despite being one of the most powerful collision avoidance tools on the bridge, trial maneuver is consistently underused in real-world operations. Studies of VDR (Voyage Data Recorder) playback from collision incidents show that in a significant majority of cases, the officer of the watch had sufficient time to test multiple avoidance scenarios using trial maneuver — but did not. The reasons cited include: time pressure, unfamiliarity with the function, buried menu access paths, and overconfidence in mental simulation.

From a system design perspective, this points to a clear engineering requirement: the trial maneuver function must be a one-touch, always-available control — not buried in a submenu. It must be usable within 3 seconds of the decision to try it. The best implementations place a dedicated physical button or a persistent on-screen softkey labeled “TRIAL” directly adjacent to the course and speed indicators.

❓ Frequently Asked Questions