An adaptive seismic random noise attenuation method based on Engl criterion using curvelet transform

Random noise attenuation method based on sparse inversion require accurately estimating a threshold matching the noise energy to ensure reliable denoising. However, noise properties vary across datasets, necessitating manual tuning which can be computationally expensive and labor-intensive. To address this issue, this work proposes an adaptive seismic random noise attenuation technique based on Engl criterion using Curvelet sparse transform. By leveraging the inherent trade-off between solution sparsity and fitting error during the iteration, the method automatically determines an optimal threshold for attenuation without requiring prior noise estimation. Effectiveness is demonstrated through synthetic and field data experiments, with results showing comparable performance over state -of -art approaches. The adaptive nature eliminates tedious manual tuning while the Curvelet-based sparse formulation handles nonlinearity of seismic events. This noise -agnostic automated approach enables efficient high -quality denoising across diverse datasets with minimal user intervention. The simplicity, efficiency, and reliability signify its viability for practical workflows demanding customizable random noise suppression.