The Exoplanet Challenge!

[toml42]

Gentlemen (and Ladies),

This summer I'm conducting a project on transiting extrasolar planets at my University, and it has come to my attention that there are several such planets outside of our solar system in reach of amateur equipment! Indeed, some of the very first were discovered with relatively modest equipment.

For example, here is a page about work done on HD 189733b, a planet larger than Jupiter and closer to it's star than Mercury - done with a 60mm refractor and a DMK camera! I'm sure many of you have a comparable set up, it really is possible

Homepage of Michael Theusner - Images

This page also has a lot of useful background information, although much of it is a little heavy!

The difficult bit is the processing, but as part of my project i'm working on software to ease the experience and it could always use additional test data

Not only is this quite exciting in its own right, but useful science could come of this - and it's perfectly possible that additional, previously undetected planets might crop up in the data!

And so, I submit this challenge to you, Stargazers:

We've all seen and imaged the planets of this solar system - why stop here? I'll be more than happy to supply additional info if anyone would like to have a shot.

Tom

[chemtom24]

tom, sounds interesting - could you give a quick rundown of the procedure needed to gather the data?

thanks,

tom

[toml42]

Sure,

One simply needs to find out when a transit will occur (there are utilities that will allow you to do this), then take regular exposures over the transit period, say every 3 or 4 minutes and for a reasonable length on either side to allow you to find the baseline.

A transit in general will last about an hour, so you'll be looking at a sequence of about 40 images over a period of approx 2 hours, more if you can.

you need to have a reasonably bright star in the same FOV to use as a reference star during processing.

What you'll find during processing is that the brightness of the star will drop several percent over the course of the transit, if you plot the brightness against time, you can then fit a curve to your data which will allow you to infer the size of the planet you are observing

[Capricorn]

HD 189733b has an orbital period of 2.2 days.

So if you imaged for 1 or 2 hours you would not get a determinable amount of difference in brightness.

For what you are describing you will have to pick an exoplanet that has a short orbital period and for the time span mentioned you are looking at a planet that orbits in a few hours. Not aware of any that short.

Silly question but as this is your project shouldn't you be doing it?

[toml42]

HD 189733b indeed has a period of 2.2 days, but a transit - where it eclipses its parent star - lasts about an hour, this is the period where you will notice a dip in brightness.

Due to the short period of many discovered exoplanets so far, transit events aren't too rare.

My project is to analyze data on GJ 1214b that was gathered at the GTC telescope and i'm writing software to make that task easier.

This post has nothing to do with my project, I'm just pointing out that the detection of exoplanets is within reach of the amateur and offering to help with the processing now i have some practical experience.

[FraserClarke]

Sure,

One simply needs to find out when a transit will occur (there are utilities that will allow you to do this)

ETD - Exoplanet Transit Database

This is the best utility I've seen online for working out which transits are observable. Just put in your latitude and longitude, and it will give you times of ingress and egress (start and end) of the transit for each night. It only shows ones which are at least partly visible from your location.

Aim for ones which have a transit depth greater than 0.01 mag (1%) -- anything smaller is really tricky! and ones which have V mag brighter than 11 -- anything fainter is going to get really tricky! The duration of most events is 2--3 hours, so you need to find one ideally where you can get ~5--6 hours continous observing with 1--2 hours before and 1--2 hours after the transit. There are enough objects known now that you should get a good candidate every night or two...

Doing 1% photometry with a modern CCD is very achievable with a little bit of practice. You don't even need a big telescope -- a lot of the transiting exoplanets have been found using 200mm camera lenses!! (quite literally!).

I'm sure I've seen some exoplanet transit light curves on here before? Maybe in the variable stars section?

[toml42]

That's a really useful tool, cheers. Amazing that there are usually at least 4-5 candidates every night! I looked briefly on here to see if it had been done and didn't see any, i might have missed one though.

[FraserClarke]

I found a couple buried in the variable stars section. Unfortunately they are so old the images have disappeared

e.g.;

http://stargazerslounge.com/observing-imaging-double-variable-stars/82094-tres-3b-transit-morning.html#post1238193

[toml42]

Very cool! a whole load has happened in the 2 years since that post, there must be hundreds of suitable new candidates now. I'll certainly be trying as soon as i have viable equipment!

[hemihaggis]

really cool.......

[lukebl]

This is fascinating. I'd love to try myself, but presumably the dip in magnitude is almost imperceptibly small? How is it done and what equipment do you need?

[toml42]

If you follow the link Teadwarf posted earlier you'll see that generally you're talking a few percent of a magnitude dip over a period of about 2 hours, so difficult, but not impossible.

In my original post i linked to a page where observations had been made with a 60mm refractor and a DMK camera, equipment that a lot of people have access to

[FraserClarke]

Here's the sort of thing you can achieve with a 16-inch telescope + a CCD. The top curve shows the light curve of the transit (starts ~0.38, ends ~0.5) and the bottom panel shows the light curve of a comparison object -- to check the signal really comes from the planet, not some other effect. This is Wasp33-b -- which is an odd system because the primary star is also a variable (you can see the slight wiggles in the top light curve, superimposed on the transit signal). The comparison star was much fainter, so the comparison light curve is much noisier -- but still flat. The gap in data is because the telescope lost guiding while I was asleep

You could get very similar with a much smaller telescope here, because the noise is not photon-limited (i.e. a smaller telescope wouldn't necessarily give poorer results).

[toml42]

That's fantastic! When were those taken?

Have you tried dividing the lightcurve by the reference star? that may smooth it out a little. From that lightcurve it would be relatively trivial to calculate the size of the transiting planet.

If it were plotted as ccd counts rather than magnitude, the fractional change in count rates from the baseline to the center of the lightcurve would equal the ratio of the radius of the transiting planet to its host star - which can be readily looked up

[FraserClarke]

I think they are from last October, just after Wasp33 was discovered. They are already differential photometry Tom -- the raw lightcurve (without dividing out reference stars) changes by ~0.3 mag over the night, just due to atmospheric extinction.

We've already done fits for radius etc (not quite as trivial as it might at first seem ) -- it works very very nicely.

[toml42]

Heh, i only meant as a first approximation. I'm finding first hand that a complete solution isn't simple! limb darkening isn't fun...

[themos]

It helps to be on a mountain-top...

Exoplanet observations

[toml42]

That's very impressive, i would think radial velocity measurements of that accuracy would be a far more difficult task

[lukebl]

So what software do you need to produce the graph of the light curve?

[chemtom24]

So what software do you need to produce the graph of the light curve?

might be wrong, btu isnt that what hes writing atm?

[toml42]

There are fitting routines available for the task, they're not particularly user friendly though. There's a version supplied on the first link i posted, but you need to be able to code in fortran or IDL, there's also a python version.

I'll be happy to make my version available when i'm done with it

There seems to be an online fitting tool at

ETD - Exoplanet Transit Database

i haven't used it myself, so i don't know how effective it is

[4lefts]

I'm just pointing out that the detection of exoplanets is within reach of the amateur and offering to help with the processing now i have some practical experience.

Really glad you did - I had no idea this kind of things was even possible with amateur stuff (though I will say the line between pro and amateur is quite blurred as kit gets cheaper (sort of), and lots of people on here seem to have a lot more money than me).

i was really impressed that the stuff in the link in the original post was unguided and with a 60mm refractor. even i can afford that! i don't think my astronomy (or astrophysics) chops are up to it though...

[Chris]

Hi Toml42,

I produced some light curves for the planetary transit of HD209458b as part of my final year project in 2004. The only thing I would have done differently in hindsight is that I would have taken many more images prior, during, and after the transit. I too only took an image every 4-5 minutes and when it came to the statistical analysis I was gutted to find that I didn't have a scientifically significant P-value, so even though I could clearly see the the dip in magnitude on the light curve I could not prove it beyond a shadow of doubt, I would hate to see someone undertake the same project and make the same mistake

I've been meaning to "get back on the horse" and redo the HD209458b transit with images every minute, but I would have to do it with a DSLR, I can't think of a reason why a DSLR wouldn't work?

Kind Regards,

Chris