Sea surface interaction, if not properly handled during processing, penalizes the resolution of the seismic data. The delayed signals coming from the reflection at the sea surface will interfere destructively or constructively with the subsurface reflections. Conventional data processing techniques that handle sea surface ghosts are generally limited to the flat sea surface assumption (Fokkema and van den Berg (1993); Amundsen (2001)). In recent years, some of these techniques have been extended to account for the variable sea surface profile (Laws and Kragh (2002); Orji et al. (2012); Asgedom et al. (2016)). Neverthless, the algorithms still neglect the time dependency of the sea surface. With the advent of the dual sensor acquisition concept, which enables the separation of the acquired wavefield into its upgoing and downgoing components (Tenghamn et al. (2007)), this assumption can be relaxed on the receiver side.
The first step for accurate removal of sea surface ghosts starts with a thorough investigation of rough weather ghost effects by forward modelling. Cecconello et al. (2016) present an integral approach to model the receiver side ghost effects from time-varying rough sea surfaces. In this paper, we follow the same method to model source side ghost effects. Finally, we present an integral inversion method for source deghosting that handles time-varying sea surfaces.