[Previous] | [Session 5] | [Next]
J. Caldwell (York University), W. Borucki (NASA Ames Research Center)
Kepler will detect Earth-like planets by monitoring 100,000 stars over four years for planetary transits. The required photometric precision is one part in 100.000. It is expected that if such ``Earths" are common, about 200 will be detected. In order to achieve the necessary precision, Kepler will be intentionally unfocussed, spreading the light of a single star over an area of 25 pixels. This will minimize the effect of space-craft jitter on photon counting. However, it will also allow the possibility of confusion with background objects which may be in the line of sight to a Kepler target. The greatest concern is that there may be a distant eclipsing binary star which could introduce a photometric signature that is similar to a planetary transit. For the brightest stars in Kepler's intended magnitude range, which is 9 to 14 mv, this will not be serious, because the profiles are different: eclipses have a ``V" shape, transits are flat-bottomed, and Kepler will differentiate the two. However, in this magnitude range, the number of stars per magnitude doubles at each fainter magnitude. More than half of Kepler's discoveries will be in the magnitude which is the faintest in which the precision of the photometry will be able to reveal a transit. That is, most of the discoveries will be low signal to noise events, in which the reality of a small decrease in the light from the region of the target star is certain, but the details of the decrease are not. Hubble Space Telescope images indicate there will be, on average, 0.5 background objects in the magnitude range that could be a problem for Kepler in the 25 pixel blur region of Kepler's optics. Approximately half of the stars will be binaries. The probability that a binary will be eclipsing is the same as that a planetary orbit will be transitting. In order to reduce the chance of a misidentification, various strategies can be used. Rather than integrating the signal over the 25 pixels and returning only the sum, the entire pixel set can be returned for some or all of the target stars. The spectral bandpass can be filtered to maximize contrast between target stars and background ones. Dedicated Hubble imaging could eliminate all uncertainty for over 90 per cent of Kepler's target stars in one HST orbit per star. Further, moving to high galactic latitude would reduce the chance of confusion faster than the decrease in the number of targets stars. Our and other studies indicate that at high galactic latitudes, a large fraction of the background objects are galaxies rather than stars. Galaxies cannot produce a photometric event which mimics a planetary transit. Finally, our studies have shown that a large fraction of the stars in the magnitude range of concern to Kepler may in fact be cool white dwarfs, from which the probability of a confusing event is small. Nevertheless, we acknowledge that a few per cent of the 200 Earths that Kepler is expected to find may be erroneous, and we urge travellers to confirm their hotel reservations directly before setting out to visit one of them.