Spitzer Space Telescope observations of hot Jupiters

Abstract

Currently the Spitzer Space Telescope is the most reliable telescope for conducting secondary eclipse observations of exoplanets. The depth and the time of mid-eclipse are two important parameters that come from a secondary eclipse analysis. The eclipse depth gives information on the temperature of the atmosphere, and can provide evidence for the presence of molecules in the atmosphere of the planet. If multiple wavelengths have secondary eclipse depths measured then it is possible to constrain the spectral energy distribution (SED) of the atmosphere given some assumptions on, for example, the metallicity of the planet's atmosphere. The time of mid-eclipse gives e cos which, with an analysis including transit, radial velocity and secondary eclipse data, can strongly constrain the eccentricity of the planet's orbit. To fully understand the conclusions drawn from these two parameters realistic error bars must be quoted
on the measurement of these parameters. It is generally understood that error bars that come from MCMC analyses of secondary eclipse observations are underestimated
because the correlated noise in the data is not accounted for in the analysis. The goal of this thesis was to find a method to improve the estimates of the uncertainties on these two parameters as derived from Spitzer secondary eclipse lightcurves at 3.6 um and 4.5 um. This work was conducted through the generation and fitting of semi-synthetic Spitzer secondary eclipse light curves. I estimate the amount the uncertainties on these parameters need to be inflated by and show how my results compare with other similar work in the field. I show that the amount of inflation does affect the conclusions drawn
when fitting these data with model atmospheres. This could also mean that for systems where complex chemistry is invoked to explain the observed data, simpler model can now fit the data due to the increase in the error bars. I also find that when multiple realisations of the same, simulated, secondary eclipse lightcurve are fit with the standard MCMC code, the amplitude and time of mid-eclipse can be recovered and found to be more than 3 away from the true value of the injected signal. This can mean that, because usually only 1 lightcurve is obtained per observation of the secondary eclipse, some detections of eccentricity and molecules may not be real detections but simply a result of noise in the data.