Emily Lakadawalla over at the Planetary Org has an interesting post on a recent paper by Sabine Stanley,1* Linda Elkins-Tanton,2 Maria T. Zuber,2 E. Marc Parmentier (in Science). Another of Zuber's students and her had an earlier article arguing for the origin of the dichotomy in a giant early impact based on gravity data (specifically, showing a thinned crust in a circular region -- new because they showed the thin crust ran under Tharsis and thus Tharsis is really "north" of the boundary).

http://www.planetary.org/blog/article/00001710/

The new paper models what happens to a core dynamo if you have higher heat flow in one hemisphere due to that thinner crust (if I'm getting it right), and they find that in the conditions to be expected after a huge impact, you can end up with an assymmetric dynamo that only produces much field strength in the non-impacted hemisphere.

This potentially explains several things:

1. Why remnant magnetism is only observed in the south (it lines up well with Zuber et al's revised dichotomy boundary).

2. Since the field would *not* have had axes lined up with the rotational poles, it makes doing paleomagnetic studies moot and so trying to infer any pole wander from what we know of the fields is wrong.

3. Since there would have been a hemisphere without a magnetic field shielding it, you'd get atmosphere erosion by the solar wind even while the dynamo was still going, and so you could get rid of volatiles much earlier (which seems increasingly consistent with what else we've figured out about the distant past).

Kind of neat. I'd want to see a bunch more models show the same thing, but to the extent more studies seems to be showing how an early giant Northern impact can explain apparent puzzles, the more likely that looks as an answer.

Original paper (subscription): http://www.sciencemag.org/cgi/content/abstract/sci;321/5897/1822