Incorrect imaging of internal multiples can lead to substantial imaging artifacts. It is estimated that most seismic images available to exploration and production companies have had no direct attempt at internal multiple removal. Previously, we have considered the role of sparsity promoting transforms for improving practical prediction quality for algorithms derived from the inverse scattering series (ISS). Furthermore, we have developed a demigration-migration approach to perform multidimensional internal multiple prediction with migrated data and provided a synthetic proof of concept. Now, we consider application of the demigration-migration approach to field data from the Norwegian Sea and compare it to a poststack method (from a previous related work). Beyond application to a wider range of data with our approach, we consider algorithmic and implementational optimizations of the ISS prediction algorithms to further improve the applicability of the multidimensional formulations.