Abstract

The accurate monitoring of CO₂ migration in subsurface reservoirs is essential for the long-term security of geological sequestration projects. Inverse modeling of seismic data using optimization algorithms offers a powerful tool to resolve reservoir properties and track plume evolution. In this study, we apply and compare two widely used global optimization methods—Genetic Algorithm (GA) and Particle Swarm Optimization (PSO)—for seismic inversion in the context of CO₂ monitoring at the Sleipner field, a pioneering saline aquifer CO₂ storage site in the North Sea. Both GA and PSO are used to optimize acoustic impedance—sensitive parameter used to track CO₂ saturation changes. GA, inspired by the principles of evolution, performs robust global searches and is effective in dealing with complex, nonlinear problems. PSO, driven by swarm intelligence, shows rapid convergence with fewer iterations but can be prone to premature convergence if not properly tuned. Application to time-lapse seismic data from Sleipner indicates that PSO offers slightly improved vertical resolution of thin CO₂ layers, particularly under noisecontaminated conditions, excels in computational efficiency, enabling quicker analysis suitable for near real-time monitoring. On the other hand, GA provide similar information with lesser resolution and long convergence time. Importantly, no evidence of CO₂ leakage has been observed in the Sleipner dataset up to the year 2006, reinforcing confidence in seismic monitoring and storage integrity. These findings suggest that both GA and PSO provide complementary strengths for CO₂ monitoring, and their hybrid implementation may further enhance inversion performance and storage assessment.