Noisy data and land data

  In previous chapters, all of the interpolation examples were run on marine data. Interpolating marine data is a good application because of the trouble with aliased multiples. Marine data is also relatively easy to interpolate, because it typically has little noise and regular geometry. Land data is much more difficult. Land data often has discontinuous arrangements of shots and receivers, where the survey is forced to work around and over surface features. Land data also tends to be noisier. Noise and statics make it difficult to predict a seismic trace from its neighbors, so it is more difficult to interpolate. Nonetheless, because of the expense and effort of acquiring land data, it is worthwhile to try interpolating it.

Statics and irregular geophone placement can significantly reduce the window size that can be realistically assumed to be stationary. That makes it attractive to use tiny micropatches rather than patches large enough to calculate a PEF independently. However, it also suggests that PEFs should not be gradually varying, because the dips of events might change very quickly where there are statics.

In this chapter, I apply the method of the previous chapter to interpolate noisy marine data and land data with noise and irregular geometry. The problems are more difficult than earlier examples, but the results are still encouraging.