Demultiple techniques with improved AVO compliance

Seismic data must be optimally processed for accurate reservoir characterization, and therefore each step of the processing sequence should be carefully designed and assessed for AVA compliance and primary preservation. In this paper, we focus on the demultiple, a critical step as it can be difficult to identify and separate primaries and multiples, especially in shallow water settings and at the low frequencies typical of broadband data. We first propose a new AVA quality control via a pre-defined pre-stack classification method. In a second step, we show three examples of directly incorporating the AVA preservation into the cost functions of existing algorithms, such as the Radon transform and a new adaptive subtraction scheme.