Data editing to remove the high-amplitude noise is the first step in more refined signal and noise separation processes, as well as for other data processing. This step will often be required, since high-amplitude noise corrupts the least-squares calculations used in the more sensitive processes presented in later chapters. The technique shown here appears robust enough as a standard processing tool.
While removing high-amplitude noise is necessary for the techniques discussed in this thesis, there will be advantages in removing high-amplitude noise before other processes, even if the samples removed are not accounted for in the processes that follow. While zeroed samples may be considered as noise, this noise may be preferable to the higher-amplitude noise that sometimes occur in real data. One example of where this noise removal is desirable, would be standard single trace deconvolution in the presence of noise spikes. If these spikes have amplitudes that are high enough, the deconvolution operators will be ineffective, since the spikes give the data a white spectrum. Another example would be velocity analysis, where a high-amplitude spike in the input may generate curved noise trains in the velocity analysis.
While the method presented in this chapter appears useful enough, it is not the only approach that could be taken to remove high-amplitude noise. In particular, high-amplitude coherent noise may be simply muted out of a data record. The final editing method will depend on the nature of the noise and on the data. For small data volumes, manual muting by the processor may be the most effective. The important issue is that high-amplitude noise is removed and that the edited data be marked so it will not be treated as good data and can be restored later. In chapter , an inversion to restore the edited data while separating noise and signal will be demonstrated.