Insights on tephra settling velocity from morphological observations.

In this study we present a systematic and detailed morphological characterization of tephra particles from different eruptions (Fontana Lapilli, Masaya, Nicaragua; Keanakāko'i Formation, Kilauea, USA; recent dome explosions of Soufriere Hills volcano, Montserrat) and the calculation of their Terminal Fall Velocity (TFV) as obtained based on different drag prediction models (i.e., [Wilson and Huang, 1979], [Haider and Levenspiel, 1989], [Ganser, 1993] and [Dellino et al., 2005]). In particular, particle sphericity, and, therefore, particle surface area, is essential for the calculation of TFV of irregular-shape particles but is of complex determination. Various attempts have been proposed. According to our results, 2D morphological characterization of volcanic particles is a fast and simple application for a wide range of particle size and provides consistent sphericity and settling-velocity values. 3D scanning also provides a promising tool for lapilli-sized tephra (> 2 cm). In contrast, gas-adsorption-derived surface area is not suitable for the calculation of TFV of volcanic particles mostly because it mainly describes the surface contribution of nanometric pores that are not expected to affect significantly TFV and because bulk-sample analysis is representative of neither individual particles nor of the whole particle population. Settling velocities calculated using values of surface area derived from gas adsorption analyses are up to two orders of magnitude lower than the values obtained through 2D analysis. In addition, our results also show how the influence of particle shape on TFV increases with particle size. In particular, calculated TFV converges at small particle sizes (≥ 3 ϕ) regardless of the model applied, suggesting that the spherical assumption is appropriate for this size fraction (discrepancies with the spherical model are within 10%). Discrepancies with the spherical model increase with particle size up to about 50% and depend on the choice of both the TFV model and the morphological parameterization used. In particular, the drag prediction model of Ganser (1993) is sensitive to the effect of particle morphology on TFV and is well suited for all sizes and Reynolds numbers of typical tephra particles. Finally, our results show how individual size categories (whole- and half-ϕ) are not associated with individual TFV values but with a range of values, which increases with class size. Nonetheless, the half-ϕ system is associated with a smaller standard deviation than the whole-ϕ system, and is therefore more appropriate for the modeling of tephra dispersal. In any case, for dispersal modeling purposes, it is more appropriate to indicate a range of settling velocities for each size class rather than giving an average value