Body of water inland of Southwold, UK and PGS data

Over the weekend I surveyed alongside the river in Southwold. Combined with existing GPS tracks, the PGS sourced river was clearly in the wrong place. PGS is noted as having offsets, so an initial misalignment was not unreasonable. Adjusting the river and harbour mouth was straightforward, but what of the large body of inland water which is part of the cosatline boundary ?

As the errors with PGS are reported to be offsets, I offset the inland water by the amount that had been needed to line the river up with the bridge about 1 km inland. The result looks reasonable, but I suspect it could be better. Does anybody know enough about PGS data to advise on how to do better? At the western end the water now abuts the road and one stream has an unlikely kink.

[In the current data, the ways are split just upstream of the bridge. Below that node by node edits were made, above that the boundary was moved as a unit such that the resulting shape of the river was as close as possible to the orginal PGS data, just not in quite the same place.]