Migrating and merging polarised GPR profiles: does it matter if migration is before or after merger?

A group of radar profiles was gathered across what may have been the site of a canal that was part of the late 16th to early 18th century landscaping at Castle Ward, an estate in Northern Ireland. The canal was filled early in the 19th century, and its exact location was unknown since then. Two sets of profiles were acquired for each of the three survey lines: one set had the antennas perpendicular to the line direction, the common cross-line survey configuration; the other set had the antennas parallel to the survey direction, which we call here the inline orientation. In all cases, transmitting and receiving antennas were parallel to each other. The raw data were compared and there are the obvious differences expected when acquiring data using different polarisations. In addition, we also tested migrating then merging the profiles, versus merging then migrating, to look for any systematic difference in the results. In principle, the final profiles should be the same. However, it appears that migrating first, then merging yields a clearer image of the shallower subsurface (the upper part of the profile image), whereas merging then migrating yields a clearer image of the deeper parts of the profiles. The same features are readily apparent in both profiles; there is no net loss of information nor any difference in interpretation in either case. Nonetheless, the interpretation is aided by processing using both orders – migrating then merging and merging then migrating– so that all relevant features are clearly identified.