Knowledge / Navigation and collision

Radar and AIS disagreement: how can target information be trusted?

Compare radar and AIS through time alignment, reference points, motion frames, target association and shared sensor dependencies.

On this page

Radar and AIS describe a vessel through different observation chains. Radar measures returned electromagnetic energy and develops a track; AIS transmits information produced or entered aboard another station. When their symbols or vectors disagree, the first task is to identify what each value represents. Choosing whichever display looks smoother can hide the actual problem. This article explains a structured comparison using original timing and geometry examples. It is educational material, not a live collision-avoidance procedure, equipment-adjustment instruction or claim that either source is universally correct.

Start with the observation mechanism

A radar echo is an observation of a reflecting object relative to the radar installation. Its displayed motion can depend on tracking, stabilization and own-ship sensor inputs. AIS is a cooperative broadcast service carrying identity, position and other fields. A received AIS position is a report from another system, not a direct measurement of that hull by the receiving vessel. Both paths contain assumptions and possible errors.

The IMO AIS overview explains carriage and the system’s information role. It does not establish that every nearby object carries a working transmitter or that all transmitted values are correct. Comparing radar and AIS is therefore a comparison of evidence, not a competition between one “real” source and one “digital” source. Each can reveal limitations in the other.

Distinguish position, heading, course and speed

AIS fields should retain their separate meanings. Heading describes orientation; course over ground describes the direction of ground velocity; speed over ground is its magnitude. A turn or current can make these differ without any instrument failure. The USCG Class A message description lists separate fields and unavailable-value conventions. A special code for unavailable data must not be interpreted as a valid bearing or speed.

Radar vectors also require a stated reference. A relative vector describes motion with respect to own ship; a true vector uses the selected stabilization and input basis. Sea-stabilized and ground-stabilized information answer different questions. Before comparing arrows, establish units, reference frame, vector time and the quantity being plotted. Two different arrows can both be correct if they intentionally represent different motion definitions.

Time alignment can explain a visible offset

Suppose an invented target moves steadily at 12 knots and one displayed position is twenty seconds older than another. In that interval, it travels 12 × 20/3,600 = 0.0667 nautical mile, approximately 123.5 m. An offset of this order can arise from the stipulated timing difference alone. The calculation does not establish an AIS update interval, an acceptable error or the cause of an actual mismatch.

Real systems may predict or extrapolate between reports, while other displays retain a received position. Record the source timestamp, receipt time, display behavior and any extrapolation rule before comparing. During a maneuver, constant-velocity propagation may also become inaccurate. The correct question is whether two positions represent the same object, reference point and time, rather than whether the symbols happen to overlap at one screen refresh.

Reference points matter on a large hull

An AIS reported position relates to the positioning reference configured for the transmitted data. Ship dimensions and antenna offsets help a display place the outline relative to that point. A radar track may follow a changing effective part of the echo rather than the identical physical point. Even with sound data, these observations need not coincide exactly for a large vessel or changing aspect.

The USCG AIS encoding guide describes position-reference dimensions and their units. Incorrect configuration can distort a displayed outline. The educational implication is to separate reference-point displacement from navigational position error. A symbol centered on the wrong point is not repaired by merely drawing a larger ship icon, and an attractive scale outline is not proof that the transmitted dimensions are correct.

Own-ship errors can affect several displays together

A heading input contributes to radar stabilization and transformations into a ground-referenced display. Position and speed inputs may also feed several bridge systems. If those systems share a faulty source, agreement among them may be a shared error rather than independent confirmation. Count evidence paths, not screens. A second application using the same sensor message is still dependent on that message.

For an original geometric illustration, a one-degree angular error at a range of one nautical mile corresponds to about 1,852 × sin(1°) = 32.3 m of transverse displacement. This is a small-angle geometry example, not an instrument accuracy specification. It shows how an angular input can produce a range-dependent positional mismatch. The MCA’s electronic-aids guidance emphasizes accurate own-ship inputs for plotting.

Correlation is a hypothesis that needs evidence

Associating an AIS identity with a radar echo is an interpretive step. Nearby targets, changing geometry, weak returns and data delay can make that association uncertain. A name beside an echo can become psychologically persuasive even when the match has not been established. An analysis should preserve the association status rather than presenting every matched symbol as confirmed identity.

In a fictional scene with two vessels close together, ask what evidence supports each possible association: time-aligned position, relative movement, dimensions, visual observation where available and continued consistency. Do not force a match simply because the display prefers one combined symbol. If evidence remains ambiguous, the uncertainty itself is relevant. The article does not provide a universal matching radius or automated fusion rule for navigation.

Check static and manually entered information separately

Not every AIS field comes from a continuously measured sensor. Some information is configured or manually updated. A stale destination or navigational-status entry has a different meaning from a noisy position measurement. One incorrect field does not logically prove every other field is wrong, but it is a reason to avoid uncritical reliance on the complete message. The evidence should be assessed field by field.

A useful data-quality record distinguishes unavailable, stale, implausible and conflicting information. Missing data should remain missing; substituting a guessed heading can turn an acknowledged gap into false certainty. Similarly, an apparently precise numeric field may still be wrong because its upstream sensor or configuration is wrong. Precision of transmission is not accuracy of the underlying physical observation.

Radar has its own failure and ambiguity mechanisms

Weak reflections, clutter, shadowing, multiple targets, tracking behavior and changes of aspect can affect radar information. An echo’s apparent centroid can move while the hull’s reference point follows a smoother path. A newly acquired or recently maneuvering target may have a less stable vector. The correct response is not to assume that radar always resolves every AIS disagreement automatically.

The MCA radar and plotting guidance provides equipment-use context. A technical review should state which observations were available and under which operating conditions. It should avoid retroactively treating the best final track as information that existed throughout the encounter. Retaining raw or appropriately recorded evidence helps distinguish acquisition difficulties from later interpretation errors.

Radio identification is another evidence path

Calling a vessel by the AIS name does not itself prove that the responding voice belongs to the intended radar target. Misidentification and misunderstanding remain possible. The MCA’s VHF and AIS guidance cautions against these risks and against allowing radio exchanges to displace COLREG compliance. Communication can add information while also consuming scarce attention and time.

For a training debrief, identify exactly what the exchange established. Was identity confirmed? Was the same situation understood? Was an intention stated, and was the subsequent action observed? Do not collapse these stages into “VHF contact made, problem solved.” The example is about evidence quality, not a recommendation to negotiate a passing arrangement or a script for an active encounter.

A structured desk-review example

Consider a fictional record containing a radar target, an AIS position and a displayed ship outline that appears displaced. First establish whether both records concern the same time. Then compare their coordinate and motion references. Next examine antenna offsets, own-ship inputs and the certainty of target association. These steps organize evidence; they are not a mandatory real-time sequence or permission to defer necessary navigational action.

Suppose time correction explains part of the offset but a range-dependent lateral difference remains across several targets. That pattern may support investigating a shared angular-reference issue. It still does not prove a particular defective sensor. Alternative explanations and measurement uncertainty remain. A good desk report distinguishes observations, calculated effects, hypotheses and the additional evidence needed to discriminate among them.

Preserve provenance when combining sources

A combined target record should retain where each value came from, its time, validity and processing. If a fused position is shown, the original disagreement should remain available for appropriate review rather than disappearing into an unexplained average. Averaging two values is not a general cure: different reference points, stale observations or a common bias can make the average physically misleading.

For an educational dataset, label observed versus propagated positions and record changes in association. This makes later calculations reproducible. It also prevents a model from being evaluated against its own processed output while claiming independent confirmation. Any operational integration system has additional approved requirements; this article describes general evidence principles rather than a design or certification method.

Common errors and the useful conclusion

Common errors include comparing different vector frames, treating heading as COG, ignoring timestamps, assuming an outline is centered on the hull, and counting shared displays as independent sensors. Another is interpreting a missing AIS report as proof that no vessel exists. A discrepancy can be useful evidence precisely because it challenges an overly simple picture.

The useful conclusion is usually a bounded statement: which values disagree, how much of the difference the known timing and geometry explain, which hypotheses remain, and what evidence is needed next. It should never replace the full navigational assessment with a rule that one source always wins. The underlying aim is traceable understanding, supported by qualified ship-specific assessment and procedures.

Sources