Professional Documents
Culture Documents
222222222 2 222
222222222222222222222222222222222 222 222 222 222 22222222 222 222 282 2
2 22 222 2222222222222222222222222 2 22 22 22 22 22 ,2
2 H2 i222222222 222 2222222222222222 22 22 2ds the central value.
Empirical and theoretical work [Lissauer, et al. 2011] indicates that for neighboring planets, period
ratios below 1.25 are uncommon. The rate of observed T3S1Ms, as well as cases with higher
arity (3 to 8 per Kepler target) indicate that the great majority of T3S1Ms are spurious, and their
presence in a sample indicates the presence and relative frequency of FPE. We henceforth use
the term single T3S1s to indicate T3S1s that are not part of a T3S1M.
The large areas outside hot zones, henceforth cool zones, are not free of FPE, as indicated by
the T3S1Ms distributed widely across the detector surface. It is not possible to expand in any
straightforward manner the geometric extent of the hot zones so as to arrive at large cool zones
relatively free of FPE.
TCEs are detected as a series of three or more putative transits with a favorable signal-to-noise
ratio (SNR). As conceived, a threshold of SNR>7.1 should result in an extremely low rate of FPE
[Jenkins, Caldwell, & Borucki 2002]. A complication in this framework is that Kepler targets have
noise that varies as a function of time, as seen in the logs of quarterly CDPP [Jenkins, et al.
2010]. If the putative transits associated with a TCE occur at a time of elevated noise, and the
noise in the SNR calculation is calculated according to a lower baseline, the calculated noise will
be less than the noise at the time of the putative transits and therefore the calculated SNR will
exceed the true SNR. This allows many spurious TCEs with true SNR below threshold to have a
calculated SNR over threshold. A modest discrepancy in SNR corresponds to a large increase in
the probability of spurious detections. For example, a 20% overestimate in SNR, given a
Gaussian distribution of occurrence and a threshold of SNR=7.1, increases the occurrence rate of
random events over threshold by a factor of 104, from 6.210-13 to 6.710-9.
A useful factor for analyzing the occurrence of T3S1s is seasonal noise. Certain regions on the
Kepler instrument detector surface have noise variation with a strong seasonal cycle as quarterly
rolls change the direction of solar irradiance and therefore temperature-dependent performance
characteristics of the instrument [Jenkins, et al. 2010]. Each T3S1 was observed only during a
particular season, named for the approximately corresponding seasons in the Northern
hemisphere. The electronic artifacts that cause FPE are largely defined by a strong seasonal
cycle that varies locally across the detector surface and can be used to predict likely FPE in the
cool zones.
When a region that is particularly noisy in one season is used to observe the same Kepler targets
three times, with four quarterly rolls between putative transit events, the noise is systematically
underestimated, SNR is systematically overestimated, and FPE occur in abundance.
In principle, a detailed account of instrument characteristics and target noise would allow for an
analytical solution to this problem, estimating SNR correctly in all cases such that a threshold of
7.1 would, as intended, preclude virtually all FPE. It is beyond the scope of this paper to address
all relevant characteristics of the data processing pipeline. We show that a simple model using
quarterly CDPP data [CDPP cite] models FPE generation sufficiently to produce a subset of
TCEs with a greatly reduced proportion of FPE.
Model
Given the detection of P transiting planets, the intrinsic frequency, , of such planets is given by
Equation 1, where CG is the fraction of geometric completeness of observations, CT is the
fraction of such planets that would transit their star at least three times during the time of
observations, and CS is the fraction of such planets whose transit series would exceed the SNR
threshold applied to observations [F12].
444444444 4 444
444444444444444444444444444444444 444 444 444 444 44444444 444 444 484 4
4 44 444 4444444444444444444444444 4 44 44 44 44 44 ,4
4 H4 i444444444 444 4444444444444444 44 44 44 44 444 444
4444444444 4 44 44 44 L4 444444 444444 444444
44444444444444444444444444 44 444444 44 :4,
4444444444444444444444444444444444444 44444444
444444 L4 444444444
04 44 .4 44 44 4 44 444444444444444444444444444444444444444444 444
4444444444444444444444444444444444444444444444 44 444444444444444444
444444444444444444444444444
444444444444444444444444444444444444444444444444444444444444 44444444
44444444444444444444444444 44 4
5 5 555555555
5 555 5
55555555555555555555555555555555555555555555555555555555555555555555555555555
55555555555555555555555555555555555555555555555555555555555555555555555555555
555555555555555555555555555555555555555555555555555555
r
55555555555555555555555555555555555555555555555555555555555555555555555555555
5555555555555555555555555555555555555555555 red (arity3). For astrophysically real
objects of planetary mass, this should be a rare occurrence because the range of allowable
periods varies only 14% from a mode of 370d; furthermore, the weight of probable periods is
significantly concentrated towar
ere observed
at least six total quarters. For these targets, we consider T3S1s as a function of target-season
NObs. It is moreover useful to separate the data for Cool and Hot zones, and also display cases
with single and multiple T3S1s per target-s