The physical size of the crescent Earth as seen from that distance was smaller than one pixel, the reflected light was bright enough to register across a single pixel.

And El Reg...

If you're going to use the 'pale blue dot' meme, it's worth linking to Carl Sagan's commencement address where he used the phrase for the first time. It's a wonderfully evocative piece of prose:

http://en.wikipedia.org/wiki/Pale_blue_dot#Reflections_by_Sagan

...bigger than the pool of honest MPs in Palriament, then.

So Deep Impact has a 30cm mirror and images Earth from 10s of millions of miles. The closest exoplanets are 10-20 light years, which is about 6 million times further away.

Does that mean they will need a mirror 2 million metres in diameter??

Lewis,

After reading that article, I was sorely tempted to suggest and advise that you change your medication, but then thought better of it, as it would need to be addressed to NASA Boffinry instead.

Fine mad reporting is it then yours to claim. :-)

on more than just an intellectual level."

You mean interesting on a Kirk level.

Yes, I think you need water for that.

Arp has observed that once you get a universe with >90% dark matter, observation is irrelevant. The subject should forthwith be spelled asstronomy.

Light collection should scale with area. To get a million times more area we need to scale radius up by thousands -- i.e. a mirror kilometers across. Much easier, but to do it in space might not be so easy unless you can make a mirror out of reflective film and rotate it to keep it spread out; whilst somehow controlling its shape. Oh yeah, with a separate space craft maneouvred accurately to the focus to take pictures.

I've read the paper and their result looks robust, but there's nothing really quantitative about how to apply it to exoplanets, there's basically just a throwaway sentence "we conclude that it should be possible to infer the existence of water oceans on exoplanets with time-resolved broadband observations taken by a large space-based coronagraphic telescope." Hmm.

OK then put in some numbers. Their imaging is done at ~0.5-1.0 microns, at which wavelengths the Earth is about 1E10 times fainter than the Sun, so at 10 parsecs the Earth would be about 30th magnitude. For this technique to work you need to image a few times a day to see the rotation - you can't just sit on the target for a long time to build up S/N.

So, ballpark, we want to get a 30th magnitude target at optical wavelengths in an hour. For comparison, the Hubble Deep Fields get to about ~29th magnitude at these wavelengths with 1E5 second exposures in each band, and a 2.5m mirror. So with S/N proportional to sqrt(time) that's about 27 mag in an hour with 2.5-m mirror, we want to go three mags deeper which is a factor of 15, which gives a ~10-m mirror (S/N proportional to D^2). That might be on the small side depending on how strict we are with our S/N requirement. Anyway - big, but not technologically impossible.

However this does assume that you can blot out the light from the 5th mag star 0.1 arcsec away!

One of the primary reasons that planet hunters are aiming for the IR rather than the optical is that the contrast between star and planet is "only" about 1E6 rather than 1E10. I'm not sure how easy it would be to divert their attention to short wavelengths...