The present invention relates to a method of processing seismic signals comprising: receiving a set of seismic signals, applying a wavelet transformation to the set of signals and generating transformed signals across a plurality of scales. Then for each scale determining coherence information indicative of the transformed signals and generating a comparison matrix comparing the transformed signals, then outputting seismic attribute information based on combined coherence information.

Systems and methods for noise-robust time-domain multi-scale full waveform inversion using convolved data are disclosed. The methods include obtaining, using a seismic acquisition system, an observed seismic dataset pertaining to a subsurface region of interest; obtaining, using a seismic processor, a seismic velocity model; and iteratively, using the seismic processor, until a stopping criterion is met: selecting a wavelet with a frequency parameter, wherein the frequency parameter increases with each iteration, forming a convolved seismic dataset based on a convolution of the wavelet with the observed seismic dataset, and updating, using a full waveform inversion, the seismic velocity model based, at least in part, on the convolved seismic dataset. The methods further include forming a seismic image of the subsurface region of interest using the updated seismic velocity model.

Techniques to reduce noise in seismic data by receiving a set of seismic data comprising a plurality of input volumes each inclusive of positional data and at least one additional attribute related to the seismic data, selecting a first input volume of the plurality of input volumes having a first additional attribute related to the seismic data, and generating a pilot volume by selecting a range of input volumes of the plurality of input volumes and stacking input volumes of the range of input volumes with the first input volume. Additionally, generating a trained dictionary based upon transformation of the pilot volume, transforming the first input volume into transformed data, imposing a sparse condition on the transformed data utilizing the trained dictionary to generate sparsified data, and inverse transforming the sparsified data to generate an output data volume as a portion of a set of modified seismic data.

A method for determining earth property data from field vertical seismic profiling (VSP) data. The method includes obtaining a survey dataset regarding a geological region of interest encompassing a set of drilled wells. The survey dataset includes VSP data and well data corresponding to the drilled wells. The method also includes: extracting, from the VSP data, a first wavelet; constructing a set of pseudo-wells; determining a reflectivity series for each pseudo-well based on the well data; and generating a first synthetic seismic dataset for each pseudo-well based on its reflectivity series and the first wavelet. The method further includes training a set of machine learning models to predict earth property data given a VSP dataset using the first synthetic seismic dataset and target data. The method further includes determining, with the set of machine learning models, predicted earth property data from a field VSP dataset and planning a wellbore path.