Exoplanet atmospheric characterization is still in its early stages. Large surveys like the Kepler Mission provide thousands of planet candidates, but follow-up observations to characterize the individual candidates are often difficult to obtain. In this thesis, I develop a method to detect small atmospheric signals in Kepler’s planet candidate light curves by averaging light curves for multiple candidates with similar orbital and physical characteristics. I also consider two applications of this method: at secondary eclipse, to determine the average albedo of groups of planet candidates, and near transit, to determine whether on average the planets have cloud-free, low- mean-molecular-weight atmospheres, or cloudy/hazy/high-mean-molecular-weight atmospheres. This approach allows the measurement of properties that are un- measurable for candidates individually, because of their low signal-to-noise, and it prevents biasing of the results by false positives (candidates that are not actually planets) and outliers by not depending on only a few measurable candidates.

I first develop the method and apply it to the secondary eclipses of 32 close-in Kepler planet candidates between 1 and 6 R⊕ with short cadence data available, in

order to determine their average albedo. I then adapt the method to the long cadence data, accounting for the effects of the longer integration time. The increase in the number of candidates available in long cadence allows for finer radius groupings of 1 to 2 R⊕, 2 to 4 R⊕, and 4 to 6 R⊕. The short cadence average includes 6,238 individual eclipses, while the long cadence averages contain 80,492 eclipses from 56 candidates in the 1 to 2 R⊕ bin, 22,677 eclipses from 38 candidates in the 2 to 4 R⊕ bin, and 4,572 eclipses from 16 candidates in the 4 to 6 R⊕ bin. In both studies, I find that these planet candidates are generally dark, though there are bright outliers like Kepler-10b, and I discuss the implications of these results for understanding the atmospheres of these planets.

Finally, I apply the method to Kepler planet candidates in short cadence near transit, looking for a brief brightening due to light that is refracted through the atmospheres of the planets and directed toward the observer just before and just after transit. Refracted light is strongest in planetary atmospheres that are cloud-free and have a low mean molecular weight. Preliminary results suggest this strong refraction effect is not present in the selected group of 10 candidates with radii between 0.8 and 3 R⊕, but I begin to develop a more detailed model and sketch out future plans to improve the model and to continue testing for the presence of refracted light with greater sensitivity.