Large explosive volcanic eruptions are commonly associated with caldera subsidence and ignimbrites deposited by pyroclastic currents. Volumes and thicknesses of intracaldera and outflow ignimbrites at 76 explosive calderas around the world indicate that subsidence is commonly simultaneous with eruption, such that large proportions of the pyroclastic currents are trapped within the developing basins. As a result, much of an eruption must penetrate its own deposits, a process that also occurs in large, debris-filled vent structures even in the absence of caldera formation and that has been termed “gargling eruption.” Numerical modeling of the resulting dynamics shows that the interaction of preexisting deposits (fill) with an erupting (juvenile) mixture causes a dense sheath of fill material to be lifted along the margins of the erupting jet. This can cause an eruption that would otherwise produce a buoyant plume and fallout deposits to instead form pyroclastic currents as the dense sheath drives pulsing jet behavior. Increasing thickness of fill amplifies the time variation in jet height. Increasing the fill grain size relative to that of the juvenile particles can result in a much higher jet due to poorer mixing between the dense sheath and the dilute jet core. In all cases, material collapses along the entire height of the dense sheath rather than from the top of a simple fountain. These gargle dynamics provide strong backing for processes that have been inferred to result in intraplinian ignimbrites and simultaneous deposition from high- and low-energy pyroclastic currents.Large-volume, explosive volcanic eruptions eject tens to thousands of cubic kilometers of magma. Calderas—subsidence features formed by rapid evacuation of large volumes of magma—are ubiquitous features of the larger eruptions. Studies of eroded calderas (e.g., Lipman, 2000) and drill cores in young examples (e.g., Nielson and Hulen, 1984) indicate that poorly sorted, pumice-rich deposits of pyroclastic currents (ignimbrites) within many calderas compose significant portions of the total volumes erupted (Fig. 1A; Item S1 in the Supplemental Material1); the intracaldera deposits are normally significantly thicker than ignimbrites that flowed out of the calderas from the same eruptions (outflow; Fig. 1B). These data are consistent with pyroclastic currents having been partially trapped within progressively deepening calderas during eruption and implies that portions of the eruptions had to penetrate their own fresh deposits, a process referred to as “gargling eruption” by Wilson and Hildreth (1997). Eruptions that do not involve major caldera collapse may also have very wide, debris-filled vent structures through which continued eruption must penetrate (e.g., 1912 CE Novarupta vent in Alaska, USA; Hildreth and Fierstein, 2012).We present numerical modeling that explores the effects of gargling on eruption dynamics. Although the simulations are greatly simplified compared to natural cases, they provide strong theoretical backing for processes that have been inferred from field studies and that have major effects on explosive eruptions and their deposits.We model the dynamics of eruption through particle layers with thicknesses between 50 and 100 m that represent freshly deposited caldera-fill ignimbrite or vent debris. Fluid flow is modeled with time-dependent, compressible-flow conservation equations solved for both gas and particles, which are coupled through momentum (drag) and heat exchange (as in Sweeney and Valentine, 2017; Valentine and Sweeney, 2018). The same approach was used to study discrete phreatomagmatic explosions in debris-filled vents (Sweeney and Valentine, 2015; Sweeney et al., 2018), but here we focus on sustained discharges. The simplified two-dimensional (2-D), axisymmetric model domain extends to an altitude of 3 km (5640 m in one case) and radially away from the axis of symmetry to a horizontal distance of 6 km (Fig. 2). The 2-D approach does not account for the three-dimensional (3-D) structure of eddies that govern entrainment; simulated jet heights and transitions from dense jet to buoyant plume are therefore approximate but are expected to be reasonable (e.g., Nourazar and Safavi, 2017). Ambient air in the domain has density and temperature determined initially by the standard atmospheric profile (Sparks et al., 1997). The eruptive mixture of hot particles and H2O vapor enters the domain from the bottom boundary adjacent to the symmetry axis at a constant rate; these particles are referred to as juvenile particles. In all simulations, this inflow boundary (vent) has a 100 m radius. The remainder of the bottom boundary has a no-slip condition while the top and right boundaries allow outflow. Caldera fill is represented by a bed of particles (porosity 40 vol%) that initially extends from the symmetry axis to the inner edge of a caldera rim; these preexisting particles are referred to as fill particles. The rim is simply represented as a rectangular obstacle of a defined height, located between 1000 and 1500 m from the axis. Although natural calderas have more complex 3-D topography and can erupt from vents with a range of sizes, shapes, and distances from developing caldera margins and which can be active individually or simultaneously, it is necessary to use our abstracted approach in order to gain a basic understanding of processes. Governing equations, material properties, information about the numerical code, boundary conditions for simulations, and comparison of grid-size effects are detailed in Item S2.To emphasize the important effects that caldera fill can have on eruption dynamics, we first focus on two simulations that have identical conditions of the erupting mixture (Fig. 2). The first (Fig. 2A) is an eruption into air (no fill); this erupting mixture forms a vertical jet at its base. In the absence of expansion or interaction with surrounding air, this jet would lose its kinetic energy at an altitude of ∼500 m and collapse to the ground to form a fountain structure and outgoing pyroclastic currents. However, in this case, which erupts with a relatively low density of ∼4 kg/m3 (corresponding to 4 wt% H2O and 0.2 vol% particles), the mixture attains an altitude of ∼700 m, where it spreads laterally (Fig. 2A). Interaction with the atmosphere reduces the mixture density so that instead of collapsing to the ground, it rises as a buoyant plume. The dynamics are very different when the mixture erupts through a 50-m-thick bed of fill particles (the mass flux and H2O content are the same as the Figure 2A case, but pressure, gas density, and speed are adjusted to account for the lithostatic load of the fill). In this case, in which the fill particles have the same size and density as the juvenile particles, the erupting mixture emerges from the top of the fill as an over-pressured jet of gas and juvenile particles, which expands and accelerates in its lower hundreds of meters (Fig. 2B). This structure is maintained throughout the simulation because fill particles are constantly closing in over the vent, in effect pinching the basal tens of meters of the jet. The jet drags up (entrains) a dense sheath or annulus of fill particles, which then promotes collapse of the outer margins of the jet. Momentum is variably imparted to material within the sheath, with parcels that are in direct contact with the jet gaining larger upward speeds than on the outside of the sheath; as a result, parcels of gas-particle mixture rise to a range of heights, and collapse of sheath material occurs continuously along its height. This collapsing mixture, which is dominated by recycled fill particles, feeds laterally flowing pyroclastic currents (Figs. 2B and 2C). Collapse of a dense outer sheath contrasts with a simple fountain structure that results when a dense mixture erupts unimpeded into air, where material collapses from the top of a fountain via a stem structure that impacts the ground some distance from the jet (Fig. 2B inset; Valentine et al., 1992; Neri et al., 2003). A buoyant gas-particle mixture rises from the top of the currents and along the outer part of the complex jet structure. Thus, the presence of a relatively thin (50 m) intracaldera deposit can cause pyroclastic currents in an eruption that would otherwise produce a buoyant plume and fallout deposits.Increasing the thickness and grain size of initial fill amplifies the effects described above, including the time-dependent behavior. For example, a simulation with the same eruptive mixture as described above (adjusted to account for different overburden pressure) but with a 100 m initial fill thickness instead of 50 m produces a jet that initially ascends to ∼1.5 km (Fig. 3A). The dense sheath of entrained fill particles results in tendrils that collapse from various heights along the jet margin, much different from a simple fountain. As time progresses, the jet decreases in height to several hundred meters then increases again to ∼1.5 km, producing a pulsing behavior with a period of 20–40 s, which contributes to unsteady flow in pyroclastic currents, along with other factors such as small plumes that rise from the currents.A simulation with the same eruptive mixture as above and with 50 m initial fill thickness (as in Figs. 2B and 2C) but with fill particles having 1 mm diameter instead of 0.1 mm shows that the poorer coupling between gas and larger particles also has an effect on jet dynamics (Fig. 3B). Here the jet drags fill particles upward along its margin as above, but the coarser particles in the dense sheath mix into the core of the jet much more gradually than when they are the same size as the juvenile particles. The juvenile jet core, over-pressured and accelerating to ∼300 m/s as it emerges from the fill, maintains a bulk particle density of 3–4 kg/m3 to a height of ∼1500 m before the coarser fill particles begin to contribute appreciably to its density. Above ∼1500 m, the bulk particle density is ∼5 kg/m3 up to a height of ∼3500 m, where the density abruptly drops to values that are similar to or less than that of the surrounding ambient air (∼0.35 kg/m3). This density drop corresponds to an abrupt transition from negative buoyancy below to positive buoyancy above (Fig. 3B). At the level of the transition, particles fall back toward the ground along the outer margin of the dense sheath. The buoyancy transition maintains a relatively constant height, fluctuating by only a few hundred meters. Material from the dense sheath collapses along its outer margin during the entire evolution of the jet and overlying buoyant plume, feeding thin pyroclastic currents that are dominated by coarser fill particles. The finer juvenile component mainly falls from the jet top, reaching the ground and increasingly contributing to the currents at later times.The simulated eruptions described above illustrate the complexities that can result from eruption through fresh caldera- or vent-fill deposits, i.e., gargle dynamics. First, the presence of fill can cause an eruption that would otherwise produce a buoyant column and fallout deposits to instead collapse and produce pyroclastic currents and ignimbrites (Fig. 2) with no change in mass flux or volatile content. Unlike in simple fountains that form in the absence of fill deposits, material in these eruptions collapses from all heights along the dense sheath of entrained fill particles. Second, modest increases in fill thickness when the fill and juvenile grain sizes are similar can amplify the complex and transient (pulsing) behavior of the erupting jet even if the eruptive mass flux is constant in time. Third, if the fill material is composed of larger (and/or denser) particles, the resulting jet attains greater heights compared to the same fill thickness with similar juvenile and fill grain sizes due to the poorer coupling of larger particles and slower mixing with the fine-grained juvenile jet core. Collapse height and its time variations are expected to influence pyroclastic current behavior and resulting depositional facies.Detailed field studies of large ignimbrites, although small in number, demonstrate that low-energy, dense and hot pyroclastic currents can be coeval with higher-energy currents that are much more mobile with respect to topography and that are emplaced at lower temperatures (Wilson and Walker, 1985; Fierstein and Hildreth, 1992; Wilson and Hildreth, 1997). Wilson and Hildreth (1997) specifically linked such processes recorded in the Bishop Tuff (California, USA) with eruption from a basin (caldera) containing unconsolidated deposits, coining the term “gargling eruption.” As noted above, material in the dense sheath around a gargling eruption can collapse from a continuous range of heights, and it is to be expected that material from a few kilometers' height would feed low-temperature, mobile pyroclastic currents, in contrast with material that collapsed from a few hundred meters, depending upon the fill temperature (e.g., Fig. 3B). While our modeling is axisymmetric, in nature it is reasonable to expect spatial variability in availability of fill or vent and/or caldera wall debris, which would lead to asymmetry in development of the dense sheath and pyroclastic currents. There are likely conditions where the dense sheath on one side of a jet feeds pyroclastic currents while the juvenile jet core, still able to mix with air on its unaffected side, becomes buoyant and produces coeval fallout deposits, resulting in so-called intraplinian pyroclastic currents (see also Wilson and Walker, 1985; Neri and Dobran, 1994; Wilson and Hildreth, 1997; Esposti Ongaro et al., 2008). Similar processes were inferred as the origins of complex interstratified fallout and pyroclastic current deposits in proximal products of the 1912 Novarupta eruption (Alaska), which had a very large, debris-filled vent structure as much as ∼2.5 km wide (Hildreth and Fierstein, 2012). Houghton et al. (2004) and Hildreth and Fierstein (2012) interpreted these deposits to have resulted from collapse of overloaded annular zones caused by recycling of vent fill. Our work provides theoretical backup in favor of such processes; it also strengthens the point made by Houghton et al. (2004) that common density profile assumptions for one-dimensional eruption-column models are incorrect in cases like these.Material in the dense jet sheath that forms during eruption through fill feeds simulated pyroclastic currents with very low juvenile particle contents. At a horizontal distance of 2000 m from the symmetry axis, the poorly resolved, thin pyroclastic current in the coarse-fill case (Fig. 3B) had no juvenile particles at 150 s and contained only ∼1% juvenile particles at 250 s. In contrast, at 150 s, the case with 50-m-thick fill but with equal juvenile and fill particle sizes had ∼4% juvenile particles at the same distance (Figs. 2B and 2C), and the 100-m-thick fill case had ∼2% juvenile particles (Fig. 3A). Thus, thicker and coarser fill material reduces the juvenile component in outgoing pyroclastic currents and their deposits, and in cases where the eruptive jet penetrates very coarse lithic debris, these currents would deposit lithic breccia horizons in the outflow ignimbrite (e.g., Druitt and Bacon, 1986; Yasuda and Suzuki-Kamata, 2018; Valentine et al., 2019). Petrologic and geochemical interpretation of zoned ignimbrites should take these effects into account.Simulations reported in this paper were conducted at the University at Buffalo's (New York, USA) Center for Computational Research. We thank E. Breard, T. Druitt, and reviewers of a previous version of this work for their helpful suggestions.

中文翻译：

---

爆炸性火山口形成喷发和充满碎片的通风口：漱口动力学

大型爆炸性火山喷发通常与火山口下沉和火山碎屑流沉积的熔结岩有关。世界各地 76 个爆炸性火山口的火山口内部和流出的火山灰的体积和厚度表明，沉降通常与喷发同时发生，因此大部分火山碎屑流都被困在发展中的盆地内。因此，大部分喷发必须穿透其自身的沉积物，即使在没有火山口形成的情况下，这个过程也会发生在充满碎片的大型喷口结构中，这被称为“漱口喷发”。由此产生的动力学的数值模拟表明，预先存在的沉积物（填充物）与喷发的（幼年）混合物的相互作用导致沿着喷发射流边缘的填充材料的致密鞘层被提升。这可能会导致喷发，否则会产生漂浮的羽流和沉降物沉积物，因为密集的护套驱动脉冲喷射行为，从而形成火山碎屑流。增加填充厚度会放大射流高度的时间变化。由于致密鞘和稀射流核心之间的混合较差，相对于幼年颗粒的填充粒度增加填充粒度会导致更高的射流。在所有情况下，材料都会沿着致密护套的整个高度而不是从简单喷泉的顶部坍塌。这些漱口动力学为已被推断导致内部火山灰岩和高能和低能火山碎屑流同时沉积的过程提供了强有力的支持。大体积、爆炸性的火山喷发喷出数万至数千立方公里的岩浆。火山口——由大量岩浆快速排出而形成的沉降特征——是大规模喷发的普遍特征。对侵蚀破火山口（例如，Lipman，2000 年）和年轻实例中的钻芯（例如，Nielson 和 Hulen，1984 年）的研究表明，许多破火山口内的火山碎屑流（燃烧岩）的分类不良、富含浮石的沉积物构成了总的重要部分体积爆发（图 1A；补充材料 1 中的项目 S1）；火山口内沉积物通常比从相同喷发中流出火山口的火山灰岩厚得多（流出；图 1B）。这些数据与火山碎屑流在喷发过程中部分被困在逐渐加深的火山口内一致，并暗示部分喷发必须穿透它们自己的新鲜沉积物，Wilson 和 Hildreth (1997) 将这一过程称为“漱口喷发”。不涉及大火山口坍塌的喷发也可能具有非常宽的、充满碎片的喷口结构，持续喷发必须通过这些喷口结构（例如，美国阿拉斯加 1912 年 CE Novarupta 喷口；Hildreth 和 Fierstein，2012 年）。我们提出了探索的数值模型漱口对喷发动力学的影响。尽管模拟与自然情况相比大大简化，但它们为从现场研究中推断出的过程提供了强有力的理论支持，这些过程对爆炸性喷发及其沉积物具有重大影响。我们通过厚度在 50和 100 m，代表新鲜沉积的火山口填充物或喷口碎片。流体流动建模与时间相关，针对气体和粒子求解的可压缩流动守恒方程，它们通过动量（阻力）和热交换耦合（如 Sweeney 和 Valentine，2017 年；Valentine 和 Sweeney，2018 年）。相同的方法被用于研究充满碎屑的喷口中离散的潜水岩浆爆炸（Sweeney 和 Valentine，2015 年；Sweeney 等，2018 年），但这里我们关注的是持续排放。简化的二维 (2-D)、轴对称模型域延伸到 3 公里的高度（在一种情况下为 5640 米）并径向远离对称轴到 6 公里的水平距离（图 2）。2-D 方法没有考虑控制夹带的涡旋的三维 (3-D) 结构；因此，模拟的喷射高度和从密集喷射到浮力羽流的过渡是近似的，但预计是合理的（例如，Nourazar 和 Safavi，2017 年）。域中的环境空气具有最初由标准大气剖面确定的密度和温度（Sparks 等，1997）。热粒子和 H2O 蒸汽的喷发混合物从与对称轴相邻的底部边界以恒定速率进入域；这些粒子被称为幼年粒子。在所有模拟中，该流入边界（通风口）的半径为 100 m。底部边界的其余部分具有无滑移条件，而顶部和右侧边界允许流出。Caldera 填充物由最初从对称轴延伸到火山口边缘内边缘的颗粒床（孔隙率 40 vol%）表示；这些预先存在的粒子称为填充粒子。边缘简单地表示为一个定义高度的矩形障碍物，位于距轴 1000 到 1500 m 之间。尽管天然火山口具有更复杂的 3-D 地形，并且可以从具有各种尺寸、形状和距离发展中的火山口边缘的喷口喷出，并且可以单独或同时活跃，但有必要使用我们的抽象方法以获得对流程的基本了解。控制方程、材料特性、有关数值代码的信息、模拟的边界条件以及网格尺寸效应的比较在第 S2 项中有详细说明。为了强调火山口填充物对喷发动力学的重要影响，我们首先关注两个模拟它们具有与喷发混合物相同的条件（图 2）。第一个（图 2A）是喷发到空气中（无填充）；这种喷发的混合物在其底部形成垂直射流。在没有膨胀或与周围空气相互作用的情况下，该射流将在约 500 m 的高度失去其动能并坍塌到地面，形成喷泉结构和向外流出的火山碎屑流。然而，在这种情况下，喷发的密度相对较低，约为 4 kg/m3（对应于 4 wt% H2O 和 0.2 vol% 颗粒），混合物达到约 700 m 的高度，并在那里横向扩散（图 3）。 2A)。与大气的相互作用降低了混合物的密度，因此它不会坍塌到地面，而是作为一个漂浮的羽流上升。当混合物通过 50 米厚的填充颗粒床喷发时，动力学非常不同（质量通量和 H2O 含量与图 2A 的情况相同，但调整了压力、气体密度和速度以考虑填充物的岩石静载荷）。在这种情况下，在填充颗粒与幼年颗粒具有相同大小和密度的情况下，喷出的混合物作为气体和幼年颗粒的超压射流从填充物顶部喷出，在其下方数百米处膨胀和加速（图.2B)。这种结构在整个模拟过程中都保持不变，因为填充颗粒不断地靠近通风口，实际上挤压了喷流的基础数十米。射流向上拖动（夹带）填充颗粒的致密鞘或环，然后促进射流外缘的塌陷。动量被不同程度地传递给护套内的材料，与射流直接接触的包裹获得比护套外部更大的向上速度；结果，气体颗粒混合物的包裹上升到一定范围的高度，并且护套材料的塌陷沿其高度连续发生。这种以回收填充颗粒为主的坍塌混合物供给横向流动的火山碎屑流（图 2B 和 2C）。致密外护套的坍塌与简单的喷泉结构形成鲜明对比，当密集的混合物不受阻碍地喷发到空气中时，材料从喷泉顶部通过杆结构坍塌，杆结构撞击地面距离喷射器有一段距离（图 2B 插图） ；瓦伦丁等人，1992 年；内里等人，2003 年）。一种漂浮的气体粒子混合物从气流的顶部沿着复杂的喷射结构的外部上升。因此，相对薄 (50 m) 的火山口内沉积物的存在会在喷发中引起火山碎屑流，否则会产生浮力羽流和沉降物沉积物。增加初始填充的厚度和晶粒尺寸会放大上述影响，包括与时间相关的行为。例如，使用与上述相同的喷发混合物（根据不同的上覆压力进行调整）但初始填充厚度为 100 m 而不是 50 m 的模拟产生的射流最初上升到约 1.5 km（图 3A）。夹带填充颗粒的致密鞘导致卷须沿着喷射边缘从不同高度坍塌，这与简单的喷泉大不相同。随着时间的推移，射流的高度降低到几百米，然后再次增加到 1.5 公里，产生周期为 20-40 秒的脉冲行为，这有助于火山碎屑流的不稳定流动，以及其他因素，如小从水流中升起的羽流。使用与上述相同的喷发混合物和 50 m 初始填充厚度（如图 2B 和 2C 所示）但填充颗粒直径为 1 mm 而不是 0.1 mm 的模拟表明，气体和较大颗粒之间的较差耦合也具有对射流动力学的影响（图 3B）。在这里，喷射流沿其边缘向上拖动填充颗粒，如上图所示，但与幼年颗粒大小相同时，致密鞘层中的较粗颗粒混合到喷射流的核心中要缓慢得多。当从填充物出现时，超压并加速到约 300 m/s 的幼年喷射核心在较粗的填充物颗粒开始贡献之前保持 3-4 kg/m3 的堆积颗粒密度到 1500 m 的高度可观其密度。在 ∼1500 m 以上，堆积颗粒密度为∼5 kg/m3 到∼3500 m 的高度，其中密度突然下降到接近或小于周围环境空气的值（~0.35 kg/m3）。这种密度下降对应于从下方负浮力到上方正浮力的突然转变（图 3B）。在过渡层，粒子沿着致密鞘的外缘落回地面。浮力过渡保持相对恒定的高度，波动只有几百米。在喷射流和上覆浮力羽流的整个演化过程中，来自致密鞘层的物质沿着其外缘坍塌，供给以较粗填充颗粒为主的薄火山碎屑流。较细的幼体部分主要从喷气机顶部落下，到达地面并在以后越来越多地对洋流做出贡献。上述模拟喷发说明了通过新鲜的火山口或喷口填充沉积物喷发可能导致的复杂性，即漱口动力学。首先，填充物的存在会导致喷发，否则会产生浮柱和沉降物沉积物倒塌并产生火山碎屑流和熔结岩（图 2），而质量通量或挥发分含量没有变化。与在没有填充物沉积物的情况下形成的简单喷泉不同，这些喷发中的物质沿着夹带填充颗粒的致密鞘从各个高度坍塌。其次，当填充物和幼年颗粒尺寸相似时填充物厚度的适度增加可以放大喷发射流的复杂和瞬态（脉冲）行为，即使喷发质量通量在时间上是恒定的。第三，如果填充材料由更大（和/或更密）的颗粒组成，由于较大颗粒的耦合较差以及与细颗粒的混合较慢，因此与相同的填充厚度相比，所产生的射流将获得更高的高度，并且具有相似的幼龄和填充颗粒尺寸。粒状少年喷射核心。坍缩高度及其时间变化预计会影响火山碎屑流的行为和由此产生的沉积相。 对大型火山灰岩的详细现场研究，虽然数量很少，但表明低能量、密集和热的火山碎屑流可以与更高能量的流同时存在。相对于地形而言，它们的移动性要大得多，并且在较低温度下放置（Wilson 和 Walker，1985；Fierstein 和 Hildreth，1992；Wilson 和 Hildreth，1997）。Wilson 和 Hildreth（1997）将 Bishop Tuff（美国加利福尼亚州）记录的此类过程与包含松散沉积物的盆地（火山口）的喷发联系起来，创造了“漱口喷发”一词。如上所述，漱口喷发周围致密护套中的物质可以从连续的高度范围内坍塌，预计几公里高度的物质将供给低温、流动的火山碎屑流，与物质形成对比。取决于填充温度（例如，图 3B），从几百米处坍塌。虽然我们的建模是轴对称的，但在本质上，可以合理地预期填充或通风孔和/或火山口壁碎片的可用性的空间变化，这将导致致密鞘流和火山碎屑流发展的不对称。在某些情况下，喷流一侧的致密鞘层会供给火山碎屑流，而年轻的喷流核心仍然能够与其未受影响的一侧的空气混合，变得有浮力并产生同时期的沉降沉积物，从而导致所谓的内层火山碎屑流（另见 Wilson 和 Walker，1985；Neri 和 Dobran，1994；Wilson 和 Hildreth，1997；Esposti Ongaro 等，2008）。类似的过程被推断为 1912 年 Novarupta 喷发（阿拉斯加）的近端产物中复杂的层间沉降物和火山碎屑流沉积物的起源，它有一个非常大的、充满碎屑的喷口结构，宽约 2.5 公里（Hildreth 和 Fierstein， 2012）。霍顿等人。（2004 年）以及 Hildreth 和 Fierstein（2012 年）将这些沉积物解释为由排放口填料回收引起的超载环形带坍塌造成的。我们的工作为支持此类过程提供了理论支持；它还加强了 Houghton 等人的观点。(2004) 一维喷发柱模型的常见密度剖面假设在此类情况下是不正确的。在喷发过程中通过填充物形成的致密喷射鞘中的材料为模拟火山碎屑流提供了非常低的幼年颗粒含量。在距对称轴 2000 m 的水平距离处，粗填充情况下解析不佳的薄火山碎屑流（图 3B）在 150 s 时没有幼年颗粒，在 250 s 时仅包含约 1% 的幼年颗粒。相比之下，在 150 秒时，填充 50 米厚但幼虫和填充粒子大小相同的情况下，相同距离处的幼虫颗粒约为 4%（图 2B 和 2C），而 100 米厚的填充物情况下，幼虫颗粒约为 2% （图3A）。因此，较厚和较粗的填充材料减少了流出的火山碎屑流及其沉积物中的幼年成分，并且在喷发射流穿透非常粗的岩屑碎片的情况下，这些流会将岩屑角砾岩层沉积在流出的熔凝岩中（例如 Druitt 和 Bacon， 1986 年；Yasuda 和 Suzuki-Kamata，2018 年；Valentine 等人，2019 年）。分区熔凝灰岩的岩石学和地球化学解释应考虑这些影响。本文报告的模拟是在布法罗大学（美国纽约）计算研究中心进行的。我们感谢 E. Breard、T. Druitt、