My first attempt at detecting an exoplanet

[Dave Smith]

Here is my first attempt at detecting an exoplanet. This was a tricky one for a first attempt as the drop in brightness was only 0.01 magnitude which is right on the limit of my uncertainties. With a little imagination it is just visible. The red lines indicate the predicted start and finish of the transit times (21:54 and 02:18 UT) There are some exoplanets that produce a drop in brightness of 0.02 mag or even greater but on any particular night there is a limited choice of stars to go for. This one is called XO-4b and was found on the Exoplanet Transit Database. My results could look more convincing with more data either side of the transit but dusk and dawn get in the way.

[carastro]

Great stuff Dave.  Not sure I could do it.

Carole 

[Dave Smith]

Thanks Carole. I am sure you could, all you need is the suitable software, I use Muniwin which is free. You just take loads of exposures making sure the star in question is not saturated and then using Muniwin to process it. In this case I used a V filter but that was not necessary.

Dave

[DirkSteele]

Fabulous piece of work. 

[beka]

Very interesting, I think there is no doubt about the dip. May I ask how did you determine that your uncertainty was around 0.01 magnitude?

Best

[swag72]

Wowsers Dave - That's very interesting

[Pig]

Nice work ?I am amazed such small measures of change can be realised from the Earths surface given all the atmosheric variables.

[Dave Smith]

25 minutes ago, beka said:

Very interesting, I think there is no doubt about the dip. May I ask how did you determine that your uncertainty was around 0.01 magnitude?

Best

Thanks Beka. It was the software that calculates the uncertainties. I feed in the camera a and scope details and it calculates the uncertainties and produces an error bar of the appropriate length.

Dave

[Dave Smith]

26 minutes ago, swag72 said:

Wowsers Dave - That's very interesting

Thanks Sara. It certainly makes a change from imaging.

Dave

[lukebl]

Very cool!

I've used Muniwin (on your recommendation) on Variable stars and asteroids, but must try on some exoplanets. Some of them have orbits of only a few hours, so they can't be much like 'our' planets!

[Dave Smith]

Very frustrating but I have just discovered that tonight there is an exoplanet that will cause a drop of 0.029 mag and of course it is cloudy !!!

Dave

[Owmuchonomy]

Very interesting.  Is there a method to validate the set up associated with the software and hardware?

[Dave Smith]

5 minutes ago, Owmuchonomy said:

Very interesting.  Is there a method to validate the set up associated with the software and hardware?

Thanks Chris. I am not absolutely sure what you are asking but I have used the same software and hardware for quite a few variable stars and originally had my results checked by my BAA mentor and for stars well studied have been in agreement with results of others. In this case I didn't go as far as getting the absolute magnitude of the star as I was only interested in detecting that small change. Is that what you meant?

Dave

[Owmuchonomy]

5 minutes ago, Dave Smith said:

Thanks Chris. I am not absolutely sure what you are asking but I have used the same software and hardware for quite a few variable stars and originally had my results checked by my BAA mentor and for stars well studied have been in agreement with results of others. In this case I didn't go as far as getting the absolute magnitude of the star as I was only interested in detecting that small change. Is that what you meant?

Dave

Hi Dave, yes that answers my query very well.  Basically you have increased the confidence in the method by validation on the variables and a third party check.  Great effort.

[dokeeffe]

Great work! Its there alright. I'm wondering is there any way you could share the fits files? Maybe dropbox or something?

Id love to run them through the system I used recently to plot exoplanet light curves. Its called Lemon (https://github.com/vterron/lemon) It uses a special algorithm to find the best comparison stars to use and generally I get a better light curve from it than astroimagej + manual comparison star selection. The only problem with lemon is that it has a lot of dependencies and its a nightmare to install.

Derek

 

[robin_astro]

On 4/23/2018 at 17:37, Dave Smith said:

Thanks Beka. It was the software that calculates the uncertainties. I feed in the camera a and scope details and it calculates the uncertainties and produces an error bar of the appropriate length.

Dave

Hi Dave,

The detection looks significant by eye but the software most likely just calculates the uncertainty based on the counts (ie that due to the random nature of photons arriving. That is only one of many possible contributions to the total uncertainty though and there will be other systematic variability which will add to the uncertainty.  Since you have many repeat points though you can make a true measure of the uncertainty by taking the measurements during (say 23:00-01:00) and outside the transit and doing a t test on the two populations to determine if your observation that there was a dip is statistically significant.

https://en.wikipedia.org/wiki/Student's_t-test

Cheers

Robin

[Dave Smith]

Thanks Robin. That looks most useful. I will give it a go.

Dave

[robin_astro]

Hi Dave,

If you need a t test calculator there are a few on line which will do it for you eg this one which has a nice expanation of the results

https://www.mathportal.org/calculators/statistics-calculator/t-test-calculator.php

You just cut and paste the values for the two sets of data into the boxes. A 1 tailed test is what you want to test if there was a statistically significant drop. (the 2 tailed test just tests if the means are significantly different) Just leave the other settings as  the default

Cheers

Robin

 

[Grotemobile]

That is really amazing what you have done there Dave. I thought that type of

thing was only possible for the professional astronomers.  

[Dave Smith]

56 minutes ago, robin_astro said:

Hi Dave,

If you need a t test calculator there are a few on line which will do it for you eg this one which has a nice expanation of the results

https://www.mathportal.org/calculators/statistics-calculator/t-test-calculator.php

You just cut and paste the values for the two sets of data into the boxes. A 1 tailed test is what you want to test if there was a statistically significant drop. (the 2 tailed test just tests if the means are significantly different) Just leave the other settings as  the default

Cheers

Robin

 

Just tried that and here are the results.

I don't understand the meaning of t or the critical value but it would indicate a successful detection I think?

The mean values indicate a change of 0.01 mag which is spot on !!

Dave

[Dave Smith]

1 hour ago, Grotemobile said:

That is really amazing what you have done there Dave. I thought that type of

thing was only possible for the professional astronomers.  

Thanks Steve. Quite a few amateurs have detected exoplanets e.g. https://stargazerslounge.com/topic/312775-my-3rd-exoplanet-observation/

Dave

[Chris]

Well done Dave, I know what a buzz this can be  You can just about see a dip, I'm just wondering if it would be even clearer if you took the mean average of every 4 or five data points?  

I love that we amateurs our doing this stuff!

[Dave Smith]

Thanks Chris, I did try averaging each batch of 5 data points but it didn't look any better.

All good fun.

Dave

 

 

[robin_astro]

7 hours ago, Dave Smith said:

I don't understand the meaning of t or the critical value but it would indicate a successful detection I think?

The mean values indicate a change of 0.01 mag which is spot on !!

Dave

Yes a t value of 5.9 (with 217 degrees of freedom) is very high so this means that for all intents and purposes,  the measured brightness was definitely lower during the transit period. ie it is statistically highly improbable that a difference in mean this large could have happened by chance.

Cheers

Robin

[johnfosteruk]

Excellent piece of work. In theory you (or anyone) could be discovering exoplanet candidates with this setup.