Joint sparsity recovery for noise attenuation
Yue Tian, Lei Wei, Chang Li, Shauna Oppert, Gilles Hennenfent · 2018
We present a distributed compressive sensing technique called Joint Sparsity Recovery (JSR), which is adapted for post-stack data with a focus on attenuating coherent and incoherent noise. The method applies sparsity-promoting joint inversion on multiple datasets that are represented on the same regular grid. It quantifies the common and unique parts between all input datasets, and simultaneously removes the data misfit as unwanted noise. We present multiple field examples that demonstrate the effectiveness and flexibility of JSR in attenuating coherent and incoherent noise in different types of geophysical data. JSR is a powerful tool to attenuate 4D noise in time-lapse volumes of seismic stacks, inverted 4D elastic properties, etc. It is especially useful for improving 4D interpretability when the 4D noise is strong due to poor repeatability. JSR can also be used as part of a workflow to attenuate noise that is not common to the multiple inputs, especially the coherent noise that is hard to deal with otherwise. Presentation Date: Monday, October 15, 2018 Start Time: 1:50:00 PM Location: Poster Station 19 Presentation Type: Poster