Horizon refinement by synthesis of seismic and well log data

Subsurface rock strata are often conceptualized in three dimensions as depth surfaces, or horizons. By picking the corresponding reflection event from 3-D seismic data, an approximate “seismic horizon” can be obtained, but in general the inconsistencies in the approximation are nontrivial. Picks from well log data give the depth to the horizon to high accuracy, but only at a few points. To achieve the final goal, which is a representation of the actual horizon more accurate than the seismic horizon, I present a least squares optimization scheme to optimally “warp” the seismic data to match the well log measurements. I then test the method on a small group of data from offshore West Africa, consisting of two picked seismic horizons data from about 15 well logs. The method shows promise in more far-reaching problems such as anisotropic parameter estimation and global velocity update.