Percolation and Its Effect on EchoQuilt Readings

When the Ceiling Rains Silt Into Your Data

A Ox Bel Ha survey team noticed the same chamber produced two completely different quilt reconstructions on two sequential dives, twenty minutes apart. The first dive showed a clean phreatic room with regular wall geometry. The second dive showed the same room cluttered with apparent reflectors that had no physical correspondence. Review of the audio log identified the cause: the second dive had spent longer under a ceiling pocket, and percolation had generated a shower of dislodged silt particles that produced pseudosound throughout the reconstruction.

How to avoid percolation — Inspired To Dive explains the mechanism: exhaled bubbles rise, hit the ceiling, and dislodge fine silt that then drifts down through the passage. DAN Low-Visibility Diving confirms that percolation produces the same kind of turbidity as fin-kick silt but from a different direction — down rather than up — and often at a different grain-size distribution. ProTec's overhead diving hazards notes turbidity from percolation occurs even with good technique, particularly in passages with loose fill or dry-portion contact surfaces.

For an acoustic survey instrument, percolation is not just a visibility problem — it is an active signal contaminant. Drifting silt particles act as small scatterers in the water column. A hydrophone receiving ambient conduit noise sees that noise arrive with extra reflections off the moving particle field, and naive processing would treat those extra reflections as wall returns. Output: a quilt cluttered with ghost walls.

The contamination has its own signature. A particle field drifting downward through the conduit at a few centimeters per second produces a Doppler-shifted reflection pattern with characteristic time-variability. Static walls produce stable returns at fixed travel times; drifting particles produce returns that walk through travel times as the particles fall. The two are physically distinct, and a properly-modeled instrument tells them apart by the temporal structure of the reflection. EchoQuilt's percolation handler applies exactly this logic, but the quality of the separation depends on having enough percolation-trained baseline data for the local sediment chemistry. Loose carbonate flour from a Yucatán cenote ceiling falls differently than silt-clay mixtures from a Florida sinkhole roof, and the instrument benefits from a few minutes of percolation calibration on first deployment in any new system.

The exhalation rate makes percolation worse on open-circuit dives than on rebreather dives. A backmount diver running open-circuit at 40 meters depth exhales every two to four seconds, and each exhale sends a bubble train upward toward the ceiling. A JJ-CCR diver on the same dive recycles the loop and exhales only when manually flushing — which can be once every ten or fifteen minutes — so the bubble-to-ceiling impingement rate drops by an order of magnitude. WKPP push teams running CCR specifically note this side benefit of the rebreather configuration: cleaner ceilings, cleaner percolation profile, cleaner survey product. Open-circuit teams on CCR-restricted sites have to manage exhale cadence more carefully if percolation is a known concern.

Filtering Percolation From True Reflections

EchoQuilt handles percolation as a known signal class with known characteristics. Percolation-driven pseudosound has a distinct spectrum — broadband, time-variable, and tied to the diver's own bubble train. True wall reflections have a spectrum tied to the ambient flow noise and stable over the duration of a station hold. The two are separable with standard spectral filtering once you know what to filter for.

The quilt metaphor works cleanly here: percolation adds a noisy yarn to the weave that EchoQuilt identifies and pulls out before the quilt gets stitched. What remains is the clean warp-and-weft of motion and acoustic returns, producing a reconstruction of the conduit itself rather than of the silt cloud. The pulled percolation data is not discarded — it becomes a diagnostic for later analysis of which chambers are silt-prone and which ceilings are stable.

Research on related signal-filtering problems supports this approach. Flow-noise and turbulence in tidal channels (ResearchGate) documents how flow-noise from particles and turbulence creates characterizable pseudosound spectra that can be modeled and subtracted. Busy Urban Soundscape Underwater (MDPI) covers acoustic indicators for how particulate matter degrades hydrophone data in other domains, confirming the cross-domain applicability of the filtering approach. Silt out — Wikipedia notes that percolation is distinct from fin-kick disturbance — the two produce different particle distributions and different acoustic fingerprints, which EchoQuilt reads as separate signal classes.

For cave diving survey teams, the operational implication is that not all silt-outs are equal for the instrument. A fin-kick silt-out leaves the ceiling clean and the floor disturbed — the quilt continues to work well because the ceiling reflections carry most of the conduit geometry information. A percolation silt-out is worse for the instrument because the noise source is in the same region as the signal. Teams surveying high-ceiling chambers with loose fill need to be aware that the instrument performs better after a percolation layer settles than during active percolation. The fin-kick case is covered in detail in the silt interference writeup, which contrasts the floor-disturbance pattern with the ceiling-disturbance pattern handled here.

Conservation biologists working dry caves face a related problem with guano clouds and visitor-kicked dust, and the Pd surveillance visits workflow for pathogen-surveillance trips uses comparable filtering logic to handle particle clouds that confuse acoustic picture.

Advanced Tactics for Percolation-Prone Dives

Three practices improve EchoQuilt quilt quality in percolation-prone caves. First, time the survey station holds to avoid bubble trains at the ceiling. A backmount diver exhaling once every four seconds sends bubbles up on a regular cadence. Holding station for 45 seconds and surveying during the exhale gaps produces cleaner acoustic data than holding for a continuous 45 seconds that overlaps multiple bubble arrivals at the ceiling.

Second, choose rig orientation for the conduit. A sidemount diver in a low-ceiling chamber directs exhaust to the sides rather than straight up, avoiding the ceiling entirely. Where ceiling height is marginal and percolation is a concern, switching from backmount to sidemount for the survey push can meaningfully reduce ceiling-silt contamination. Teams planning Ox Bel Ha or Sistema Huautla dives often configure by chamber rather than by default rig preference.

Third, log percolation events explicitly. EchoQuilt's filtering removes the signal from the quilt, but the logged percolation events themselves are valuable survey-product metadata. A chamber that consistently percolates is useful geology — the ceiling is loose, the sediment is fine, and the dive planning should account for reduced visibility windows on future visits. Archival percolation logs become a map of cave maintenance-state over seasons. Multi-season analysis of percolation hotspots integrates with the seasonal cave changes workflow to track which ceilings are loosening and which are stabilizing across years.

One further tactic: maintain a target bubble-path angle. A diver who swims at a steep upward angle drives bubbles into overhead rock, maximizing percolation. A diver who keeps trim neutral and swims flat minimizes bubble-to-ceiling interaction. Good trim was always a cave-diver virtue; on sound-mapping dives it becomes an instrument-performance factor. The team that trims flattest collects the cleanest quilts. GUE Cave 1 and Cave 2 cohorts running trim drills in spring-pool basin work transfer the discipline directly into percolation-management on full-cave penetrations later in the program.

Join the Waitlist for Cave Diving Survey Teams

If your team has ever watched a chamber get unusable for survey because a ceiling pocket rained silt for the entire dive, EchoQuilt's percolation filter keeps your reconstruction going. We are prioritizing access for teams working Sistema Sac Actun, Sistema Dos Ojos, and WKPP push dives where percolation is a known hazard across long segments. Drop your email below with a brief note on which chambers you most often abandon to percolation, your typical rig configuration in those chambers (sidemount versus backmount), and your team's GUE Cave 1 or Cave 2 trim baseline.

Early cohort members get their field data reviewed by our signal-processing team for percolation-spectrum tuning specific to the local sediment, and we will scope a per-chamber bubble-cadence template for your most active percolation-prone passages, including the rig-orientation guidance, the percolation hotspot archival schema, and the multi-season ceiling-stability tracking dashboard you can use to plan future maintenance dives. Priority access goes to NSS-CDS, GUE, NACD, and QRSS-affiliated teams with active mapping goals in known percolation-heavy systems.