Many studies have been published concerning classification techniques of seabed surfaces using single beam, multibeam, and side scan sonars, while few paid attentions to classify sub-bottom layers using a non-linear Sub-Bottom Profiler (SBP). Non-linear SBP is known for its high resolution images due to the very short pulse length and aperture angle for high and low frequencies. This research is devoted to develop an energy based model that automatically characterizes the layered sediment types as a contribution step toward ‘‘what lies where in 3D?”. Since the grain size is a function of the reflection coefficient, the main task is to compute the reflection coefficients where high impedance contrast is observed. The developed model extends the energy based surface model (Van Walree et al., 2006) to account for returns reflection of sub-layers where the reflection coefficients are computed sequentially after estimating the geo-acoustic parameters of the previous layer. The validation of the results depended on the model stability. However, physical core samples are still in favor to confirm the results. The model showed consistent stable results that agreed with the core samples knowledge of the studied area. The research concluded that the extended model approximates the reflection coefficient values and will be very promising if volume scatters and multiple reflections are included.