Prediction of Elastic Properties Within CO₂ Plume at Sleipner Field Using AVS Inversion Modified for Thin‐Layer Reflections Guided by Uncertainty Estimation

Abstract

Existing amplitude variation with offset (AVO) and slowness (AVS) theories fail to interpret observed amplitudes in terms of actual elastic properties of the media for a reflection event resulting from a stack of thin layers. We propose a method to replace the stack of thin layers with an equivalent medium of elastic properties using the Backus averaging theory. Numerical examples show considerable deviation in intercept calculated from the modified AVO and AVS theories from the conventional methods. AVS theory is preferred as it can be applied irrespective of impedance contrast in both precritical and postcritical reflections, and no spherical divergence correction is required. P and S wave velocities, density, and thickness of thin layers are determined using the inversion scheme of very fast simulated annealing (VFSA). Uncertainty in prediction is evaluated using an approximate marginal posterior probability density function and a parameter correlation matrix. We demonstrate our methodology on synthetic and real seismic data from the Sleipner field, which is a good example of enhanced seismic amplitudes due to interference among reflections from thin layers. We select 11 common depth point (CDP) gathers for amplitude analysis of four identified reflectors along a line of 3-D seismic data acquired in 2008. Predicted mean models of CO 2-saturated sand layers at all locations show that the thickness, density, and P and S wave velocities vary from 5-7 m, 1.8-2.0 g/cm 3 , and 1,460-1,490 and 630-650 m/s, respectively. Our analysis suggests that the model parameters are well constrained and independent except at a few locations.

Estimation of porosity and gas hydrate saturation by inverting 2D seismic data using very fast simulated Annealing in the Krishna Godavari offshore basin, India

“…VFSA has been applied by many researchers to solve the parameter estimation problems (e.g. Zhao et al ., 1996; Srivastava & Sen, 2009; Ghosh & Ojha, 2020). Simulated Annealing (SA), Artificial Neural Networks, Genetic algorithm and Particle Swarm Optimization are the widely used global optimization techniques for the inversion of various geophysical data (Rothman, 1985, 1986; Juan et al ., 2010; Sen & Stoffa, 2013; Biswas & Sharma, 2017).…”

Variational Experimental Design Methods for Geophysical Applications

“…We focus on questions relating to the CO 2 saturation in a subsurface layer. The setup of the physical parameters represents a simplified model related to the Sleipner field (Ghosh & Ojha, 2020;Dupuy et al, 2017). The upper layer is described by its seismic properties (P-wave velocity α 1 , S-wave velocity β 1 ), density ρ 1 , and depth d, all with ranges given in Table 1.…”