Groundroll prediction by interferometry and separation by curvelet-domain matched filtering

The removal of groundroll in land based seismic data is a critical step for seismic imaging. In this paper, we introduce a work flow to predict the groundroll by interferometry and then separate the groundroll in the curvelet domain. Thus workflow is similar to the workflow of surface-related multiple elimination (SRME). By exploiting the adaptability and sparsity of curvelets, we are able to significantly improve the separation of groundroll in comparison to results yielded by frequency-domain adaptive subtraction methods. We provide synthetic data example to illustrate our claim.