Downstream Fining of Polydispersed Gravity Currents Along a V-Shaped Valley.

Friday, 19 December 2014
Cathy Kurz Besson, Technical University of Lisbon, Instuto Dom Luiz, Faculdade de Ciência da universidade de Lisboa, Lisbon, Portugal and Catherine Anne Marie Dominique Meriaux, Universidade de Lisboa, Lisbon, Portugal
Turbidity currents belong to the class of currents transporting sediments, whose deposits exhibit

downstream grain size fining. In this study, the objective was to better understand the relationship

between downstream fining and grain sizes at the source. To this end, we performed four lock-
exchange experiments of polydispersed and turbulent gravity currents flowing along a 5-m long

V-shaped valley. The particle volumetric concentrations were typically 3%. The four currents

were made of 1) Silicon Carbide (SiC), 2) Glass Beads (GBs), 3) a combined poorly sorted SiC/GBs,

and 4) a moderately sorted combined SiC/GBs. We used the Morphologi G3 tool developed by

Malvern Instruments Corporate (Malvern Instruments Ltd, UK) for the grain size analyses. We first

established a criterion for identifying the appropriate number of grain size classes nbclass

for characterizing the grain size distributions. We considered the four statistical indicators that are

the arithmetic mean size dmean, the standard deviation σd, the skewness Skd, and the kurtosis Kd,

and show that the four indicators for the initial grain size distributions reach plateaux when nb_class

≥ 20. Hence we chose nbclass=20 as being our appropriate bin width. These four indicators were

then calculated for samples taken along the deposits to establish the grain size distributions along

the deposits. The subsequent profiles of dmean, σd, Skd, and Kd with distance from the lock show

highly variable behaviours between the different initial distributions. In particular, the distance

over which the loss of the largest grains of the initial distribution occurs [dmean> dmean(initial)],

can cover up to half the runout length. Curiously, the different rates of fining estimated from the

curves (dmean/ dmean(initial)) as a function of downstream distance (x/x_runout) do not appear to

be well correlated with the sedimentation velocities based on dmean(initial). This is currently being