We present a pipeline for automatic detection and analyses of flares, primarily using the times series data obtained from the Kepler Mission. Both short and long cadence data from Kepler are very useful as they are continuous for a long time. We developed layers of detrending to remove various astrophysical and systematic variations, using a variable-degree polynomial function, an ingenious iterative Fourier detrending method and trend estimation by spline-fitted Hodrick Prescott filter. Data was lightly filtered using an original algorithm to subtract noisy variations and cosmic rays, accompanied by a Wiener filter. Flares were detected using sigma clipping algorithm with a 2.5-sigma threshold. Some other conditions were also applied to the candidate events to minimise false detections. Detected flares were further analysed in order to get various flare parameters like flare duration, flare energy, etc.
Project student at ARIES