It's EXTREMELY hard to find out what is the composition of a planet. They don't emit light of their own, and they are extremely dim (only on very very very favorable results they might actually measure their presence, but it's just one pixel in the image!!). Measuring the light from a body is crucial to find out its composition, by spectroscopy. And spectroscopy is quite hard on its own.
There are other ways to find out the body's temperature (e.g. by measuring its brightness) under different "color filters" (visible, near infrared, mid-infrared, far-infrared, etc etc) but the information from such techniques can also be very limiting when compared to spectroscopy.
Usually this is all done from the ground. There are plenty of great ground-based telescopes (with relatively modest sizes) which are involved in planetary astronomy. But this post from the blog Bad Astronomy can illustrate how spectroscopy can be useful for finding out a "candidate"'s conditions:
http://blogs.discovermagazine.com/badast…
Unfortunately, this is also a poor example. The LCROSS probe orbited the Moon, so what they've done here (which was only an instrument calibration exercise, really) was akin to photographing a lion with the camera three inches from its head... which is not what happens in real life. The images shown there are several pixels across, and quite bright.