Evolutionary data assimilation at Long Valley Caldera, CA

"Despite advancements in volcanic modeling, the time-dependent evolution of volcanoes is still poorly understood. Of particular need are methods for combining extensive monitoring data sets with dynamic models. Sequential data assimilation has been shown to be powerful approach for linking models and data to improve the use of both. One such approach, Evolutionary Data Assimilation (EDA), previously used in hydrological predictions [Dumedah, 2012], is adapted. EDA provides a ""snaphot"" of parameters such as location and volume change at each timestep, allowing users to update a dynamic model’s trajectory as new observations become available. To test the application of EDA to volcano monitoring, we first develop a series of synthetic numerical experiments to track the ability of the EDA to back out chosen model parameters. Specifically, synthetic GPS and interferometric synthetic aperture radar (InSAR, a satellite based measurement of deformation) data are created from an analytical model with prescribed values for geometry and volume change. We find that EDA performs well in synthetic tests using GPS and InSAR data. After establishing the EDA method with synthetic tests, the EDA is applied to investigate the recent unrest observed at Long Valley Caldera in California using GPS from 1995-2015 and InSAR data from 2012-2014. EDA performed reasonably well at finding the location of the chamber and estimating volume changes. However, due to the analytical Mogi model used and the inherent nonuniqueness of parameters such as depth vs pressure vs radius, EDA was not able to resolve the depth or radius of the chamber. With more robust models, EDA is a powerful method that could be used to track evolution of volcanoes."

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Geology

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2017

Author dc:creator

• Monical, Therese

Contributors dc:contributor

• Gregg, Patricia

• Data assimilation
• Long Valley

Statement dc:rights

• Copyright 2017 Therese Monical

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/97438

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/97438