Improving fluid-induced time-lapse seismic monitoring using local orthogonalization

Time-lapse seismic surveys are critical for monitoring subsurface reservoir changes due to fluid movement and geological alterations. However, conventional methods often suffer from noise and misalignment between repeated surveys, reducing interpretation accuracy. We propose a local orthogonalization-based method that isolates dynamic fluid-induced changes from static geological structures by reinforcing local signal orthogonality within seismic differences. The technique leverages adaptive similarity and regularized weighting to suppress artifacts while preserving meaningful subsurface changes. Validation using synthetic and field datasets from CO 2 sequestration and steam injection projects confirms the method’s effectiveness in improving reservoir monitoring and its robustness to noise and acquisition inconsistencies.