In order to separate and/or attenuate noise contaminating seismic data, it is necessary to identify and use some characteristic of the noise which distinguishes it from "legitimate" seismic reflection energy. Sometimes, effective use of this characteristic requires the transformation of the input data samples to a new coordinate system via one or more transform or mapping operations. One unique characteristic available to analyze/separate certain noise modes on multi-component seismic data is that of particle motion polarization, since modes exhibiting characteristic polarization can be simultaneously and independently detected on sensors recording motion in two or more orthogonal directions.
One seismic noise mode well known for its distinctive polarization is the Rayleigh Wave or "ground roll", so it constitutes a good target for a noise attenuation algorithm based on some aspect of polarization. Another prominent example of strongly polarized noise is the ice flexural wave often encountered on seismic data recorded on floating ice in the arctic; it is also a prime candidate for some form of polarization filtering. A simple common-mode noise estimation technique was devised and applied to a multi-component field example of each of these two types of coherent noise. Based on our relatively simple tests, however, it appears that separation of coherent noise based on its polarization on two independent particle motion component wavefields is no more effective than estimation of the noise in each wavefield separately, based solely on the coherence characteristics of the noise.
View full article as PDF (3.11 Mb)