# Kinematical filters

Seismic wave filed is a superposition of waves and noises, and noises can be both coherent and non-coherent. Algorithm of 2D filtering is one of the most efficient approaches for both coherent and non-coherent noises. There are three basic methods of filtering implemented in Prime: in time-spatial domain (x - t), in spatial-frequency domain (x - w) and in 2D Fourier domain (k - w). Local tau-pi transformation is used for kinematical filtering in t-x domain.

## Kinematical filters based on the signal pass

Suppose, that travel time curves of useful seismic waves were approximately obtained after preliminary analysis. Filter must pass these waves and also those waves which rely to the specified fan without any distortions. All other waves are considered to be noises and they must be removed.

In cases of noisy data or travel time curves non-hyperbolicity it could be difficult to estimate the signal and noise kinematics. Adaptive filtering is very useful in this case. It allows analyzing wave field for every point and finding the signal and noise.

More or less exact model of noise waves (for example multiples) may be known when we attenuate coherent noises. In this situation adaptive filter-mask can be applied. It can kinematically analyze the model of noise and make decision about presence or absence of noise for every data sample. The main advantage of using this method than using those methods, which directly subtract model of noise from initial field is no necessity of full correspondence of modeled dynamical features to the real dynamical features.

## Filtering in offset-domain, stacked and migrated sections

Described filtering approaches can be applied not only to the CDP gathers, but also to the offset-sections, time sections or results of migration in t-x domain.

Polarization filtering - is a separation of two or three-component CSP gathers with extraction of P-waves and S-waves. A priori velocity in upper part must be known. Kinematical filters can be used for noise extraction before elimination.