More than 800 million people live in proximity to active volcanoes and could be directly impacted by potential eruptions. Mitigation of future volcanic hazards requires adequate warning of a pending eruption, which, in turn, requires detailed understanding of the fundamental processes driving volcanic activity. In this Review, we discuss the processes leading up to volcanic eruptions, by following the journey of magma from crustal storage zones to the surface. Magma reservoirs can feed volcanic eruptions if they contain sufficiently hot and mobile magma and are able to supply sufficient energy for the magma to reach the surface. Young volcanic plumbing systems favour volcanic activity, whereas storage becomes more likely in mature volcanic systems with large reservoirs (hundreds of cubic kilometres). Anticipating volcanic activity requires a multidisciplinary approach, as real-time monitoring and geophysical surveys must be combined with petrology and the eruptive history to understand the temporal evolution of volcanic systems over geological timescales. Numerical modelling serves to link different observational timescales, and the inversion of data sets with physics-based statistical approaches is a promising way forward to advance our understanding of the processes controlling recurrence rate and magnitude of volcanic eruptions.

超过 8 亿人居住在活火山附近，可能会直接受到潜在火山爆发的影响。减轻未来的火山灾害需要对即将发生的喷发进行充分警告，而这反过来又需要详细了解驱动火山活动的基本过程。在这篇综述中，我们通过追踪岩浆从地壳储存区到地表的旅程来讨论导致火山爆发的过程。如果岩浆储层含有足够热和流动的岩浆，并且能够为岩浆到达地表提供足够的能量，它们就可以为火山喷发提供能量。年轻的火山管道系统有利于火山活动，而在具有大型水库（数百立方公里）的成熟火山系统中，储存变得更有可能。预测火山活动需要多学科方法，因为实时监测和地球物理调查必须与岩石学和喷发历史相结合，以了解火山系统在地质时间尺度上的时间演变。数值建模用于将不同的观测时间尺度联系起来，数据集的反演与基于物理学的统计方法是推进我们对控制火山爆发复发率和规模的过程的理解的有希望的方法。