Seismic frequency enhancement through spectral shaping.

摘要

Seismic vertical resolution is a function of subsurface velocities and frequency content of the data. We present spectral shaping (SPS), a new technique that enhances bandwidth, and thus stratigraphic resolution, on seismic data. The fundamental problem of extending the seismic frequency band can be simplified to one basic idea: improving the bandwidth of the frequency spectrum of a seismic trace while preserving its phase spectrum. The phase spectrum controls time localization of seismic events while the amplitude spectrum controls how energy is partitioned among these events. By comparing the amplitude spectrum of a low frequency (target) trace with that of a high frequency (reference) trace, we can extract an enhancement factor for each frequency component. This factor can then be multiplied by the real and imaginary parts of the target trace to modify its amplitude spectrum while preserving its phase. The result is a seismic trace that has the frequency spectrum of the reference trace while maintaining its own original phase spectrum. The main application presented here is compensating for loss of frequencies due to NMO stretch.; Although the results are very promising, a side effect of SPS is the introduction of artifact noise caused by boosting weak frequencies. Several solutions are proposed. Damped SPS invokes a numerical stability factor "epsilon" to avoid dividing by hard zeros. Using epsilon results in only partially shaping the amplitude spectrum at each frequency component. Another solution is bandpass SPS (partial spectral shaping or PSPS) where the user can define the range of frequencies to be used from the reference trace. Either solution, or a combination of both, proved very effective in minimizing the artifact noise while maintaining the benefits of the process.; Spectral shaping can also be used as a generic tool with many applications. For example, we can use the shallow, high frequency part of seismic data as a reference to compensate for loss of frequency with depth due to attenuation. This application is presented and tested on real data with promising results.