Trends and noise removal using a Wavelet based modification of the Trend Filtering Algorithm: Application to wide-field surveys
Author: Daniel del Ser
Trends, systematics and noise are present in any time domain data. Their presence results in a decrease of the photometric precision and are often cause of other unwanted effects such as incorrect signal characterization or a lower signal detection probability. Algorithms such as the Trend Filtering Algorithm (TFA) or SysRem have been used for some time in astronomical time domain surveys to deal with these systematics and improve the detection and precision of any periodic signal. In this talk we will briefly explain how these algorithms work and present a modification of TFA, called TFAW, that makes use of a Stationary Wavelet Transform based filter that tries to eliminate, as much as possible, any noise contribution and improve the characterisation of the noise- and trend-free signal. The use of TFAW results in a clear improvement of the signal-to-noise ratio, a better period estimation through an improved frequency spectrum and better signal characterization without any signal loss or changes in the shape of the signal. In addition, TFAW can be used to deal with multi periodic signals allowing to separate the different signal contributions and improving their respective light curves. TFAW is a generic algorithm that can be used to any survey provided a sample of reference time series is available. Simulated periodic signals will be presented as well as real light curves coming from the TFRM and Evryscope databases in order to show the efficiency of the algorithm.