Wave sepraration using robust mathematics morphology based on signal lateral continuity

Seismic data are always contaminated with noise. There are generally two categories of seismic noisy: regular undesired signal and random noise. Some regular noise, such as multiples, may be useful for other seismic data processing. Wave separating method helps researchers to extract this information. Bandpass frequency filtering, Time-variant spectral whitening, F-K or tau-p filtering are widely used to attenuate regular noise, but with infidelity and not amplitude preservation. A robust mathematics morphological filter is presented in this paper with no need to transform other domain and scarcely damage useful seismic events. Synthesized data example demonstrates the processing procedure of wave separating filter, and show the feasibility of this method. The validity has been proved by the actual data processing. The proposed method is effective not only to regular noise with homologous seismic, but also outsource data. Modified morphological filter allows the picking up time windows not precise, because of suitable structural element, which also reduce the difficulty of processing.