Acoustic waves are modeled using a 2D frequency-domain finite-difference scheme with three different artificial damping boundaries: the absorbing boundary condition (ABC), perfectly matched layer (PML), and a hybrid strategy combining PML and ABC. We evaluated the features of each simulation for a given artificial boundary condition for homogeneous and heterogeneous media. For shorter PML layers, the hybrid absorbing scheme gave better results when compared with the traditional PML technique, with a significant reduction of the spurious reflections and a shorter computational run time.

Neural networks have been used in various computer vision applications, including noise removal. However, removing seismic noise via deep learning approaches faces a specific issue: the scarcity of labeled data. To address this difficulty, this work introduces an adaptation of the Noise2Self algorithm featuring a one-shot learning approach tailored for the seismic context. Essentially, the method leverages a single noisy image for training, utilizing a context-centered masking system and convolutional neural network (CNN) architectures, thus eliminating the dependence on previously labeled data. In tests with Gaussian noise, the method was competitive with established approaches such as Noise2Noise. Under real noise conditions, it demonstrated effective noise suppression removal for a smaller architecture. Therefore, our proposed method is a robust alternative for noise removal that is especially valuable in scenarios lacking sufficient data and labels. With a new approach to processing seismic images, particularly in terms of denoising, our method contributes to the ongoing evolution and enhancement of techniques in this field.

Deep Learning-based models such as Convolutional Neural Networks, have led to significant advancements in several areas of computing applications. Seismogram quality assurance is a relevant Geophysics task, since in the early stages of seismic processing, we are required to identify and fix noisy sail lines. In this work, we introduce a real-world seismogram quality classification dataset based on 6,613 examples, manually labeled by human experts as good, bad or ugly, according to their noise intensity. This dataset is used to train a CNN classifier for seismic shot-gathers quality prediction. In our empirical evaluation, we observe an F1-score of 93.56% in the test set.

Although modern supercomputers are composed of multicore machines, one can find scientists that still execute their legacy applications which were developed to monocore cluster where memory hierarchy is dedicated to a sole core. The main objective of this paper is to propose and evaluate an algorithm that identify an efficient blocksize to be applied on MPI stencil computations on multicore machines. Under the light of an extensive experimental analysis, this work shows the benefits of identifying blocksizes that will dividing data on the various cores and suggest a methodology that explore the memory hierarchy available in modern machines.