A new way of monitoring volcanic activity by integrating ongoing satellite measurements into dynamic models has been demonstrated by researchers from France. Based on data assimilation, the method might one day allow for real-time eruption forecasts in volcanic regions.

As magma moves beneath the Earth's surface – such as under a volcano – the ground above flexes. These ground movements can be measured using both GPS and satellite-based radar data, and used to develop models of the depth and shapes underlying magma reservoirs.

A limitation of many of these models, however, is that they are kinematic in nature – focusing on motion alone. This makes them unable to yield information on the pressures of the underlying magma system. This is important in determining when a magma chamber will rupture and the volcano's capacity to feed any resulting eruption. The surface disruption caused by a small pressure change in a large magma chamber may look identical to a large pressure change in a small magma chamber, for example – even though the latter case is more likely to lead to an eruption.

To distinguish between such scenarios, volcanologists must use dynamic models that can consider how the surface displacements change with time. A small chamber, for example, would pressurize much faster than a large chamber. Most dynamic models tend to be based on data inversion and require extensive calculation and the incorporation of all observations beforehand. This makes them unsuitable for real-time eruption forecasting because they are unable to incorporate ongoing measurements.

In their new study, geophysicist Mary Grace Bato and colleagues at the Institut des Sciences de la Terre in Grenoble have addressed this issue by turning to data assimilation. This is a time-stepping approach that combines models, observations and error statistics to forecast the state of a dynamic system. Data assimilation has long been used to produce weather forecasts and predict the effects of greenhouse-gas emissions.

To read more, click here.