Full text loading...
We proposed a method combining coloured inversion, constrained sparse-spike inversion, and the concept of harmonic extrapolation. We redefined the theoretical and practical limits of seismic resolution. ML methods for relating seismic information with well data and for interpolation between wells add extra regression information pushing the envelope for resolution of seismic inversion. However, ML methods always have generalization gap that is very challenging to quantify. In other words, “super-resolution” of well interpolation very much dependent on geostatistical properties of well information and possesses substantial uncertainty that is important to quantify. The application of our method to the Volve data demonstrated successful blind well test.