Exoplanet transit analysis using the point spread function

摘要

The behavior of the point spread function could be analyzed in images of stars with known exoplanets. A transform plot of the log intensity versus the Gaussian or sinc function value was used in the analysis. The intercept was extrapolated for brightness information and the slope or adjacent log intensity difference was also fitted. The robustness of the algorithm was evaluated using a bootstrap statistical method that does not require the error to follow a Gaussian distribution. The bootstrap confidence interval was used to assess the changes in the signal before and during transit with simulations that include Gaussian errors as well as non-Gaussian ones. Simulations using Gaussian noise suggested that transit signal changes at about 1-sigma noise level are detectable with a 90% confidence interval, while non-Gaussian noise simulations suggested a lower confidence level. This bootstrap method and the standard Chi-square method used in the analysis of the point spread function of data taken with space telescopes such as Spitzer was compared. The impact of using under-sampled CCD observations with super-sampled point spread functions on the bootstrap analysis is discussed. The analysis of images such as HD 209548 (captured by our 16-inch telescope) suggested that the point spread function approach is comparable to the photometric approach but requires fewer images. The application to other images from telescopes such as Hubble is also discussed.