Help

discussion

How-To: How do I spot planetary transits?!

Started by arralen
Default_user
over 8 years ago

At first let me say - I won't tell you what to do en detail.

To figure out how you can spot those transits best is up to you, because not everyones perception and thought processes are the same. Sorry for luring you in with the promise of an easy way ;-)

What I am going to do is show you the basics 'bout lightcurves, how they come into being they way we get them to see, and what to look out for on the search for those ****** elusive planetary transits.

page 1

(1) a nice, quiet star without any transits - the diagram basics

(2) a closer look at our measurement - what does noise do

(3) We've got Jupiter! - our first, nice and easy to spot transit

(4) But what about Earth? - much more exciting then gas giants

(5) Reality is not that nice ... - how hard it usually is to spot earth-likes

(6) Neptune appears for the rescue - what you can expect to spot on PH.org

(7) Neptune, sovereign of the Sea of Noise? - more difficult detection because of noisier signal

(8) Is Neptune inclined? - not all transits happen centrally

(9) Neptune - hot and speedy! - inner orbit, short transit, multiple occurence

page 2

Default_user
over 8 years ago

(1) Let's look at a nice, quiet star without any transits as a first step. You'll notice that all measurement points line up just fine, on the 1,0000 line - that means the relative brightness of all measuring points (represented by dots in the diagram) is the same, and therefore equal to the average. I've cut the time axis (x-axis) to 3 days because this should be sufficient for our means.

lc without transit

Default_user
over 8 years ago

(2) A closer look at our measurement: Let's "zoom in" a bit on our dots and let's see if they really line up that well ... obviously, they don't! I have only changed the scale of the y-axis from 0.9-1.02 to 0.9996-1.0003, just like the (auto-)zoom of the Planethunters interface would do - and all of a sudden our nice simple line explodes into thousands (ok, rather 145) fragments.

What happend? Actually, we see the "noise" in the measurement. Noise comes from tiny fluctuations of the star, our photo sensors, the amplifiers etc., and maybe a myriad other sources which needn't bother us. All we need to know is - noise makes our dots randomly jump above or below the average (1.0000 line). Yet, it's still the same data as in the first pic, and there's still 1 dot for every 30min (0.5 hours) of "recording time", just as in the Kepler data. And no transit.

To anyone who has looked at some lightcurves already, the picture will look rather familiar. As you'll have already noticed, the "lightcurve" looks rather tame - no wonder, because a +/- 0.0002 noise amplitude is as quiet as things can get with the Kepler data.

lightcurve with random noise

Okay, someone might argue that this is a linear noise distribution - and the noise in the original lightcurves isn't linearily distributed. Right. I made things simpler for me - and I doubt it makes much of a difference for this tutorial. If you have better -and easy to understand- pics, send them over, and I'll put them in.

Default_user
over 8 years ago

(3) We've got Jupiter! Now I added a central° transit of Jupiter to our sun. "Our sun" in every way - I used the values of the solar system, but moved Jupiter into much closer orbit, so its crosses the suns disc much faster than it really would. For details, look at the screenshot on the bottom of this post. ( I do not recommend that planet transit simulator, as it's "simulated measurements" are crap and missleading!) °"central" means just that - that planet crosses right in the middle of the stars' disc

The scale on the y-axis has changed again (like PH would do, too), to accomodate the deep U dip that Jupiter causes. Yet we can still see that our dots deviate from the theoretical lightcurve (red line) because of the noise - noise does not vanish because a planet transits in front of the star!

That means that all dots which are not part of the dip have not change vs. the pic above, and those that are part of it only have been moved down the appropiate amount - if things look different on the upper part of the lightcurve, it's simply because the scale of the y-axis has changed again!

theoretical jupiter transit plus noise.

If you wonder how we get that trapezoidal theoretical lightcurve, look here

And here's some info on the simulated planetary transit:

The transit simulator screenshots were made with the Extrasolar Transit Simulator of the University of Nebraska-Lincoln.

Default_user
over 8 years ago

(4) But what about Earth? Ok, let's have a look at Earth transiting in front of the Sun. Note the changed y-axis scale: the signal is much much smaller.

Do you see anything? In front of the Sun? On the lightcurve? Then you most likely are a gifted planet spotter or an expert anyway and don't need to read any further. In fact, really earthlike planets have been found mostly impossible to spot by eye in the Kepler data of the first 3 quarters of observation time (0,1 and 2).

Remember our example here is a notedly quiet star and data pipeline, in reality the signal most times is much noisier, and the star may fluctuate, have flares or starspots, which can make the noise several times bigger than it is here.

I have sometimes been asked "how to spot those", especially as there are quite some of them in the simulated transits people get to classify. Answer is - you don't, unless you're very very lucky and have lots of experience.

theoretical earth transit plus noise

Default_user
over 8 years ago

(5) Reality is not that nice ... You still think you would spot earth easily? No problem - let's add a bit more noise, up to a realistic level. I'll get back to the noise level later and in more detail - now, simply marvel at earths ability to hide in plain sight ;-)

This is how transits of earthlike planets will appear to you most times on PH.org :

theoretical earth transit plus more noise

Sometimes you'll get lucky : Chat: It IS possible to see Earth-sized planets, sometimes

Default_user
over 8 years ago

(6) Neptune appears for the rescue: With Jupiter transits just sooo obvious and Earth mostly impossible to spot, we should move on to planets everyone could spot, although not without problems, right? Neptune fits quite well in this regard - there are (or should be, according to theory) more Neptunes than Jupiters, and we should be able to spot them in the Kepler data easily enough. There should be even more Super-Earthes out there, but things are getting trickier around them ;-)

So, here's Neptune. The y-scale has changed a bit again, and we have a nice, clear U-shaped transit. Yet we can see the influence of the noise, which still is +/-0.0002 what, as I said, is the minimum you'll encounter on PH.

Note that single dot on the left (ingress) flank of the transit - why is the right one missing?

  • our transit isn't properly centered on the 1.5 day mark

  • Neptune needs a bit less than 0.5 hours to get completely out of the way in front of the sun, so in case we're "unlucky", the telescope will just not catch the moment when Neptune is right on the edge.

theoretical earth transit plus noise

Again, I have pulled in Neptune to inside the Mercury orbit, so we get a nice well-arranged transit in our 3-day timeframe (7.5hours) - in reality it has a transit time of 71.3 hours - that would pretty need our whole diagram ...

Default_user
over 8 years ago

(7) Neptune, sovereign of the Sea of Noise? Now, let's up the noise level a bit, to a more typical, annoying level - yet still no pulsations, flares or starspots.

At first, I just went from +/- 0.0002 to +/-0.0005 noise amplitude, but that looked plain ugly and unrealistic, so I added both noise sources up, and the result is pretty near to the real thing, IMHO, as we have

  • periodically appearing outlyers

  • "knots" where the noise seems to vanish

  • linear gatherings of dots which are just random

  • fake dips

  • false transit flank steepness (on the left)

Default_user
over 8 years ago

(8) Is Neptune inclined? Up to now, we only looked at planetary systems exactly edge-on. Chances to encounter those are pretty slim, as anyone would agree without doubt.

Here's what happens if Neptunes orbit would be inclined (tilted) vs. our line of view by only 0.82° ... any bigger, and it would not pass completely in front of the sun any more, and at 0.885° there wouldn't be any eclipse any more!

What is remarkable here?

  • there's nearly no "floor" left, turning the U-shape of the transit dip into a V-shape

  • the transit has gotten much narrower - only 5 of 14 dots left

  • the overall depth has not changed, means we still get a strong signal from a Neptune-sized planet

Default_user
over 8 years ago

(9) Neptune - hot and speedy! Let's move Neptune around a bit, towards the Sun again. Nearer to the Sun means it has to move faster on its orbit, so not only it's "year" -the time to do one circle around the sun- gets shorter, it also passes in front of the Sun much faster.

Like in the previous example, we're down to 5 dots per transit. Theoretically, we still have U-shaped transits, though - but the random noise turns those Us into something .. random .. yet still "boxier" than the v-shaped grazing transits we've seen above in (8).

Most important: Short, U-like transits mean a fast orbit, mean a short orbital period - and we should get to see several transits in our diagram, especially if it spans 30 days. If we don't see them, it's either a grazing transit (not very probable) or no transit (much more probable)!

In this case, Neptune has a transit time (eclipse) of 3 hours, and an orbital period of 4 days! This is a so-called Hot Neptune (wikipedia) , which appears as small trench-like dip every 4 days.

Please Log In to make comments.