However, when picked velocities for structured area in the middle of the line, we encountered reflection events that have dramatic velocity changes. Crossing features were shown in these NMO corrected CMP gathers (Fig. \ref{650073}). Firstly we tried to flat all the reflections as best as we could though it's hard to decide what velocities to use for these reflections. Thus the seafloor velocities of this dipping area were not constrained to 1500 m/s. We used this velocity model (Fig. ) to finish our initial stacking and migration.