Distinguishing Inflation Drivers at Shallow Magmatic Systems Using Statistical Data Assimilation
Abstract
Recently, the application of statistical data assimilation to volcanic deformation data has produced successful forecasts and hind-casts of volcanic eruptions. Although these results highlight the potential of such techniques to revolutionize forecasting approaches, much work remains to be done in assessing their full capabilities and limits. In particular, non-uniqueness in how source parameters are reflected in surface observations can significantly impair the assimilation's ability to resolve the magma system's true state and the likelihood of eruption. Parameters such as radius and overpressure, for instance, have critical implications for the mechanical stability of the reservoir as it evolves. For deep spherical reservoirs, however, variations in these two parameters produce identical deformation patterns at the surface. In such end-member cases, it therefore becomes challenging to determine model parameter values using geodetic data alone. However, the degree to which this effect holds for shallow reservoirs of varying shape has not yet been delimited. In this study, we investigate how well the Ensemble Kalman Filter (EnKF) data assimilation framework can distinguish between changes in reservoir size and pressure at a shallow system. Two time series of synthetic GPS and InSAR data are generated, one in which inflation is driven by excess pressure and another in which it is driven by the lateral expansion of the reservoir. We find that although the EnKF does not accurately reproduce the original synthetic model, it can still distinguish between the two datasets by recording relatively larger variations in the correct driving parameters. Moreover, the misfit with the synthetic values can be reduced by improving data availability and by introducing slight changes in how the EnKF scales model parameters or handles catastrophic divergence. Ultimately, while data assimilation remains a powerful tool for assessing the accumulation of stress in the subsurface and the long-term stability of volcanic systems, future studies will need to consider the effects of non-uniqueness, either in their conclusions or by incorporating additional a priori constraints on their parameters.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.V13D0205G
- Keywords:
-
- 1295 Integrations of techniques;
- GEODESY AND GRAVITY;
- 7280 Volcano seismology;
- SEISMOLOGY;
- 8178 Tectonics and magmatism;
- TECTONOPHYSICS;
- 8485 Remote sensing of volcanoes;
- VOLCANOLOGY