Fig. 5: A grain size fractionation model for Cordón Caulle volcano.

a Compiled data in this study reflect a captured in-conduit fine ash fraction, characterised by a mean (as-measured) diameter d of 9.94 \(\times {10}^{-7}\) m. For each dataset, best-fit lognormal curves have been overlain. Distribution assumed by Reckziegel et al.³¹ are approximately unimodal and described by mean of d = 3.10 \(\times {10}^{-4}\) m (based on lognormal assumption of the ash mass fraction). Data of Costa et al.³⁰, reconstructed from field data, are bimodal, described by lognormal peaks at d = 1.34 \(\times {10}^{-5}\) and d = 5.39 \(\times {10}^{-3}\) m. Mean and \(\pm 2\sigma\) range are highlighted for data of this study and ref. 31; these values are shown for each of the peaks of ref. 30. Note different bin size and y axes between panels. b In-conduit processes summarised schematically. Magmatic particles are transported in a gas phase at a mean velocity of \(\langle u\rangle\). Flow turbulence results in clustering and decoupling of particles (1) of St \(\ge 1\) from the eddy motion, in addition to turbophoresis (2) which results in elevated concentration of particles at the wall. Particles may then impact the wall (3) at velocity \({u}_{\theta }\)—dependent on \(\theta\)—then sinter rapidly (4). c Energetic fragmentation within the conduit generates a population of both fine and coarse particles; due to the processes outlined in b, the relatively fine fraction is preferentially captured at the walls of ash vents; the emitted ash therefore predominantly reflects the coarser fraction of the original population. Conceptual model of the shallow subsurface depicted in c is from the so-called cryptic fragmentation model for silicic hybrid and effusive eruptions described in ref. 2. Scales are approximate, as the crater—and presumably the conduit—diameter and morphology changed throughout the course of the eruption.