A machine learning approach for calibrating seismic interval velocity in 3D velocity model

Velocity model technique is routinely used to convert data from the time-to-depth domain support prospect evaluation, reservoir modelling, well engineering, and further drilling operation. In Vietnam, conventional velocity building workflow oversimplifies interval velocities as only are populated into 2D grids for depth conversion or oversimplified calibration by applying a single scaling factor function. This study explores 3D obtain accurate high-resolution using machine learning approach both fields A B in Cuu Long basin, offshore Vietnam.
 To design an effective conversion, anisotropy analysis was performed understand differences between seismic geological layer structural model. The multiplied achieve velocity. velocity, elastic attributes, geometric stratigraphic attributes were training features (variables) predicting supervised algorithm Supervised offers opportunity develop expert-knowledge-based automated system, which incorporates knowledge quantitative mining [1]. random forest regression algorithms selected after evaluating several algorithms. provide insight uncertainty of final conducted blind test 24 wells 7 horizons.
 comprehensive built first time Vietnam. result showed can address disadvantages create highly representations subsurface including measure uncertainty.