Here in the UK, we have many areas where marshland or salt-marsh has been drained in stages, sometimes over hundreds of years. The flood-bank which formed the edge of an earlier stage usually remains between different blocks of drained land. Only the bank at the current edge of the drainage is in use -- all the others are marooned in the middle of the drained land.
The same process might happen with banks made to retain water, rather than exclude it. This would leave apparently purposeless banks in the middle of the paddies as you describe.
Alternatively, could your banks be used to allow some areas of flooded land to be drained for harvesting or planting, while leaving others flooded?
All this is assuming you mean dike in the sense of a flood-bank -- in the UK we also use dyke to mean a water-filled ditch! (And "downs" are hills...)
Regards,
RichardActually, it is just a perception that they are not following the topography. The change in elevation is so slight (perhaps 1 ft per mile) that the land appears to be flat. However, if there is a slope at all, even such a small one, there will be a point that if you are trully making the land level, there will have to be a break, and it may occur in the middle of the section and it will be curved if the land is undulating at all. I want to add that it is also possible the levee is following an old stream bed.