One of the most useful anomaly visuals is not a dramatic failure at all. It is a disagreement.
The vessel is turning, but one motion source reflects that turn differently from another. In the demo scene, the IMU and GPS vectors begin aligned, then diverge.
That divergence tells an important story: trust is no longer something you can assign to a single feed in isolation.
What the scene shows
The animation uses two vectors to make the problem immediately understandable.
At first, GPS and IMU motion evidence support the same picture. Then the GPS-derived course begins to lag and diverge while the inertial picture continues to show a live turn.
That makes the anomaly visible as a comparison problem, not just a sensor failure problem.
Why this matters
In practice, independent corroboration is one of the strongest ways to challenge persuasive bad data.
If GPS says one thing and motion evidence says another, the disagreement itself is operationally meaningful.
It may indicate:
- a degraded GNSS picture
- delayed or stale course behavior
- a spoofing-related trust problem
- a need to slow down and cross-check before relying further on the displayed track
The key point is that no single source should automatically win the argument.
The disagreement is the signal
A disagreement does not always tell you immediately which source is wrong. But it does tell you something important:
the trust state has changed.
That is often enough to justify:
- more scrutiny
- more cross-checking
- less reliance on a single displayed answer
Good monitoring helps crews see that trust change early instead of discovering it after the consequences grow.
The practical takeaway
GPS vs IMU disagreement is a good example of why corroboration matters. The bridge does not just need data sources. It needs a clearer way to see when those sources stop supporting the same operational picture.
GeoWatch is designed around that kind of corroboration problem, helping operators spot disagreement sooner and keep a stronger record of how trust changed during the event.