| 23 Sep 2026
Ash Fingers Modify Tephra Sedimentation During a Coarse-Ash-Dominated Eruption: Evidence from the 23 July 2024 Etna Paroxysm
Abstract. Ash fingers are descending particle-laden plumes that develop beneath volcanic clouds and have been observed during several explosive eruptions worldwide. By promoting collective settling, they modify the atmospheric transport and deposition of volcanic ash and remain a significant source of uncertainty in tephra-dispersal forecasting. However, direct measurements of ash-finger properties are rare, and their depositional signature remains poorly constrained. Here, we investigate ash fingers generated during the 23 July 2024 paroxysmal eruption of Mt. Etna (Italy) using high-resolution visible-wavelength video observations, detailed deposit characterization from ground-based and drone-collected samples, thin-section analyses, disdrometer observations and numerical simulations. The resulting deposit is dominated by coarse ash and shows no evidence of aggregation, providing a unique opportunity to isolate the influence of ash fingers on sedimentation processes. Our observations demonstrate that ash fingers can develop in the absence of abundant fine ash (<63 µm), challenging the commonly held assumption that fine particles are required for their formation. Comparison between observed ash-finger descent velocities and settling velocities calculated for irregular particles indicates that ash fingers were capable of entraining particles up to ~950 µm in diameter. This threshold is highly sensitive to particle shape, highlighting the importance of non-spherical settling behaviour in controlling ash-finger transport. Particle concentration also appears to be a primary control on finger formation, as ash fingers were observed only during phases characterized by high mass eruption rates (MER > 2.5 × 10⁵ kg s⁻¹). Deposit analyses show that ash fingers locally modified proximal sedimentation by increasing the proportion of coarse ash and shifting grain-size distributions toward finer values than expected from individual particle settling. In contrast, simulations of TEPHRA 2, a tephra dispersion model, predict only minor differences between scenarios with and without ash-finger settling. This limited sensitivity could arise because the model assumes spherical particles, leading to an underestimation of the grain-size fraction susceptible to ash-finger entrainment. The discrepancy highlights a fundamental limitation of standard dispersal models based on the assumption of individual settling of spherical particles and suggests that particle morphology must be explicitly considered to realistically represent ash-finger transport and deposition. These findings provide new constraints on ash-finger dynamics and demonstrate that particle shape exerts a first-order control on the interaction between collective sedimentation processes and tephra deposition, with important implications for volcanic ash forecasting and hazard assessment.
Summary and general assessment
The manuscript combines video observations, field and drone sampling, particle characterization, disdrometer measurements, a WRF reconstruction and TEPHRA2 simulations for the 23 July 2024 Etna paroxysm. The authors interpret descending particle-rich structures beneath the cloud as settling-driven ash fingers and conclude without a solid support that they can form without abundant fine ash, that particle shape exerts a first-order control, and that ash fingers are a key sedimentation mechanism in mafic explosive eruptions.
The observational dataset is potentially valuable, but the physical interpretation is not adequately supported. The manuscript treats the applicability of a framework derived largely from analogue experiments with controlled suspensions as established. Morphological similarity, however, does not demonstrate that the same instability controls structures beneath turbulent volcanic clouds. Alternative explanations are just mentioned but not tested against the data, particularly the eddy-controlled particle heterogeneity documented by James et al. (2022) and applied to a volcanic plume by Mingotti and Woods (2024). In addition, the modified TEPHRA2 simulations produce only minor changes and do not improve agreement with the deposit. These results do not support several of the stronger mechanistic claims in the abstract and conclusions.
The central issue is therefore not whether finger-like structures were observed, but whether the available evidence identifies their governing mechanism and demonstrates a significant effect on tephra deposition. In its present form, the manuscript does neither with sufficient confidence. The main issues are detailed below.
Major comments
The manuscript must engage quantitatively with James et al. (2022) and Mingotti and Woods (2024). James et al. show that turbulent plumes in a cross-flow develop particle-rich and particle-poor regions whose scale grows with distance from the source (δ/z = 0.84 ± 0.10), including in single-phase plumes. Particles sedimenting from these clusters fall at their individual Stokes velocity: clustering produces fluctuations in particle load but no measurable enhancement of the mean settling speed, unlike the convective sedimentation described by Hoyal et al. (1999). Mingotti and Woods show that the 2010 Eyjafjallajökull plume, including the sequence used by Manzella et al. (2015), behaves quantitatively like these laboratory plumes, with its shape and concentration structure modulated by the dominant eddies and rapid adjustment to wind.
This framework provides a direct alternative explanation for the Etna observations: particle-rich structures may reflect the heterogeneous mass distribution generated by dominant plume eddies rather than a settling-driven gravitational instability. It is consistent with their scale and onset relative to the cloud, their occurrence without abundant fine ash, and their descent at approximately the individual settling velocity of the particles carried. The authors should compare the predictions of the two mechanisms using their measurements of finger spacing and coarsening, width relative to cloud thickness, descent velocity relative to both particle and turbulent velocity scales, and particle concentration and grain size inside and outside the structures. If the available data cannot distinguish the mechanisms, this uncertainty must be stated. A citation or brief mention of the alternative would not be sufficient.
The estimated onset of the first structures about 1.6 km from the source, at a distance comparable with the cloud scale, is particularly relevant to this comparison. The manuscript currently uses this observation only to suggest that wind-driven rolls might promote a gravitational instability. It should instead test whether the measured onset, width and spacing follow the eddy-scale relationship documented for plumes in a cross-flow. The characteristic scale h should also be defined consistently: Sect. 4.1 appears to use a width of about 1.3 km, whereas Table 1 reports a cloud thickness of 1.8–2.1 km.
Both interpretations draw partly on laboratory studies, and volcanic plumes introduce additional complexity. The manuscript should therefore explain why the experiments supporting the settling-driven interpretation are considered applicable to Etna while the plume-in-cross-flow results are not. Relevant literature on wind-blown plumes and turbulence-induced clustering should be treated consistently rather than mentioned only in passing.
The formation of fingers without abundant fine ash is presented as a surprising result that challenges a prevailing assumption. It is surprising only within the settling-driven framework adopted by the authors; an eddy-controlled mechanism predicts no grain-size requirement. The coarse, aggregate-free character of the eruption may therefore challenge the assumed mechanism rather than simply extend it to coarser particles.
The measurements make this issue especially clear: the deposit has a median grain size of about 860 µm, less than 5 wt% is finer than 355 µm, the OPC data show no substantial fine-ash population, and no aggregates were identified. These observations are important, but they do not by themselves demonstrate a coarse-particle extension of a settling-driven instability. They are equally, and perhaps more directly, compatible with structures whose scale is imposed by plume turbulence rather than particle size.
The manuscript must also distinguish ash fingers from aggregation in terms of depositional consequences. Aggregation selectively transfers fine ash into larger, typically faster-falling aggregates and can therefore strongly modify the distal deposit. If ash fingers do not selectively act on grain size, their main effect should instead be a redistribution of mass in space and time, potentially increasing deposition over proximal distances but not necessarily changing the distal tail. This distinction is particularly important because the simulations show only a limited response to the finger parameterization. The authors should identify a measurable depositional signature that separates ash-finger transport from aggregation and establish whether the proposed process affects distal dispersal or mainly redistributes rapidly settling coarse particles near the plume.
The dmax calculation does not resolve this question and does not identify the formation mechanism. By defining Vp(dmax) = Vf, it only determines which particles could remain within any structure descending at the measured speed. Moreover, dmax-ia (785 µm) is practically identical to the deposit median (about 860 µm), implying that the structures descend at approximately the individual terminal velocity of much of the transported mass. An enhancement relative to individual settling, the defining feature of collective sedimentation, is therefore not demonstrated. Similarly, the relationship in Fig. 9 between median grain size and the fraction finer than dmax is not independent evidence because both variables derive from the same grain-size distribution, while the fining of the disdrometer signal coincides with the eruptive peak.
Increasing dmax to about 950 µm by adopting a sphericity of 0.7 broadens the potentially entrained fraction but does not alter this logical limitation: the calculation concerns retention within a descending structure, not the process that created it. A necessary kinematic condition should not be presented as evidence for a specific instability.
The claim that particle concentration is a primary control rests mainly on the observation that fingers occurred when MER exceeded 2.5 × 10⁵ kg s⁻¹. This inference is not robust. MER is calculated from plume height, so the apparent relationship is also a relationship with plume and cloud scale, as expected under an eddy-controlled interpretation. Only two fingers were measured, and visibility varied with backlighting, making the absence of fingers at lower MER difficult to establish. In addition, Mingotti and Woods (2024) estimate 0.6–1.1 × 10⁴ kg s⁻¹ for the Eyjafjallajökull sequence containing the observations of Manzella et al. (2015), substantially below the value used here to support an MER threshold. Although the estimates concern different methods and times, their disparity shows that such a threshold is not a reliable discriminant.
More fundamentally, a plume-integrated MER cannot represent the local concentration relevant to an instability (apart that MER description should be treated, for what is, i.e. as a statistical variable with huge variability). Turbulent eddies can produce local concentrations that differ substantially from the mean, whereas laboratory onset criteria refer to controlled particle layers. The authors should compare their estimated concentration inside the fingers with the surrounding cloud and with appropriate instability criteria for particles of this size. They should also acknowledge the confounding between MER, plume height and observational visibility rather than interpreting MER directly as a concentration threshold.
Figure 11 and Sect. 3.6 show that enhanced settling produces only minor differences from the individual-settling simulation and does not improve the fit to the deposit. Both simulations overestimate mass loading by more than 50% at the proximal sites C2–C5 and C9–C11 and underestimate it at the more distal sites C12 and C14. This significant negative result should be central to the interpretation. The manuscript instead attributes the weak response mainly to spherical particles and presents particle shape as a first-order control, although no simulation using the irregular-particle threshold was performed; Sect. 4.5 acknowledges that doing so is not straightforward. The modified fraction is also less than 5 wt% because the deposit is coarse. The particle-shape claim is therefore based on the dmax calculation rather than demonstrated by the transport modelling. Model performance should be quantified using site-by-site mass-loading and grain-size misfits, and sensitivity to other uncertain inputs should be assessed before isolating particle shape as the explanation.
The treatment of meteorology and plume dynamics also requires clarification. WRF provides a three-dimensional, time-dependent reconstruction at 1.1 km resolution, but the plume model used to derive MER employs a standard temperature profile and a single average wind speed; image correction uses ERA5 wind directions; and Sect. 2.7 does not state which fields drive TEPHRA2. The authors should specify exactly which WRF variables are used, how they are averaged or reduced, and which components are omitted.
In its standard formulation, TEPHRA2 is a quasi-steady advection–diffusion model using a horizontally uniform wind profile. It does not resolve vertical air velocity, time-dependent wind, coherent eddies, vertically directed turbulence or particle–eddy interactions—the processes most relevant to the formation of the observed structures. Applying an enhanced descent velocity to all particles finer than dmax presupposes the settling enhancement being tested. It also differs fundamentally from aggregation, which modifies the grain-size spectrum selectively. TEPHRA2 may be useful for exploring the consequence of an imposed settling rule, but it cannot identify the physical origin of the structures. The high resolution of the WRF calculation must not be presented as if it removed the stronger approximations of the plume and transport models.
This distinction also limits the hazard implications. A parameterization that modifies proximal mass redistribution without reproducing selective removal of fine ash cannot be assumed to alter distal dispersal in the manner associated with aggregation. The model results should be used to delimit, rather than amplify, the scale at which an ash-finger effect is supported by the data.
The claims that particle shape exerts a first-order control, that the deposit provides direct evidence of finger-driven sedimentation, and that ash fingers are a primary or key collective-sedimentation mechanism in mafic eruptions exceed what is supported by one event, two measured fingers, simulations showing almost no effect, and no systematic discrimination among mechanisms. The westward fining of the proximal deposit is also not uniquely attributable to fingers because the same wind shear acts on individually settling particles and the baseline model does not reproduce the proximal deposit. These statements should be removed or substantially qualified.
Essential revisions
Overall assessment
The observational dataset may merit publication in scientific journals, but the present analysis does not establish either the physical origin of the descending structures or a significant influence on the final deposit. It should be clarified that the results are at least as consistent with an interpretation based on dominant turbulent eddies and heterogeneous particle distribution rather than the settling-driven instability adopted by the authors. Consideration of this alternative, using the authors’ own data, and substantial revision of the mechanistic and depositional claims are necessary before the manuscript can be evaluated further.
References
Hoyal, D.C.J.D., Bursik, M.I. and Atkinson, J.F. (1999) The influence of diffusive convection on sedimentation from buoyant plumes. Marine Geology 159, 205–220.
James, C.B.G., Mingotti, N. and Woods, A.W. (2022) On particle separation from turbulent particle plumes in a cross-flow. Journal of Fluid Mechanics 932, A45. https://doi.org/10.1017/jfm.2021.1065
Manzella, I., Bonadonna, C., Phillips, J.C. and Monnard, H. (2015) The role of gravitational instabilities in deposition of volcanic ash. Geology 43(3), 211–214.
Mingotti, N. and Woods, A.W. (2024) Video-based measurements of the entrainment, speed and mass flux in a wind-blown eruption column. Communications Earth & Environment 5, 237. https://doi.org/10.1038/s43247-024-01402-x