It was Data, not Noise

Again it turns out that things aren’t always what they seem to be. When looking at a suspended solution in much higher resolution than before, unsuspected details emerged.

Turbidity of settling 2,000 FTU formazin measured every 20 seconds.
Turbidity of settling 2,000 FTU formazin measured every 20 seconds. Yellow area is over 2200 FTU. CLick to read more.

What appeared to be a uniform suspension in which the particles sank in their characteristic settling velocity, turned out to apparently be a liquid with significant turbulence. Small “packages” of water with higher turbidity (lighter) or lower (darker) move in opposite direction in front of the sensor. The sensor is a SediMeter, consisting of 36 infrared optical backscatter detectors (OBS) 10 mm apart. It was developed for studying sedimentation, erosion, siltation and re-suspension.

Note that at time 16:40 there is some fascinating micro-turbulence with variations of around 10% within a minute outside a given sensor. If measuring once per minute, this might have been interpreted as noise, but with this high temporal resolution, one can clearly see how discrete packets of more turbid water are sinking.

An old and time-proven method to do grain-size analyses on fine-grained cohesive sediments (such as clay and silt) is to suspend them in water, and measure how fast they sink. This can be done with either the pipette method (extracting water at a certain level and weighing the sediments after drying it), or the hydrometer method (measuring the density of the liquid). In both cases one assumes that after X seconds, all sediment size Y or coarser have sedimented (cf. Stoke’s Law). It would be interesting to see, using this instrument, what really goes on in that sedimentation vessel…