Searching for planets in the Kepler fields using parameter estimation techniques

Abstract

Man has always been curious of whether or not there is any possibility that in the universe, there are other planets of a similar size and behaviour to Earth. This led to the introduction of several space-based missions with the scope of observing windows of this immense universe for finding such planets. Among these missions, one of the most noticeable and recent is the Kepler space observatory, launched by NASA on the 7th March 2009, [53]. The Kepler spacecraft is outlined to examine part of a region of the Milky Way galaxy to uncover planets in their habitable zone, [53]. Data is then sent to Earth, which is then interpreted to detect periodic dips in the star's brightness caused by extrasolar planets eclipsing their host star - transit detection, [54]. The main aims of this project is to process time series data, with particular application to the Kepler mission, aiming to perform parameter estimation calculations through model fitting to obtain the most similar model, with the best likelihood. Despite this, the huge amount of data to be processed, introduced an increase in the requirement for a system able to process time series in a fast manner, yet still providing accurate estimations for the parameters of interest. As a result, enhancements in speed and precision are two major factors this project aims to address, together with a secondary tool, completing a pipeline approach, which is able to process and detect transiting dips, before estimation is performed. There are however numerous difficulties one would face when tackling the above problem, among which one can mention false-positives eclipsing binary stars, [51], and varying noise levels in data, [25], making it all the more challenging to detect and estimate the best fits for such data. The outputs produced by this project demonstrate the achievements in both the detection and parameter estimation of Kepler time series data, providing information about planet-star systems, together with their transit and orbital parameters. Further to this, the system presented attains comparable to even better results in a faster manner, than current similar existing techniques.