Forecasting eruptions is the main challenge for volcanology, as ultimately it can reduce the impact of volcanic activity on the population and environment. Volcanology has experienced some success in forecasting eruptions on short timescales of hours to days beforehand. Nevertheless, these timescales may be too short for appropriate mitigation measures, including the evacuation of the population, especially in densely populated areas. Ideally, volcanologists would issue reliable medium-term forecasting, on a timescale of weeks to months, in order to give authorities the opportunity to implement appropriate prevention plans. However, medium-term forecasting has not been feasible so far, mainly due to the difficulty in finding reliable eruptive precursors. 

Volcanoes usually provide warning signs of eruption, called unrest, consisting of an anomalous change in monitoring parameters, such as surface deformation, seismicity and degassing. As unrest can also cease without eruption, unrest is a necessary but not sufficient condition for eruption, and therefore understanding the outcome of unrest (eruptive or not) remains a challenge for volcanology.

A research team from Italian, British and American institutions addresses these limitations in a scientific article just published in Nature Geoscience (https://www.nature.com/articles/s41561-022-00960-z), showing that a novel parameter, which can be indirectly measured at volcanoes, is a good indicator of whether it will erupt or not (Fig. 1). This precursor is the rate of magma injection within the volcano, derived from measurements of the surface deformation of the volcano from satellites. The rate of magma injection is directly related to the possibility to start propagating magma outside the magma reservoir, a necessary condition for having an eruption. The approach has been tested at some highly-active and frequently erupting type of volcanoes, those with calderas, which are wide volcanic depressions at the surface previously formed by the partial emptying of the magma chamber below. In particular, the precursor has been tested at calderas that erupt a very common type of magma, known as “basalt”.

Results show that it is possible to successfully forecast whether magma can propagate outside the magma chamber (and trigger eruption) or not within 1 year from the onset of the deformation in basaltic calderas for 86% of the cases, which is far higher than hit rates currently obtained through any other parameter. 

In particular, if the rate of magma injection below the caldera is high (≥ 10⁻¹ km³ yr⁻¹), eruption occurs within 1 year from the onset of the deformation in all the considered cases (nucleating unrest, Fig. 2); conversely, if the rate of magma injection is low (< 10⁻² km³ yr⁻¹), eruption does not occur for 89% of cases (Plain unrest; Fig. 2). Transitional behaviours are observed in between. For high transitional magma injection rates of  5–9 × 10⁻² km³ yr⁻¹ eruption occurs within 1 year from the onset of the deformation in the 86% of the cases, while for low transitional magma injection rates of 1–4.99 × 10⁻² km³ yr⁻¹ eruption does not occur in the 87.5% of the cases (Fig. 2).  

The increased possibility to swiftly access and invert space-geodetic data permits near-real-time evaluation of the magma inflow rate in a shallow reservoir below a basaltic caldera. This allows prompt forecasting of the probability of having an eruption, weeks to months ahead, a crucial support when densely inhabited areas may be evacuated.