Abstract: Interval multiple is one of noise which is quite difficult to suppress in seismic data processing due to its irregular reflection. It appears as artifact in migration which affects the seismic processing quality. In marine seismic data processing, sea bottom related interval multiples exist widely because the strong impedance difference at sea bottom leads to sea bottom related interval multiple especially under the sea bottom where high velocity layer or scattering body exists. In order to solve this problem, a method for prediction of sea bottom related interval multiple prediction is proposed, which is based on wave equation. Thanks to the clear sea bottom reflection, sea bottom related interval multiple was predicted by convolution and cross-correlation with sea bottom reflection and those wave-field reflected under sea bottom. This is a data-driven multiple prediction method which overcomes the disadvantage of velocity dependence and has higher computation efficiency. Simple model test shows its validity. Finally, based on Sigsbee2B model test, sea bottom related interval multiple is reasonably predicted, confirming the adaptability of this method to complex substructure.