Jinsong Chen's Geophysical Inverse Problems Research Interest
Geophysical inverse problems are the inferences of physical properties from geophysical measurements. They are typically ill-posed because of non-uniqueness of the solutions and nonlinearity of the forward modeling. The main tasks of my research in the area: (1) To formulate inverse problems properly within a Bayesian framework, (2) To develop efficient (often problem-specific) Markov chain Monte Carlo (MCMC) sampling strategies to explore the joint posterior probability distributions, and (3) To implement the MCMC sampling strategies by developing fast and efficient computer codes on multiple platforms (e.g., Unix and Linux clusters). Examples of my research are:
- Developed Bayesian models to estimate reservoir parameters using seismic and electromagnetic data and stochastic rock physics models (Chen and Dickens, 2009, Geophysical Prospecting; Chen and Hoversten, 2005, SEG Expanded Abstract).
- Compared Gauss-Newton iterative and Markov chain Monte Carlo based methods for inverting spectral induced polarization data (Chen et al., 2008, Geophysics).
- Developed Bayesian models to estimate reservoir gas saturation from 1D marine seismic amplitude versus angle (AVA) and controlled-source electromagnetic (CSEM) data using MCMC sampling methods (Chen et al. 2007, Geophysics; Chen et al., 2004, SEG Expanded Abstract).
- Developed stochastic models to estimate reservoir porosity and water saturation using borehole porosity and water saturation measurements, crosswell seismic P- and S-wave traveltimes, and inverted 2D electrical conductivity data (Chen and Hoversten, 2003, SEG Expanded Abstract).