For seismic stratigraphy, a key issue is distinguishing the characteristics of seismic reflections generated by geologic events with different scales, which in turn assists in sequence stratigraphic interpretation. Data-driven signal decomposition approaches, such as variational-mode decomposition (VMD) and multichannel VMD (MVMD), can use waveform similarity to decompose seismic data into several intrinsic mode functions (IMFs). Unfortunately, it is a hard task to define the number of IMFs. To overcome the shortcoming of previous works, we have constructed an enhanced MVMD and then developed a workflow to decompose seismic data. We first determined the relationship between the IMFs and structures with different scales. Then, we developed a method to set the number of IMFs using contraction operator mapping and scale-space representation. Finally, we developed a workflow and applied it to synthetic and field data to identify seismic sequence stratigraphy boundaries. Synthetic and field data examples indicate that our workflow preserves lateral continuity and precisely extracts IMFs caused by depositional sequences with different geologic scales, facilitating the interpretation of subtle depositional patterns.