Orientation for New Members of the KELT Follow-Up Network

The KELT project operates as a scientific collaboration to make scientific use of the data from the KELT survey telescopes.  The primary goal is to discover transiting exoplanets orbiting bright stars, but we also pursue related ancillary science such as eclipsing binaries and other variable stars. 

Members of the follow-up collaboration participate at their own availability.  Members are encouraged to take observations of targets whenever the weather permits, but there is no requirement for how often observations should be made. 

When a member has the time and opportunity to take observations for KELT, the normal process is to check the KELT transit finder the day of the planned observing session, and identify any objects to observe.  Take care to verify that the observation is possible with your facility and conditions.  Some targets are faint and require larger telescope apertures; some are very shallow transits and require the ability to obtain very good photometric precision and minimal systematics, and in some cases particular filters are requested.  

We encourage observers in North America to enter planned observations on the “KELT North America Follow-up Planner” website to minimize duplication of effort in a region where we have a high density of follow-up observers. The planned observations entered by all observers are displayed in a table so that other observers can consider alternate targets or filters. If there are no good alternate targets, consider observing the same target that is on the list, but in a different filter. If all options available on a particular night are being observed by others, it is still useful to have multiple observations of the same event in the same filter to ensure data consistency. The planner website allows entries for the current night and the next night only, which avoids listing candidates before reasonably accurate weather predictions are known. If you enter a candidate for the next night, please check your forecast on the day of observations and make changes as needed.  Please keep in mind that if you list a candidate and filter combination with a 50% or higher chance of success, others are unlikely to observe the same event, so if you determine that you will not be able to observe the candidate, please remove the entry so that we do not miss the opportunity to observe the event.  

The KELT telescope pixels are very large (23”) and the KELT photometric aperture has a radius of 6 arcmin which makes it likely that multiple stars are blended in the KELT aperture. Because of the large aperture, apparent transit detections in the KELT instrument’s light curves can be caused by several scenarios including (1) a candidate star that is an eclipsing binary (EB) which has a deeper depth (generally greater than ~ 40 mmag, depending on the stellar radius) in the follow-up light curve than predicted by KELT due to multiple sources in the KELT photometric aperture (2) a nearby eclipsing binary (NEB), (3) a blended EB (BEB) or hierarchical triple in the follow-up aperture (we sometimes refer to these generically as blends), or (4) a spurious detection by the KELT instrument (i.e. there is no detection of an eclipse or transit in the target or nearby stars).  

The objective of KELT follow-up observing is to sort out which candidates are possible transiting exoplanets from the ones that are not. The first step is to determine which, if any, of the stars in the KELT aperture is experiencing a transit or eclipse. This step can usually be done with transit observations that cover 50-100% of the transit duration plus 30 minutes of baseline. The KELT aperture flux is weighted depending on the distance of the pixel from target star, out to 6 arcmin. Because of the weighting, all detectable stars within 1 arcmin of the candidate should be searched for an eclipse, all stars that are half or more as bright as the target should be searched out to 2-3 arcmin, and very bright stars should be searched out to 6 arcmin. AIJ allows for multiple target star apertures, so AIJ users can perform the target and NEB search with a single photometry run.  

If no eclipse is found within the region as defined above, the period on the transit finder may be doubled for a second check by the follow-up team. After more nondetections at the new period, the target will be expired. If an NEB is found by a greater than 50% transit coverage light curve, the target will be expired. If a transit with a depth that is too deep to be caused by a planet is found, the target will be expired (the KELT science team will determine if the depth is too deep).  

If a transit signal with depth consistent with a planetary companion is found in the target star (or even a nearby star - e.g. KELT-16b), the next step is to check for blends. This is accomplished by observing full transits in blue and red filter bands to check for a depth difference. Simultaneous/alternating blue/red filter observations of a single event from the same telescope are preferred, but larger depth differences can be determined by comparing observations from different epochs and/or telescopes. If significantly different depths are found in blue and red filters, the candidate will be expired as a blend. If a transit signal shows the same depth across filter bands, then radial velocity (RV) data is obtained as a final step to promote the candidate to a confirmed exoplanet, retire it as a blend (i.e. no high precision RV signal found), or to classify the candidate as a rapid rotator which makes RV measurements difficult/impossible (we are now using Doppler tomography to confirm some rapid rotators as planets - e.g. KELT-17b). Notes are provided on the transit finder to guide you in the observations/durations/filters that are needed next for each candidate. 

In general, observations are useful if they cover the entire predicted transit, with at least 30 minutes of baseline before ingress and after egress. If no full transit observations are available, partial transit observations which include at least 50% of the predicted in-transit time plus 30 minutes of out-of-transit baseline are useful if a target has not been previously observed at the currently listed period (i.e. 1 x BLS or 2 x BLS). Partial transit observations are not generally useful to determine filter dependent transit depths. Exposure times should not exceed ~ 4 minutes (preferably less than 2 minutes) so that the transit shape is properly sampled. Observations should use whatever filter will provide the highest photometric precision, unless otherwise specified.  Filters that will provide data usable for publications are clear, clear blue-blocking (CBB, although only observe in the clear or CBB if absolutely necessary), or the full set of Johnson-Cousins, or SDSS ugriz. 

We offer priority rankings of targets.  The source of the priority information is a combination of many factors, not all of which are obvious.  Use them as a guide or suggestion when choosing targets, but your own judgement is best.  The priorities are probably best used to narrow down a long list of choices when there are many different targets to observe at the same time. 

We find that precise telescope tracking and guiding is the most critical factor that enables the production of light curves with minimal systematics. The best results will be achieved when the field position is held within a few pixels throughout the observations. Even with proper image calibration (including flat-fielding), when the target moves around on the detector throughout the observations, systematics of 10 mmag or larger may be produced. These systematics are often larger than the signal we are trying to detect. We request that observers include a description in the submission report of any movement of the field on the detector of more than ~3 pixels from the start to finish of observations. The next most important factor is to select ideally ~8 comp stars that are closest in brightness to the target star which will mitigate the effects of CCD nonlinearly and Residual Bulk Image (RBI; i.e. image retention from one exposure to the next) on the differential photometry. If none are available, try to select some brighter and some fainter comp stars such that the average number of counts within the comp star apertures is close to the target star counts. If possible, choose comp stars that are near the target and distributed around the target star to help mitigate the effects of atmospheric gradients on your differential photometry. Third, if the field is not crowded, defocus your telescope to a FWHM of ~ 10 pixels or more, but not so much that you blend the target or any of the intended comp stars into neighboring stars. Defocusing spreads the light out so that differing pixel sensitivities average out within the aperture, and the peak count of all stars is lower, which enables the use of a wider range of comp stars. If the field is crowded or if the target star is faint relative to the sky background, avoid defocusing which in these cases may reduce photometric precision or increase systematics. If you are at the telescope during observations, AIJ can be operated in “real-time” mode so that you can see how these settings affect the light curve prior to starting the science observations. Do your best to set the exposure time and amount of defocus such that neither will need to be changed throughout the time series. A change to either may produce a baseline offset in your light curve due to CCD nonlinearity or Residual Bulk Image. If at all possible, avoid a meridian flip during your observations, or select a candidate event that would allow you to perform the meridian flip about halfway through the predicted event, such that the flip occurs during what would be the ~ flat part of an exoplanet transit.  

After observations are acquired, the observer should reduce them and send them on to the appropriate KELT follow-up email list quickly - within 1-2 days is preferred.  Results can be provided after a longer wait time, but the longer the wait, the less useful they may prove to be. 

The preferred method for data reduction is the AstroImageJ (AIJ) package.  Other tools are permitted, but we have a great deal of expertise with AIJ and can provide extensive advice about using it. See the AIJ website to install the free package on your Mac, Windows, or Linux system. If you are a beginner at photometry, Dennis Conti has written an introduction to exoplanet transit observing based on AIJ. A step-by-step guide to differential photometry using AIJ is available in Chapter 10 of the AIJ User Guide available at the AIJ website link above. An example dataset is also available for download that can be used to practice photometry in AIJ before you begin observations. For a detailed description of AIJ, see the paper. If none of the above resources answers any AIJ question you may have, post a question on the AIJ User Forum, where Karen and/or other AIJ users will provide helpful answers that may be useful to other AIJ users. If you truly think the answer to your question would not be useful to other AIJ users and you don’t want to post on the forum, email your questions directly to Karen Collins at karenacollins@outlook.com. Observers should extract differential photometry using appropriate comparison stars, and detrend against standard parameters, like airmass.  You do not need to immediately be a photometry expert; we will provide guidance as you get started, and as long as you are able to take data on a regular basis, you will learn the normal procedures quickly.  The key step is to extract the light curve of the target star, along with any other stars near the target, based on the discussion above.  

In your observing report, send an email to the follow-up email list. Please include the full target name, the UT date, the filter, and the observatory in the email subject line.  Send one email per target per night.  The email body doesn’t need to contain much, just a description of the observations and any relevant comments about the observing conditions and results. Include a description of any movement of the field on the the detector (e.g. due to bad tracking and/or guiding) of more than ~3 pixels from the start to finish of observations. Attach a plot of the lightcurve, preferably in jpeg or png format. Always include the raw differential target star lightcurve at the top of the plot. If you choose to detrend and/or fit the data, plot that result under the raw lightcurve (6/10/16 - Karen). Please do NOT paste the plot into the body of the email.  Do the same with a finder chart from the observations, with the position of the target and comparison stars marked. A finder chart that includes sky orientation and pixel scale markers is strongly preferred (AIJ provides this capability). Finally, include a space or tab-delimited ASCII table of the lightcurve, which should include the time, OOT normalized differential target star flux, normalized target star error, airmass, and any other parameters you found useful for detrending.  The brightness may be provided in magnitudes or millimagnitudes, but OOT normalized flux is preferred. 

Also see the wiki pages Information and Names of non-KELT Transit Candidates and Example Cases.

Summary

1) Check KELT Transit Finder to determine potential nightly observations {not linked above}

2) If observing, use KELT North America Follow-up Planner to "announce"

3) Use AIJ to analyse

4) Send report to appropriate follow-up email list