CAT_04 // Ocean Science

Turbidity Cascades: What 24 Contained Incidents Taught Us

Published
Reading time
2 min
A dark stormy sea under heavy clouds
Photo: @heytowner via Unsplash

Twenty-four sediment turbidity events, all detected before they became visible in the raw trace, all contained by the autonomous ballast policy. What the pattern looks like.

The Autonomous Rules panel on every Bathys console shows the same line: Automated ascent policy active. 24 incidents contained. This is the story of those twenty-four.

What a turbidity cascade is

A turbidity cascade begins when a slope of loose sediment fails and slides down-slope as a dense, fast-moving current. To a mooring in its path it looks like three things in quick succession: a sharp rise in acoustic backscatter as suspended sediment thickens the water column, a pressure gradient propagating down-slope, and finally a mechanical load on the mooring line itself.

The third one is the dangerous one. The first two are the warning.

The pattern across 24 events

Every one of the twenty-four events shared the same signature:

  • Backscatter first. Acoustic backscatter rose by at least 8 dB between 6 and 14 minutes before the pressure gradient crossed Watch.
  • Gradient second. The pressure-mapping module opened a Turbidity Cascade event in the anomaly console when the gradient propagated across two or more topologically connected moorings.
  • Sediment load threshold. The event log line Sediment load > 85% appeared, on average, 11 minutes before the mechanical load would have exceeded the mooring’s rated tension.

That eleven-minute window is what the ballast release policy is built around.

What “contained” means

Contained does not mean the sediment stopped. It means the mooring was not lost. In each of the twenty-four cases the policy either released ballast and brought the instrument package up-slope of the flow, or, in nine cases, an operator reviewed the event inside the window and held the mooring in place because the console showed the gradient flattening.

Nine holds out of twenty-four is the number I am proudest of. It means the system is not just fast; it is legible enough that people override it well.

What the ocean taught us

Two of the twenty-four events occurred on slopes our bathymetry model classified as stable. Both slopes are now flagged, and both contributed to the decision to version the terrain model. One event began during a spring tide in a canyon head where we had never seen sediment movement before. It is now a permanent watch site.

The array watches the ocean. Every so often the ocean corrects the array.