Experiments with turbidity currents – some historical context

Experimental modeling of geological processes has a long and distinguished history. Throughout, these experiments were devised to duplicate, at much reduced scales, structural elements as they relate to the Earth’s crust (e.g., folds, faults, orogenic wedges), and surficial landscapes that form in response to surface erosion, sediment transport and deposition. The general idea is that if we can reproduce earthscapes experimentally, then we can begin to understand the processes underlying their formation.

An important principle guiding these kinds of experiments is similitude – how similar are they to natural phenomena? This question inevitably requires us to consider the ‘scale’ of the experiments – is the experiment scaled with respect to variables such as length or is it scaled with respect to rock properties such as viscosity and rock strength that influence the rheological behaviour of the experimental materials used (e.g., brittle versus ductile behaviour). Scaling considerations were important even in the earliest attempts at experimentation. James Hall, in his celebrated experiments on folded strata (presented to the Royal Society of Edinburgh in 1815) used materials like cloth and pliable clay in his attempts to replicate qualitatively the kinds of structures observed in outcrop.

Experimentation came of age with the iconic paper by M.K. Hubbert (1934) who established methodologies for scaling that changed the rationale of analogue modeling from purely descriptive to quantitative. In the decades that followed, all manner of experiments were devised to unravel the dynamic intricacies of tectonic, geomorphic, and sedimentologic phenomena. Scaling was centred on the rules guiding dimensional analysis and derived dimensionless quantities (in experiments involving fluid flow, Froude and Reynolds numbers are important). The rationale for this kind of experimentation is nicely encapsulated by Middleton and Wilcock (1994) as “The design problem for the model is then to set all the (other) scale factors so that the model is dynamically similar to the original “;dynamic in this context means geometrically, dynamically, and kinematically similar.

The word problem here is well chosen because creating similarity between variables at experimental and natural (real world) scales is difficult. For example, in experiments involving sedimentation and fluid flow, Froude and Reynolds numbers are important, but it is not possible to scale both simultaneously. If grain size is downscaled too far the model sediment will become cohesive. Furthermore, natural systems are inherently variable.

Paola et al., (2009) challenge this mindset on the basis that, despite the inevitable scaling problems in stratigraphic and geomorphic models, the experiments prove to be “unreasonably effective” when compared with real world phenomena. Their conclusion is stated thus – “The observed consistency between experimental and field systems despite large differences in governing dimensionless numbers is what we mean by “unreasonable effectiveness”.”. They attribute this to natural scale independence between experiment and natural systems. In other words, if an experimental system is a small copy of a much larger natural system, then internal similarity will be similar between the two systems. This approach retains the general concept of similarity but de-emphasizes the need for exact dimensional similarity between experiment and the real world. The crux of their argument is that with this approach we can focus on deciphering underlying processes rather than losing sleep over differences in scale. Paola et al., are not saying we should dispense with scaled models (they are important in engineering studies) but to focus on similarities that will provide analytical insight.

The early experiments on turbidity currents by Phillip Kuenen and Carlo Migliorini are iconic and yet they were conducted very much as a ‘look and see what happens’ approach and to test hypotheses, without explicit reference to dimensional similarity.

Kuenen’s 1937 experiments: testing an hypothesis

Kuenen’s best remembered contribution is the paper on experimental turbidity currents and graded bedding with Carlo Migliorini (Kuenen and Migliorini, 1950) that reset our understanding of sediment gravity flows and deep ocean sediment distribution. However, Kuenen had experimented with sediment gravity flows 14 years before the iconic graded bedding paper. His 1937 paper (Experiments in connection with Daly’s hypothesis on the formation of submarine canyons) reports on flume experiments designed specifically to test the logical consequences of a hypothesis promoted by Reginald Daly in 1936, that submarine gorges (Kuenen’s name for submarine canyons) owe their existence to the erosive effects of turbulent density currents (turbidity currents) (there was a lot of discussion about the origin of submarine canyons at that time – see papers by Francis Shepard and others).

[R. Daly, 1936. Origin of Submarine Canyons.   F. P. Shepard,1936: The Underlying Causes of Submarine Canyons]

The experiments were designed to answer 5 questions (Kuenen, 1937, p. 331):

1. Will a suspension of sediment flow down a slope under water?
2. Will this type of current continue to a considerable depth without loosing its motive force in consequence of mixing with clear water?
3. Has such a current any erosive power?
4. Will such a current follow a slight initial gorge?
5. Is the rate of flow increased by enlarging the scale of the experiment? In other words, can swifter currents be expected in nature than in the laboratory?

The first four questions deal with processes associated with density currents; Question five attempts to equate experimental scale with real world scales.

The flume he used was like many modern flumes. The glass sides acted as physical boundaries. The sediment bed was constructed from sand overlain by a thin layer of gypsum (mainly for colour contrast); model bed slopes were predetermined. Initial slurries consisted of various mixtures of clay, water, and rock salt. There is no discussion of variable scaling during the initial tank setup. Kuenen argued that we can try to answer the questions regardless of the scale of the experiment and given that the very existence of turbidity currents was still being debated, there were no analogues against which variable scaling might be compared. The concentrations of the experimental mud suspensions were not judged on whether they scaled correctly to natural occurrences, but whether they produced sensible results.

The results of Kuenen’s experiments were unreasonably effective. He was able to provide cogent answers to all five questions:

1. The bottom-hugging density currents show most of the flow characteristics produced by more recent experiments, as shown in his Figure 5B (top of page):
   • Flow was turbulent.
   • There is a well-defined flow-head where the snout overhangs the base of the flow.
   • A sediment plume overlies the body of the flow; the plume billows are capable of ingesting ambient water that dilutes the flow.
   • The sediment plume contains incipient Kelvin-Helmholtz waves (not identified by Kuenen) resulting from shear between the plume and overlying water.
   • Experimental flow velocities ranged from about 3 to 20 cm/s; velocities increased with slurry specific gravity and volume.
2. The most mobile flows continued to the end of the tank (4-5 m). Kuenen indicates that mobility depends on turbulence, and that the distance of travel is related to the size or volume of the initial slurry and the maintenance of turbulence.
3. The currents had erosive power – observed where gypsum was incorporated into the flow.
4. The currents tended to migrate to and enlarge pre-existing topographic lows.
5. Question 5 was an attempt, a posterior, to extrapolate the experimental results on flow velocity to what might be expected in natural flows, at least to order of magnitude values.

Kuenen and Migliorini, 1950

The Kuenen 1937 paper is important, but it was the 1950 contribution by Kuenen and Migliorini that changed our thinking about sediment gravity flows in general, and turbidity currents and turbidites in particular. The experiments were designed specifically to test the hypothesis concerning the deposition of graded beds.

[They were not the first to suggest that graded beds were deposited by turbidity currents; that accolade goes to Bramlette and Bradley, 1940 based on their observations of some deposits in deep sea cores – also available from the USGS.]

Two flumes were used in their experiments; one 2 m long with glass sides and filled with sand so that the bed sloped towards one side – forcing channelized flow. The second flume was an open-air trench 30 m long that again confined the flowing current. The slurry in both sets of experiments was a mixture of clay, silt, sand, and gravel that produced initial conditions with different densities and viscosities; initial slopes were also varied. The initial conditions were varied by trial and error to determine which sediment mixtures, slopes, and slurry volumes produced flows with sufficient mobility.

The results spoke for themselves; the dialogue in their paper expresses some surprise at the “excellence of the grading” (p. 106). The experiments produced stacked successions of graded beds from turbulent suspensions, when successive flows were introduced to the flume. Flow velocities were as high as 80 cm/s. They also demonstrated that the coarsest fractions were deposited closer to the flow starting point, becoming progressively finer grained down-flow. Graded beds at the far end of the flume were composed of fine sand – silt – clay. Partial erosion of earlier deposits also occurred at the base of turbulent flows. They also confirmed a crude empirical relationship between flow viscosity and the presence of turbulence; too much clay and the non-turbulent slurry flowed en masse for short distances or broke apart, with no grading in the resulting deposit.

No ripple-like bedforms developed in any of the flows. However, Kuenen and Migliorini surmised that ripples might overlie the graded beds during the last stages of suspension fallout in response to normal ocean currents – note that Arnold Bouma would not publish his turbidite ‘model’ for another 12 years. Some experimental runs also developed deformation structures such as pull-apart layers and folds that look like convoluted laminae in response to sediment loading on a sloping surface and the early stages of compaction and dewatering. All of these sedimentary structures would eventually be incorporated into Bouma’s iconic model.

Unreasonable effectiveness

Sedimentological experimentation in the period to 1950 was in its infancy, its audience sceptical. In this environment, Kuenen and his colleagues had attempted to duplicate and tease apart a smidgen of understanding about two of the least understood phenomena – sediment gravity flows and graded bedding. The existence of turbidity currents had minimal confirmation at that time; bottom-hugging suspensions of sediment had been observed to flow down reservoir slopes and small lacustrine delta slopes. Oceanographers also knew something was going on across the continental slope and deep-sea floor when trans-Atlantic cables were severed immediately following the 1929 Grand Banks earthquake. But nothing was known about the flow compositions, properties like density and viscosity, flow rheology, or the depositional response to the passage of such flows.

In this light, Kuenen’s experiments were prescient. They were conducted with a sense of ‘try this and see what happens’. Even the more recent emphasis on model similarity had little to recommend it because there was virtually nothing to compare with. Their concern with scale is summarised thus “The artificial turbidity currents can give us some idea of what to expect in nature where the larger dimensions should maintain higher velocities.” (p. 103). There were no explicit attempts to design the experiments according to a priori scaling factors such as length or the dynamic properties of flows such as viscosity. Their consideration of properties like density and viscosity were directed at producing sensible results, mostly by trial and error. Multiple experiments conducted in the decades that followed have confirmed that the results Kuenen and others produced were unreasonably effective.

R.A. Daly, 1942, The Floor of the Ocean: New Light on Old Mysteries p. 136. (Available from Biodiversity Heritage Library

Other posts in the turbidite series

Experiments with turbidity currents – some historical context

Experiments with turbidity currents – scaling laws

Experiments with turbidity currents – three examples