MAGMA: Magma Accommodation and Ground Movement Analysis

Lead Research Organisation: University of Leeds
Department Name: School of Earth and Environment

Abstract

MAGMA will transform how we capture the complex geology beneath volcanoes within the numerical models of volcano ground movement that are used in eruption threat assessment; this will help drive significant improvements in forecasting eruptions, helping keep people safe and secure.

Over 800 million people live near volcanoes. To keep people safe and secure, we need to reliably forecast when volcanic eruptions may occur and what their potential size, style, and hazards will be. However, because the geology and plumbing system dynamics beneath each volcano is unique, all volcanoes behaves differently, making it difficult to reliably forecast eruptions.

As magma intrudes through the crust and accumulates, it often pushes up the overlying rock and Earth's surface. Ground movements at active volcanoes thus often herald eruption. We therefore monitor the surface elevation of volcanoes, using satellites and ground-based tools, to look for tell-tale ground movements related to magma build-up. Using sophisticated numerical models, we can estimate the amount and location of magma required to drive measured ground movement. These estimates of magma bodies provide crucial inputs for eruption forecasts as they constrain how close magma may be to the surface, its volume and pressure, and how fast it is moving.

A key flaw of many ground movement models is that they assume the rocks through which magma moves are simple and have no internal structure or compositional variation. Yet we know rocks vary physically and chemically at all scales, and how they deform changes in space and time. Critically, ground movement models that include more realistic geology (e.g. layering) show that, incorporating even small degrees of complexity can change estimated magma body properties by orders of magnitude. Such changes in magma body estimates may be the difference between forecasting an eruption or not.

Reliably using ground movement to forecast volcano eruption onset, size, style, and hazards requires models that realistically capture host rock complexity. For example, uplift above injecting magma requires the overlying rock to bend. Yet we actually know very little about how resistant rocks are to bending. We also do not know how local extension and compression within the bending rock volume changes its material properties and thus affects its response to further deformation. Critically, these controls on rock bending dictate how much and where ground movement occurs above injecting magma. To solve these problems in MAGMA, we will:

1) Conduct mechanical experiments where we load and bend different rocks to measure their resistance to bending; lab results will be 'upscaled' using tried and tested geotechnical methods so they are representative of entire rock masses.
2) Examine the geometry of and deformation within ancient areas of uplift above magma bodies exposed at the surface or imaged in 3D seismic reflection data, which provides ultrasound-like images of the subsurface.
3) Collect samples from field areas and use mechanical experiments to pull and compress them in the lab to test how bending locally changed their resistance to deformation.

MAGMA will use this information to build novel Finite Element numerical model that can capture multiple aspects of host rock complexity. These models will adopt a perturbation theory approach, which has been successfully tried and tested in complex ground movement models of earthquakes. With our developed method, we will create synthetic but realistic models of volcanic systems to simulate ground movement above pre-defined magma bodies. By incorporating host complexities and varying inputs, we will test different geological scenarios. Our work will enable a step-change in ground movement modelling, which will lead to improvements in the reliability of eruption forecasting and understanding of volcanic processes, helping keep people safe and secure.

Publications

10 25 50