Abstract

Shallow water demultiple remains a significant challenge due to the absence of recorded waterbottom reflections in seismic data acquired in shallow marine environments. This issue primarily arises from the lack of near-offset channels, which are crucial for capturing the water-bottom response. Techniques such as Surface-Related Multiple Attenuation (SRMA) and conventional water-bottom multiple suppression heavily rely on accurate waterbottom information. In its absence, the estimation of the multiple model becomes unreliable and ineffective. Therefore, a robust approach is essential to accurately map the water-bottom depth or time. Various techniques are available to address this issue, including manual picking of the water-bottom on stacked data, manual picking on the autocorrelation of the near-offset stack, and reconstruction of the water-bottom directly from the seismic data. In this study, we conduct a comparative analysis of these methods, demonstrating the superior robustness of the reconstruction-based approach over manual picking techniques. Introduction One of the primary challenges in imaging shallow water marine seismic data is the effective attenuation of multiples generated by the waterbottom and the free surface. Surface-Related Multiple Elimination (SRME) is considered one of the most effective methods for multiple suppression, as it does not require prior knowledge of subsurface geology or a velocity model. However, successful implementation of SRME critically depends on accurate acquisition geometry and reliable water-bottom information. In shallow marine environments—where the water depth is less than the source–receiver spacing—near-offset water-bottom reflections are often not recorded. This limitation significantly hinders the application of SRME in such settings. To address this issue, several methods have been explored to derive accurate water-bottom information efficiently and with minimal manual effort.