Detection of TrES-2 Exoplanet on Draco

 Nearly four weeks after successfully detecting an eclipsing binary I decided to tackle a much more challenging task: detection of an exoplanet.  The decision about what exoplanet to detect was made by the fact that the transit of TrES-2 was predicted to occur the night I was interested on trying this effort; August 18, 2007.  Very quickly I realized that even though the basic setup process was going to be the same, the level of accuracy needed was two orders of magnitude greater.

TrES-2 Data Acquisition

The equipment used was the same used to detect the eclipsing binary:  my 12” Meade UHTC OTC at f/10 mounted on the Paramount ME with the ST10XME (binned 2x2 this time) as the detector.  TrES-2 is slightly dimmer than SZ Herculi so I decided to use 10-second integrations instead of 5-seconds to ensure a high signal to noise ratio.  I do not have any “scientific” filters so I selected the Clear filter to image GSC 3549-2811 the target star around which the TrES-2 exoplanet orbits.

Luckily it was clear in Central Ohio , where I have my imaging equipment, and I had the expected time of transit for that rare clear night.

 I was surprised to note that unlike an eclipsing binary the planetary transit takes a lot longer so the expected plot is shaped like a “U” as opposed to a “V”.  This time I waited until TrES-2 had transited so I wouldn’t have to worry about the pier flip at the meridian.  As before, confirming accurate focus, getting a guide star and ensuring that the reference stars were in the field of view took me a little more time that I anticipated thus I started about 20 minutes into my planned 3 hour imaging session. 

Once the imaging session was started the system performed flawlessly and I was lucky that no clouds formed as it was very humid that night.  The total time of acquisition was the planned 3 hours and at the end I also took 10 darks, bias and (given the accuracy needed) flats.

 Processing and Data Reduction

After a good morning sleep (I made it home by 4 AM) I was anxious to start so after creating master darks, bias and flats I quickly processed all 153 images using Maxim DL’s  “Batch Save and Convert” function. 

 As mentioned above I used the guider for this imaging run so I did not have to align the images.  That allowed me to immediately use the Photometry tool in Maxim DL.  After solving the plate with Pinpoint I selected GSC 3549-2716 (mag 10.77), GSC 3549-2760 (mag 13.19) and GSC 3549-2837 (mag 12.30) as reference stars.  All the reference star magnitudes were taken from the astrometric information provided for the GSC catalog by Pinpoint.

 The first plot of the data basically showed four horizontal lines for all four targets which it was to be expected due to the good seeing conditions present that night, there was absolutely no indication of any magnitude change.  Of course, if the expected change in magnitude was about 1.5% the plot needed to be scaled so it would show that small a change.

 I adjusted the scale of the plot and was surprised to see that, even though noisy, there was a definitive “dip” in the TrES-2 star magnitude!  But was it real?

 One way to know if the apparent change in magnitude is real is to subtract the magnitude change of the object star (TrES-2) to the magnitude changes of one of the reference stars, in other words, find the difference.  This way changes that are caused by air mass extinction, clouds or other external phenomena are cancelled out and you are left with a true picture of the change, also known as the differential change.

Even though differential analysis is good a problem may occur if the  reference star(s) used are not stable.  There have been documented cases where previously unknown variable stars have been discovered due to the sensitivity and accuracy of measurement needed to analyze exoplanet data. 

Results

I learned about Bruce L. Gary (Hereford Observatory), another amateur astronomer that has worked with professionals astronomers on the detection and measurement of several exoplanets and creator of the Amateur Exoplanet Archive (AXA).  

Mr. Gary developed an exoplanet analysis method based on using a synthetic reference star and a set of spreadsheets that calculate things like Air Mass (and the extinction losses associated with it) for every observation and the ability to generate light curves that take into consideration all these factors.  I want to thank Mr. Gary for developing this great tool.

The plot above is the result of following through Mr. Gary's advice and performing what he calls "good data analysis". 

The light curve above shows the analysis of exoplanet TrES-2 transiting its parent star GSC 3549-2811 on August 18, 2007.  

There are several components that are part of the light curve generated.  The small red dots represents each actual 10-second images that passed the acceptance criteria for both, extra losses and outlier rejection. The large red dots are 9-point non-overlapping, median combines of the accepted data.  The two vertical lines at the top indicate the predicted ingress and egress times.  At the right of the legend, in the bottom, there are a few more notes; at the upper left it shows that an aperture radius of 10 was used for measuring star fluxes this resulted in an RMS for 1-minute data of 1.72 mmag, which corresponds to an RMS of 0.77 mmag for 5-minute averages.  

The thick grey line is a model of a typical exoplanet transit.  The model illustrates the “U” shape expected with the detection of an exoplanet.

The ability to detect an exoplanet with off-the-shelf commercial equipment from Central Ohio has been an extraordinary experience that adds another dimension to this wonderful hobby.  The reduced data has been accepted by and is now available at the Amateur Exoplanet Archive (AXA). 

Now that I have tasted the excitement of discovery my plans are to continue to try to better the detections.  To better my accuracy and produce scientific grade data I may have to get “scientific” filters and another filter wheel to my inventory.

Isaac Cruz

The Octadome