Decon in the log domain - practical considerations

Additional considerations

Another consequence of the source and receiver ghost signals is the presence of notch frequencies. The particular frequency value in which a notch will occur depends on the time difference between the positive and negative lobes of the Ricker, and therefore on the depths of the source and receiver, and their offset:

$$f_n = \frac{V_w}{2d\text{cos}(\theta)},$$ (4)

where $V_w$ is the water velocity, $d$ is the depth and $\theta$ is the reflection angle. Since the sources and receivers are usually towed at a depth of only a few meters, the notch frequency is quite high (above 100Hz), and increases with offset. In order to see it in the spectrum of the data we would usually require a time sampling of 2ms.

The purpose of deconvolution is to separate out the parts of the data that are a result of acquisition from the parts that are a result of geology. Therefore, if the notch is visible in the spectrum of the data, we expect that after a successful deconvolution the notch will have been filled in.

In this paper, we show the results of applying this deconvolution method to various offsets of a marine streamer dataset, with and without regularizations. The code used is the same as in Claerbout et al. (2012), and was written by Antoine Guitton.

Decon in the log domain - practical considerations