I offer these statements about the interpretation process:
A contractor's job often consists of enhancing potential-field data to make them as interpretable as possible (by either the contractor or the client). In getting data ready for interpretation, I like to confine the work to “atural processes” and avoid arbitrary processes.
I see too many people abusing the facilities available in commercial software packages and applying polygonal fits of the third, fourth, fifth, seventh, ninth order and calculating residuals until something strikes a chord, and then off they go and interpret the results. Alternatively, all too many people apply arbitrary frequency-domain filters chosen to pretty up the data without a lot of thought about the consequence. Then you no longer have a potential field.