next up previous [pdf]

Next: acknowledgments Up: Berryman: MESA Previous: Examples

Discussion

As with many procedures in seismic data enhancement and analysis, computation of power spectral estimates with MESA can be more art than science. Considerable insight into the stochastic processes involved and experience with choices of operator length is required before MESA can be considered a standard processing tool. The work summarized here was intended to eliminate some of the uncertainty in applying MESA by producing a reasonable upper bound [Equation (49)] on the operator length. In many cases, this upper bound will itself be a good choice for the operator length, since it often gives values comparable to those of other methods without requiring performance parameter monitoring.

In conclusion, I recommend that anyone wishing to obtain high resolution power spectra for seismic traces should examine both maximum entropy spectra and ordinary spectra. When both methods give peaks approximately at the same frequencies, we may be confident that the maximum entropy method is giving higher resolution of true modes in the spectrum. If the MESA peaks are not reasonably close to the peaks obtained by the better understood Fourier transform method, we should give additional and careful consideration to the proper choice of the operator length for the MESA spectrum.


next up previous [pdf]

Next: acknowledgments Up: Berryman: MESA Previous: Examples

2009-04-13