h a l f b a k e r y
It's not a thing. It will be a thing.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
The new Kepler telescope is designed to detect planets that transit in front of their stars, however this only works for about .5% of stars that are on the correct plane for viewing. For detecting planets that are on different planes, perhaps background stars could be used?
First, remove the target
star from the data, then watch the background stars for recurring light-dips, indicating a planet.
This will probably need a whole new telescope, I'm not sure Kepler could handle it.
"Methods of detecting extrasolar planets" at WIkipedia.org
While difficult, direct imaging isn't impossible. (And apparently background stars can be used to detect foreground planets - via lensing) [phoenix, Jun 05 2009]
How to identify the regular pattern of a norbit
Sorry. [phoenix, Jun 05 2009]
[simonj, Dec 07 2011]
[beanangel, Dec 09 2011]
||I'm not sure and I'm just guessing, but I don't think it
would work. (a) The planet's own sun offers a far, far
bigger target disc than distant stars, so the chances of the
planet occulting another star might be small (?) (b) I don't
think you could digitally process out the overwhelming
brightness of the planet's own sun and still see very small
changes in brightness of "adjacent" stars. This, after all, is
why it is so hard to see the planets directly (ie, you can't
subtract out the star's own light without losing the
planet). (c) all stars move relative to one another, and
also a planet's orbit will precess; I would imagine this
means that the planet will not occlude the same
'background star' repeatedly, and it will be very hard to
identify the regular pattern of a norbit.
||You could mask out the target star ie. have a physical
barrier in the centre of the lens.
||A] //chances of the planet occulting another star
might be small// If your seeing distance is large
enough, the sky is very, very crowded. If the
universe were infinite the night sky would be
bright. Of course, it's not, so there are dark
spaces. But your point A might also be used to
argue that there was no hope of ever seeing
gravitational lensing. And yet that's been done.
Intuition's not reliable here: t's a quantitative,
question, that an astronomer could answer.
However, the fact that this isn't done already
suggests that some astronomer has.
||B] //digitally process out the overwhelming
brightness of the planet's own sun and still see
very small changes in brightness of "adjacent"
stars.// You don't need to. You just have to
detect a very small AC oscillation superimposed
on a much larger DC baseline. Which is exactly
the sort of signal processing used to detect
occulting planets. Except, in this case, it's (in one
respect) easier, because the spectrum of the AC
component is different* from that of the DC
||*Even if the spectral lines are identical, the red
shift'll be different.
||C] Who says the exoplanet needs to occult the
*same* backround star on every orbit?
Complicates the signal processing, some, but
that'd be the least of ones worries -- points A & B
||[simonj]: I don't think that's the problem - you
could do the
same thing by electronically "masking" the
relevant pixels on the image. (Actually a physical
mask might be worse, because light will diffract
around its edges.)
||I suspect that there are several reasons why you
can't "subtract" the main star sufficiently well,
though I don't know which ones are most
||A) light scattering. Interstellar dust and stuff
must scatter some light, so a tiny percentage of
the main star's light would be superimposed on
the surrounding field of view. This might be
especially true for stars with planets, since they
probably have lots of dust and crap orbiting
||B) gravitational bending of light. Spacetime is all
rippley, which will have a similar effect to
scattering, in that it will tend to confuse the light
of the main star with that of nearby (in line-of-
||C) Angular resolution. The angle between the
exoplanet (and the eclipsed background stars) and
the main star is going to be incredibly tiny. The
necessary angular resolution to discriminate the
exoplanet (or the eclipsed stars) from the main
star would require an FET.
||So, overall, I think the problem is like trying to
detect a moth by watching it "eclipse" the lights
from a few glow-worms. The moth and the glow-
worms are all a few inches away from an arc-light,
and you're looking at the whole scene from fifty
miles away, through a sandstorm and heat-
||Stars come in different types, right? What if your
detector was not for visible light, but
for some other kind of radiation that the target start
is poor in, but the background stars are rich in?
||That would help. It should also be a doddle to
detect exoplanets orbiting black holes, or mauve
||This is the part of planet-hunting that has always concerned me - we are only looking for / can only see the tiny fraction of planets that have nicely aligned orbits.
It makes you think; there have been about 2000 exoplanets found now, and if [simonj]s 0.5% is correct (sounds about right...), there are a shiteload more out there!
Finding the rest is where the fun science begins!
||This might work for visual doubles i think but i
haven't done the arithmetic yet.
||Edit: Now i will. I'm not aware of any particularly
close doubles, so i'm going to imagine the Alpha
Centauri system is a optical double as well as a
ternary system. If a planet was situated at three
hundred AU and occulted or transited the
hypothetical star at maximum elongation, it would
be separated by almost four minutes of arc. So,
looking at a real optical double, Secunda Giedi is
thirty-three parsecs away and its optical double,
Algedi Prime, over two hundred parsecs away.
They are separated by six minutes of arc. To
transit Algedi Prime, a planet would therefore
have to be almost two thousand AU from Secunda
Giedi. If its orbit at that distance had a semimajor
axis of that size and orbited , that would give it a
sidereal period of thirty-three millenia. A planet
twice the diameter of Jupiter at that distance
would be travelling at six and a half kph. Algedi
Prime is a yellow supergiant with a diameter forty
times the Sun's, but is six hundred and ninety light
years away, so its apparent diameter is four
hundred thousandths of an arcsecond. At a sixth
of that distance, i.e. the hypothetical planet's
distance, the angular diameter of this planet
would be about a seventh of the more distant
star's, and would take nearly four decades to cross
the disc at average speed. This may be very
wrong incidentally, i've lost concentration.
However, since it would be foreshortened at
maximum elongation, it would take a heck of a lot
longer than that.
||So assuming that a planet could orbit that far out
from a star, and there's no information on that so
far as i know, and with these very shaky
calculations, rather surprisingly, it probably does
very occasionally happen. However, the orbital
plane of the planet would have to be oriented
correctly, just as it would with a transit of its own
||I think that if we were able to view every known
star in the sky from every other known star
system, the chances are that now and again this
would happen, but i have no idea how probable it
is that it happens in our own sky or if there's been
enough accurate observation for this to have been
observed. It'd be better from orbit of course. The
best chance would be from an optical double
whose more distant component was very large,
the separation was very close and the planet was
very distant from its primary and very large.
||just being an idiot, perhaps a post heliopause Big Laser could create quantum linked photons that had stabilities of multilightyear duration. Then when the special IR spectroscopy frequency photons reached something thrilling like water, it would create a particular quantum event at the earth system optical observer. The idiot part is mostly, is it really possible to keep huge quantities of quantum linked photons stable while the explorer photons travel light years to be absorbed. One reply is that linked photons lasting minutes are already 10^8 or more longer lived than atomic transitions thus 10^14 might work long enough to detect custom absorbants light years away