Monitoring and
reconstruction of subsurface CO2 plumes using a stochastic inversion approach
Ramirez, Abelardo
L.1, William Foxall1, Kathy Dyer1, S. Julio
Friedmann1 (1)
We have developed and tested a stochastic
inversion tool that uses multiple data types to reconstruct subsurface liquid
plumes (e.g., CO2, steam, water floods). The tool uses Bayesian
inference, a probabilistic approach that combines observed data, geophysical
forward models, and prior knowledge. It produces plume images that are
consistent with disparate data types, e.g., measurements of injected plume CO2
volume, surface tilt measurements, and cross-borehole electrical
resistivity
measurements. It uses a Markov Chain Monte
Carlo (MCMC) technique to sample the space of possible plume models, including
the shape, location and CO2 content of the plume. We present joint
reconstructions of injected CO2 volume and cross-borehole
electrical
resistivity
data collected during a real CO2
flood in the
electrical
resistivity
data. The
results demonstrate the benefits of joint reconstructions using disparate data.
They also demonstrate that our approach identifies alternative models when the
available data is insufficient to definitively identify a single optimal model,
and also provides the probability that a given model is the best explanation
for the available data. This information can guide further data collection or
integration. Furthermore, the method provides quantitative measures of the
solution uncertainty due to unknown reservoir properties, measurement error, or
poor sensitivity to the plume by the geophysical techniques. This work was
funded by the Laboratory Directed Research and Development Program at Lawrence
Livermore National Laboratory. This work was performed under the auspices of
the U.S. Department of Energy by the Lawrence Livermore National Laboratory
under contract W-7405-ENG-48.
AAPG Search and Discover Article #90063©2007 AAPG Annual Convention, Long Beach, California