A Mechanism for Stratifying Lava Flows

Relict lava flows (e.g., komatiites) are often reported to be zoned in the vertical, each zone separated by a sharp contact. Such stratifications in igneous flows, both intrusive and extrusive, can be treated as analogues of suspended loads of sediments in rivers and streams, and hence amenable to quantitative treatment derived for the hydraulic environment as long as dynamic similitude is assured. Situations typically encountered in the hydraulic environment are streams carrying a bed load at the bottom of the stream, the bed load separated by a sharp horizon from a sediment load carried above it. This sediment load may be topped by others of decreasing density as one moves to the surface of the flow, with perhaps the uppermost layer clear of any suspended matter. Rules exist for estimating the thickness D of these loads: one of them is given by D ~ 4.4V3/rgcvs where V is the shear velocity or average velocity of the flow, r = (ρs - ρl)/ρl where ρs is the density of the suspended solid matter, ρl the density of the fluid, g the acceleration of gravity, c the concentration of the particulate content and vs the settling velocity. The settling velocity is secured through Stoke's Law and the velocity of the flow is given by V = R2/3S1/2/n where R is the hydraulic radius, S the gradient along which the fluid flows and n is the Manning Coefficient. In the igneous case, the bed load would be composed of primocrysts, i.e., of the first crystals to come out of solution as the flow cools along its run. This would leave the upper portions of the flow more evolved except perhaps for a quenched crust riding atop the flow. As the viscosity of the flow is dependent not only on temperature but on composition and crystal content, the mean velocity of each layer will be different from the layer above and below it. This requires shear at the interface of adjoining stratifications, which brings into play another mechanism: dispersive pressure (the Bagnold effect). Dispersive pressure will drive primocrysts into boundary layers such as that attending the bottom of the flow and at those separating stratifications. For instance, if the primocrysts were spinals, then a Cr high might be expected at the interfaces separating stratifications. Since the melt throughout is evolving as it moves down stream, compositional variations along strike (as well is in the vertical) might be expected. Application of the above notions falls within the confines of field observation.