Conclusion

We presented an IRLS flattening approach able to flatten data cubes in the presence of laterally limited vertical faults. Our method uses iteratively re-weighted least-squares with the Geman-McClure function. The requirement that the faults terminate within the data is necessary so that dips can be summed around the faults in order to remove the structure.

For faults that do not terminate within the data cube, this method still may indicate their location. This is because dips estimated at faults tend to be more erratic than other dips away from the fault and are usually treated as outliers.

There are still several weaknesses and areas for improvement of this method. If the fault termination is close to the boundary of the data, this method has the unfortunate side effect of connecting the fault to the data boundary, creating a separate fault block. It may also have the tendency to create false faults if there is significant noise. Lastly, we still may not have determined the best weight function.