V18-1185.00 Erosion and re-deposition by pyroclastic flows: implications of experimental evidence for reworking of loose substrates and primary deposits. Rowley, P. J. 1 , Menzies, M. 1 , Kokelaar, P. 2 and Waltham, D. 1 Abstract Pyroclastic deposits can remain unconsolidated for long periods after deposition (e.g. Taupo ignimbrite, New Zealand c.1.8ka). Subsequent erosion, entrainment and remobilisation of these loose deposits by younger flows can (a) reset depositional thickness, (b) reshape palaeotopographies, (c) redistribute charcoals and phenocrysts used for absolute dating and (d) mix juveniles used in understanding provenance. An experimental approach was used to assess the interaction between dry granular charges and static loose substrate materials. Flume experiments used a lock-gate release of coloured dry granular bead charges, deposited onto a variety of loose granular substrates. The results revealed the scale of erosion from granular temperature conduction from the “younger” flow into the “older” substrate. Flow thicknesses of approximately 10 mm removed an equivalent thickness of substrate in highly unsteady flow conditions. Furthermore, experiments using polymict charges assessing particle size and density reproduce a range of internal stratigraphies observed in field deposits, and grant insight into the depositional mechanisms active in these non-uniform unsteady dry granular currents. Analogue experiments demonstrate that reworking by dry granular flow can reproduce “ignimbrite stratigraphy” identical to the primary deposits formed by the fountaining of eruptive columns, including inverse graded “pumices”/ low density particles and normal grading “lithics”/high density particles Retrogressive failure of an existing ignimbrite deposit can • generate secondary ignimbrite architectures/stratigraphies which are structurally and chemically indistinguishable from the primary ignimbrite deposit produced by the collapse of low or high fountains/eruptive columns. Shearing granular flows are capable of remobilising • large quantities of loose substrate material through granular temperature conduction (i.e. momentum transfer between mobile and stationary particles). Mixing may move particles from substrate material into • active flow, which subsequently would provide anomalous data for Temperature proxy data from included charcoals » Dating from included phenocrysts » Provenance from included juvenile clasts » Mixing may mask unit boundaries, impacting the reliability • of eruption, flow and deposit volume estimates, as well as flow propagation extents. Method The experiments are conducted in a custom-built box-section flume with lock-gate release apparatus. The 2 m long, 150 mm wide flume comprises a 700 mm long 30 degree slope, which transitions through a 500 mm radius curve to the 5 degree inclined runout surface. 0.250 mm diamter silica beads are used as the bulk flow material, 1.00 mm diameter silica beads representing large clasts (e.g. pumice clasts), while 0.250 mm and 1.00 mm diameter ceramic beads are used to investigate the behaviour of denser particles (e.g. lithic clasts). Varied colouration of substrates and sequential charges is used to distinguish mixing and sorting effects. 500 ml charges typically flow and and deposit within 1.3 seconds from release, with flow thickness varying between 0-15 mm. Introduction Scaled (see Figure 1) analogue experiments have been conducted using a variety of silica and ceramic beads in order to investigate granular sorting and substrate interaction effects in dense granular flows, to inform our understanding of processes at the basal contact in a range of geophysical currents. Pyroclastic flows, debris flows and similar shearing currents pose a significant risk to populations around the world, and developing a better understanding of the flow and deposition processes is fundamental to managing that risk (Baxter, Neri, & Todesco, 1998). Granular sorting Experiments demonstrate that well mixed poly-disperse charges are able to undergo extreme granular sorting in the very short duration and distance of these experiments. ‘Brazil Nut Effect’ (BNE), Reverse BNE (RBNE), and same-size density sorting are all observed (Figures 2A and 2B). BNE sifts larger same-density clast particles to the upper surface of the deposit (Figure 2A), often achieving monomict rafts on the top of the sediment pile. Approximately 10% of the large silica bead population was lost by the flow front, as the smooth hard flume surface encouraged bouncing of particles. When their concentration was low there was insufficient grain- grain interaction to retard these beads movement, so they were able to accelerate away from the main body of the flow. These particles might otherwise have been expected to form a distal toe of BNE sorted clasts on the upper front of the deposit. RBNE sorting of large dense particles concentrates the ceramic beads towards the centre of the flume, and produces the normal grading typical of dense clast particles within geophysical flows. Same-size high density particles follow the same concentration patterns as the RBNE sorting (Figure 2B). In this case the finer resolution provided by smaller particles highlights the deposit architecture formed by unsteady dense granular flow - i.e. numerous stacked shear planes, in line with findings by Shea & van Wyk de Vries, 2008. Reworking Despite the unsteady and brief nature of the flows, they are able to mobilise substantial thicknesses of substrate material even at low angle interfaces (Figure 3). The shearing basal flow contact encourages roll-up of vortical features, and the remobilisation and transport of substrate material downstream of its origin. Particles are lifted from the substrate into the over- riding flow by up to 10 mm, with downstream transport in the order of 50-100 mm (Figure 4). The advection and mixing of material from the upper surface of the substrate results in indistinct amalgamated contacts. Erosion is concentrated in the high-energy centre of the flume, and is associated with the transport of substrate material obliquely toward the edges where the lower velocity sidewall flow demonstrates significantly less erosive capability. Discussion These brief, unsteady and thin scaled flows are able to capture the instantaneous behaviours which occur within more sustained dense granular flows. The contrasting colouration of flow and substrate units delineates boundaries and highlights mixing more clearly than might be observed in geophysical environments, where sequential units often have very similar visual properties. The ability of homogenously mixed polydisperse charges to undergo near complete BNE, RBNE and density sorting in under two seconds of flow implies that deposits with normal grading of dense particles and inverse grading of large particles (often seen as a ‘classic’ primary ignimbrite stratigraphy) may result from deposition of secondary flows, for example those formed by retrogressive collapse of unstable primary deposits (see Figure 5, and Branney & Kokelaar, 2002). The ability of these unfluidised cold flows to sort also suggests that collapse Conclusions Figure 1. Dimensionless scaling analysis of flow parameters, comparing the flume experiments with debris and pyroclastic flow ranges (collated from Rowley 2010) References Branney, M. J., & Kokelaar, P. (2002). Pyroclastic density currents and the sedimentation of ignimbrites (Vol. 27). The Geological Society of London. Rowley, P. J. (2010) Analogue modelling of pyroclastic density current deposition. Ph.D. Thesis, London, UK Shea, T., & van Wyk de Vries, B. (2008). Structural analysis and analogue modeling of the kinematics and dynamics of rockslide avalanches. Geosphere, 4(4), 657-686. Figure 6. Locally broken and diffuse pyroclastic flow contact demonstrating remobilisation of clasts from the lower pumice fall layer into the over-riding pyroclastic flow through eroded channels in the intermediary ash fall layer. Direction of flow toward the reader. Bandas del Sur, Tenerife, Spain. A B Turbulent overpassing ash cloud Basal granular current Figure 5 - Retrogressive collapse from either single (A) or multiple (B) loose sequences are able to sort during flow to deposit as a single secondary deposit with the same broad stratigraphy typical of primary deposits. Very young (<weeks) deposits may maintain significant heat and ongoing outgassing of juvenile components, but otherwise fluidisation and elutriation of fines (and therefore volume / presence of the turbulent overpassing ash cloud) is very limited. Figure 3. Edge (A) to centre (H) sections at 10 mm intervals through the deposit, demonstrating interaction of a single monomict charge (red) run onto a horizontal stratified substrate. Edge section is unaffected by erosion and demonstrates the original substrate topography. Centre of the flume is dominated by a roll-up structure with some secondary shear deformation. Flow L-R. Figure 2. Edge (a) to edge (p) sections at 10mm intervals through the deposit, demonstrating sorting of polymict charges. A - sorting of 1 mm diameter silica beads from 0.250 mm diameter silica beads, demonstrating BNE sorting. B - sorting of 0.250 mm ceramic beads from 0.250 mm silica beads, demonstrating density sorting. The multiple inclined concentrations within the deposit represent ceramic-rich flow bases, highlighting a stacked shear formation mechanism for the deposit. a b c d e f h A B Figure 4. Single centre-flume slice from a polymict 500 ml charge (blue) containing (orange) large particles run into the flume, followed by a 500 ml green monomict charge. Substrate-flow contact is deformed by smearing of the blue material and the lifting of large clasts into the overpassing flow. Flow L-R. smearing of substrate at proximal end Large clasts from charge 1 reworked into charge 2. g of multiple stacked stratigraphies might result in single flow deposit architectures further downslope (Figure 5B). These experiments demonstrate erosion, mixing, and deformation at the contact plane between dense granular flows and loose substrate. The erosion is capable of removing thicknesses of substrate material in the order of the flow thickness, even in these highly unsteady currents. The mixing is able to produce diffuse contacts, commonly observed and referred to in geophysical deposits as ‘amalgamated’ contacts. The degree of mixing between flow units produced in these experiments, and implied for more sustained flows suggests that apparently gradational single units may in fact represent two separate events with a well mixed contact. This impacts the reliability of deposit volume calculations, and hence any derived values (e.g. eruption volume and magnitude). Furthermore, the transfer of clasts from lower units into the over-riding flow has implications for their use in deposit analysis; the use of included charcoals, phenocrysts or juvenile fragments for dating or provenance studies of pyroclastic deposits is dependant on unmixed depositional units. The tendancy for sequential flows to have similar broad chemistry and appearance results in cryptic mixing, therefore masking the true deposit structure. Figure 6 highlights a field example of erosion by pyroclastic flow through an ash layer, and the reworking of pumice clasts from below into the over-riding flow. Distinguished here by colour variation, in more similar stratigraphic units and with the ash layer absent (e.g. by more extreme erosion) this might otherwise be interpreted as a single gradational layer. 0 50 50 Scale (mm) 0 50 50 Scale (mm) 0 50 50 Scale (mm) Inverse graded pumice Normal graded lithics 0 0 0 0 0 0 0 0 0.5 0.03 0.005 0.025 0.5 45 45 1 1x10 2 90 90 1 0.05 0.01 0.06 1 1x10 1x10 1x10 1x10 1x10 Height Flow thickness Particle density Particle diameter Solid volume fraction Internal friction angle Runout slope Cohesion Viscosity