CROSS-EQUALIZATION

Since the two experiments were conducted with very similar acquisition parameters, none of the `optimal' cross-equalizing methods that I discussed in my earlier paper were useful. The simple time domain match-filtering approach was used instead.

In this approach, a filter, ${\bf A}$, is designed on a training window to map one survey, ${\bf s_1}$, onto the other, ${\bf s_2}$, minimizing the regression:

$${\bf A s_1} - {\bf s_2} \approx {\bf 0}$$ (1)

The training window should be a subset of the survey in which it is expected that all the changes are acquisition related. For this dataset the training window that was used are shown in Figures 4 and 5.

 

htrain.300Hz Figure 4 Training window for 300 Hz section. Left is without honey and right is with honey.

 

htrain.700Hz Figure 5 Training window for 700 Hz section. Left is without honey and right is with honey.