Great idea - couple of potential stickies, though, which might make this difficult to implement.

Firstly - the data-maps aren't actually 'images' of the Aurora - they're really just a map of the level of 'electrical disturbance' (for want of a better word) around the poles. That is, the maps really display 'current hotspot areas' where Aurora activity is likely to be highest, rather than displaying the Aurora themselves. This makes the resulting images slightly less, well, awesome.

Secondly, the map-projections used are Polar - and GE can currently only 'talk' Mercator - and the translation from one to the other is rather tough. It's been bugging me pretty much all the time I've been using GE (which isn't long, I grant you!) - because I live in the UK, and ALL our darned sat-maps and weather-maps are Polar Stereographic, and thus virtually impossible to overlay in GE without serious errors the further away from the centre-point you get.

However - it might be possible to overlay the KP maps which are also available on that site. These give a hint as to the likelihood of seeing Aurora (in general, not on a day-by-day, alas) from various positions around the globe. These maps are standard cylindrical (Mercator) projections, and so they SHOULD map ok to GE when overlaid.

Not quite the wonderful project that it originally sounds (and boy, wouldn't it be wonderful if it really WAS a true image of the Aurora?) - but maybe still of use to some 'aurora-hunters' out there?

If I get time after the project I'm currently stuck on, I'll give it a whirl!

Top idea in any case, so if someone else out there thinks they can solve it, please DO go ahead and try! I'd love to see the results as much as the next person!