Kepler Amateur Follow-Up Observations
Webmaster: Bruce Gary
This
web site resumed active status after a
suspension (Apr 24 to May 17). I have now
reassessed the feasibility of amateur
observations contributing to refinement
of TTV for Kepler objects. This reassessment
was prompted by an analysis by Howard Relles
in April showing that most KAFO light curves
weren't accurate enough to be useful for that
purpose. After reviewing how KAFO was
structured I found two flaws: 1) Kepler
candidates with absolutely no TTV at high
precision were included in the KAFO active
list, and amateur observations of these
objects with low precision were a "waste of
time," and 2) KAFO observers were not provided
guidance on which Kepler candidates were
realistic for observation, based on magnitude,
transit depth and telescope aperture. The new
KAFO addresses both flaws: 1) only Kepler
objects with known or suspected TTV are
included in the active list, and 2) a graph is
provided that observers can use for assessing
whether an object is both bright enough and
has a deep enough transit depth to be usefully
observed for a specific telescope
aperture.
_______________________________________________________________________________________________
This website lists Kepler exoplanet candidates with known or suspected TTV that are observable by amateurs. The Kepler project's latest release now includes >7280 candidates; 26 of these are either known or suspected of exhibiting TTV. The "KAFO active list" will show which oof these 26 objects are predicted to have transits during a 2-week interval. Observers are encouraged to select objects for observation based on a graph showing an empirical relationship between V-mag, transit depth and telescope aperture. Long period exoplanets are emphasized because they are in some respects more suitable for a network of amateur observers than a small number of professional observers since amateurs are found at a large range of longitudes and sample a wide range of weather conditions at any desired observation time. Follow-up analysis of mid-transit times reported on this web site will be performed by Sigfried Vanaverbeke (and maybe Howard Relles), and these results will be included in a section of this web page when they become available.
Recent Updates:
Oct 29: Added K00377.01 LC by Gary
Oct 28: Updated events list
Currently Active List: (light curves are at KAFO LCs)
Figure 1 (top). The "Currently Active List" above includes a column labeled "TTV [min]," kindly provided by Howard Relles (http://exoplanet-science.com/KAFO.html), showing the amplitude of TTV variation. A zero entry means that an insufficient length of Kepler observations is available for estimating the period and amplitude of TTV (i.e., period of TTV variability is too long or shape of TTV is too complicated). The last column ("Aperture Needed") is my estimate of telescope aperture [inches] needed to observe the transit with an accuracy useful for KAFO. If the required aperture exceeds 100 inches (based on my ground-based, typical performance, given as Fig. 2) then "..." is displayed. Note: K00377.01 is a special case. It is uncertain by up to 2.5 days, so I've included entries that sample this uncertainty (K003777.01' is for Fabrycky's solution, K00377.01'' is for Relles' solution extrapolated.)
Figure 2 Lower-left. Guidance for assessing feasibility of providing useful timing information for a transit event based on the star's brightness, the transit depth and telescope aperture. The area above the trace is "feasible." For example, a 11-inch telescope observing a star with V-mag = 14.0 can provide "useable" transit observations when transit depth > 10 mmag (assuming sufficient pre-ingress and post-egress data are available, and the light curve quality is good in other respects).
Figure 2 Lower-right. Empirically-based mid-transit time SE for a 12-inch telescope versus V-mag & transit depth (assuming a good quality light curve). For other telescopes I suggest multiplying transit depth by (12-inch/aperture) and sliding the traces right or left by amount = 2.5 * LOG(aperture/12-inch). For example, a 24-inch telescope observing a V-mag = 12.75 star (0.75 mag right adjust is for larger aperture) will have a 1.0 min SE when transit depth ~ 10 mmag. Note: naive modeling of timing SE doesn't allow for systematic effects that vary greatly with observer (reference stars used, photometry apertures used, seeing variability, etc). Actual SE values can be larger by a factor of 2 or 3 (if light curve quality is poor).
The chart above is based on an analysis of previously submitted LCs. I plotted quality symbols on the V/D field and found that a trace could be constructed that separated the useable from un-useable LCs with nearly 100% accuracy. All of the attempts below the trace were un-useable (10 out of 10), and 79% of the LCs above the trace were useable (11 out of 14).
Observers, try to observe the entire night (these are long period candidates which tend to have long transit lengths). Even if only an ingress, or egress, is observed, this can be useful. Combining light curve segments can be achieved if they overlap (or if same filter and reference stars are used, etc). I recommend using a clear filter (or better yet, a Cb filter, clear with blue blocking). Second choice is r'-band filter, or Rc-band filter. Third choice is V-band. Make sure the PSF has FWHM > 2.5 or 3.0 pixels (to avoid "pixel edge noise); if you feel the need to bin 2x2, and if FWHM < 3 pixels, it's suggested that you defocus to broaden PSF. Keeping the star field fixed with respect to the pixel field will reduce systematic effects, such as those related to notoriously difficult master flat fields. Anytime this can't be achieved, and star field wander exceeds ~10% of the FOV, there's no point to constructing a light curve because it is likely to be ruined by unknown master flat systematics. That's my position, and I'm sticking to it! Check to make sure the master dark frame doesn't increase noise (in area without stars); if it does, then consider using auto-optimization (I'm assuming you're using MaxIm DL).
Oct 29: Added K00377.01 LC by Gary
Oct 28: Updated events list
Currently Active List: (light curves are at KAFO LCs)
Figure 1 (top). The "Currently Active List" above includes a column labeled "TTV [min]," kindly provided by Howard Relles (http://exoplanet-science.com/KAFO.html), showing the amplitude of TTV variation. A zero entry means that an insufficient length of Kepler observations is available for estimating the period and amplitude of TTV (i.e., period of TTV variability is too long or shape of TTV is too complicated). The last column ("Aperture Needed") is my estimate of telescope aperture [inches] needed to observe the transit with an accuracy useful for KAFO. If the required aperture exceeds 100 inches (based on my ground-based, typical performance, given as Fig. 2) then "..." is displayed. Note: K00377.01 is a special case. It is uncertain by up to 2.5 days, so I've included entries that sample this uncertainty (K003777.01' is for Fabrycky's solution, K00377.01'' is for Relles' solution extrapolated.)
Figure 2 Lower-left. Guidance for assessing feasibility of providing useful timing information for a transit event based on the star's brightness, the transit depth and telescope aperture. The area above the trace is "feasible." For example, a 11-inch telescope observing a star with V-mag = 14.0 can provide "useable" transit observations when transit depth > 10 mmag (assuming sufficient pre-ingress and post-egress data are available, and the light curve quality is good in other respects).
Figure 2 Lower-right. Empirically-based mid-transit time SE for a 12-inch telescope versus V-mag & transit depth (assuming a good quality light curve). For other telescopes I suggest multiplying transit depth by (12-inch/aperture) and sliding the traces right or left by amount = 2.5 * LOG(aperture/12-inch). For example, a 24-inch telescope observing a V-mag = 12.75 star (0.75 mag right adjust is for larger aperture) will have a 1.0 min SE when transit depth ~ 10 mmag. Note: naive modeling of timing SE doesn't allow for systematic effects that vary greatly with observer (reference stars used, photometry apertures used, seeing variability, etc). Actual SE values can be larger by a factor of 2 or 3 (if light curve quality is poor).
The chart above is based on an analysis of previously submitted LCs. I plotted quality symbols on the V/D field and found that a trace could be constructed that separated the useable from un-useable LCs with nearly 100% accuracy. All of the attempts below the trace were un-useable (10 out of 10), and 79% of the LCs above the trace were useable (11 out of 14).
Observers, try to observe the entire night (these are long period candidates which tend to have long transit lengths). Even if only an ingress, or egress, is observed, this can be useful. Combining light curve segments can be achieved if they overlap (or if same filter and reference stars are used, etc). I recommend using a clear filter (or better yet, a Cb filter, clear with blue blocking). Second choice is r'-band filter, or Rc-band filter. Third choice is V-band. Make sure the PSF has FWHM > 2.5 or 3.0 pixels (to avoid "pixel edge noise); if you feel the need to bin 2x2, and if FWHM < 3 pixels, it's suggested that you defocus to broaden PSF. Keeping the star field fixed with respect to the pixel field will reduce systematic effects, such as those related to notoriously difficult master flat fields. Anytime this can't be achieved, and star field wander exceeds ~10% of the FOV, there's no point to constructing a light curve because it is likely to be ruined by unknown master flat systematics. That's my position, and I'm sticking to it! Check to make sure the master dark frame doesn't increase noise (in area without stars); if it does, then consider using auto-optimization (I'm assuming you're using MaxIm DL).
