Final Report
Total Page:16
File Type:pdf, Size:1020Kb
DRAFT VERSION JULY 31, 2018 Typeset using LATEX twocolumn style in AASTeX61 REVEALING THE VARIABILITY OF NAKED-EYE ECLIPTIC STARS WITH K2 HALO PHOTOMETRY MICHAEL GREKLEK-MCKEON1 AND DANIEL HUBER2 1Department of Astronomy, University of Maryland, College Park, MD 20742, USA 2Institute for Astronomy, 2680 Woodlawn Dr, Honolulu, HI 96822, USA ABSTRACT Using the technique of halo photometry, we analyze the brightest stars observed by K2 in campaigns 11-14, searching for previously unidentified variability. We correct and detrend the lightcurves, calculate power spectra for each star, and perform frequency analysis to extract the dominant oscillation modes. We discover seven new classical pulsators, 28 new pulsation modes for previously known pulsators, and 24 new oscillating red giants. We also find evidence for granulation-like signatures in the power spectra of the classical pulsators in the sample, and set constraints on the radii and orbital period of potentially observable exoplanets around the bright K dwarf 36 Ophiuchi, with typical lower limits of 0.8 RE radius and 40 day period. 1. INTRODUCTION as described by (Van Cleve et al. 2016). Yet, improved data The brightest stars on the ecliptic, visible to the naked processing pipelines have increased K2’s photometric pre- eye, have been observed by astronomers for millenia. Be- cision to almost the level of Kepler. Many pipelines have fore the K2 mission, they had never been observed for an been developed to correct the systematics present in the raw extended period with high-precision space-based photome- K2 photometry, including: K2SFF (Vanderburg & Johnson try. Revealing the inherent variability of stars is essential 2014), K2P2 (Lund et al. 2015), K2SC (Aigrain et al. 2016), not only for proper classification, but for understanding their K2VARCAT (Armstrong et al. 2015, 2016), K2PHOT (Eylen interior structures and evolution using tools such as astero- et al. 2016), and EVEREST (Luger et al. 2016; Luger et al. seismology (Handler 2012). By examining high-precision 2017). photometry from K2, which observed many stars of Kepler While these tools are available to alleviate the problem of magnitude 6 or brighter, it is possible to determine variability systematic pointing drift errors in the lightcurves, our sam- in these systems that was previously unseen due to the lim- ple has a much greater problem - being so bright, each of the itations of ground based observations. While there have of targets is saturated and hence suffers from significant bleed course been observations of bright ecliptic stars from space columns. In theory, the flux of the star could be recovered observatories in the past, never before have we had access to by analyzing the entire bleed column of the target, but K2’s continuous long-cadence data, collected every half hour for limited bandwidth and large distance from Earth makes it in- an entire campaign of approximately 80 days. feasible to download such a large amount of pixels. Instead, a When the original Kepler mission (Borucki et al. 2010) circular aperture around the saturated target can be requested, ended due to to the loss of two reaction wheels and by ex- ignoring the full length of the bleed column. This aper- tension the loss of the stable pointing that the spacecraft was ture has an economically feasible number of pixels, and the originally capable of, K2 was born. K2, Kepler’s extended pioneering technique of halo photometry from White et al. mission, uses the balancing pressure of sunlight of the satel- (2017) can be used to recover the flux and variability of the lite’s carefully aligned solar panels, as well as the two re- target even without access to all of the saturated pixels. Halo maining reaction wheels, to achieve pseudo-stable pointing photometry uses the flux from unsaturated pixels in the cir- (Howell et al. 2014). Still, K2 experiences systematic point- cular aperture surrounding the target to recover the total flux ing errors due to the thruster fires used to orient the space- by weighting these pixels to minimize the total variation in craft’s roll axis. This mechanism gives typical roll motion flux between successive measurements for the entire time se- in the focal plane of 1.0 pixels peak-to-peak over 6 hours at ries. It then corrects the reconstructed data with the K2SC the edges of the field, two orders of magnitude greater than pipeline. More detail can be found in White et al.(2017). In typical 6 hour pointing errors in the Kepler primary mission this work, we use the photometry corrected with this method to determine the power spectral density for 75 bright stars 2 GREKLEK-MCKEON &HUBER observed over K2 campaigns 11, 12, 13, and 14. From the the entire campaign, showing a constant increase in bright- power spectral density, we use iterative sinusoid fitting to ex- ness unlikely to be caused by true stellar variability. The tract the most dominant signals of variation from each star, majority of the lighcurves also exhibited significant non- and compare these variable modes to ones in the literature physical trends in flux at the start of each campaign, either that have been previously identified. We also search for os- displaying a steep rise or a steep drop in flux before leveling cillation signatures in the power spectra of the red giants in off for the remaining time in the campaign (see figure 2a & the sample. 2b). These features often took place on a scale of days and Our sample contains both "classical pulsators", objects that displayed up to a 40% change in relative flux, and are caused show a high amount of power at a few dominant frequen- by thermal effects as the CCD stabilized at the start of each cies and very low power elsewhere, as well as red giants, campaign. Data exhibiting clear systematic signatures such which are "solar-like oscillators". These red giants have as these even after running through the correction pipelines many frequencies over which the power is spread. Instead were simply trimmed away. of one dominant mode with power high above the noise, they Next, a convolution was found for each system and divided have a frequency where the power is highest among the en- out to remove any remaining long term trends unlikely to be velope around the nearby spread, commonly referred to as true stellar variability (these can be seen in figure 2 frames c νmax. We identify many previously unknown variabilities in & d). This was performed with astropy’s convolution model these bright stars, and detail several corrections to "known" (http://astropy.org/convolution), which uses a 2D Gaussian stellar pulsation periods. Furthermore, we analyze the rela- kernel run over the data at each point to produce the convo- tionship of low-frequency, 8hr granulation noise to the color lution function. For the Gaussian kernel, in an effort to pass and luminosity in the red giants of our sample, as well as a wide filter and retain the short-period stellar signals spe- in the prewhitened power spectra of the dwarfs. This rela- cific to each system, we varied between a 96 data point or tionship was initially established by Kallinger & Matthews 2-day standard deviation and a 240 data point or 5-day stan- (2010) and explored further by Bastien(2015), and Bastien dard deviation. The first half of the dataset was mirrored back et al.(2016). Finally, we examine the system of 36 Ophiuchi, from the start of the time series, and the second half mirrored the only K dwarf in this sample of bright stars, and inject ex- forward from the end of the time series, to remove the edge oplanet transits modeled with BATMAN (Kreidberg 2015) effects of the convolution function (visible in figure 2 frames to constrain the parameters of a detectable planet given the c & d as the blue points). After dividing out the convolu- star’s inherent variability. tion function to remove the long term variations, we began analyzing the stars in frequency space. 2. METHODOLOGY The stars in this sample include 11 targets from cam- paign 11, 8 from campaign 12, 35 from campaign 13, and 21 from campaign 14. Their visual magnitudes range from 1.62 to -0.62. In order to understand the distribution of the sample, we used each target’s EPIC ID to retrieve its co- ordinates from MAST (https://archive.stsci.edu/), and then match this to a known object in the SIMBAD database (https://archive.stsci.edu/), from which we collected parallax information, and V and B magnitudes. We used these quan- tities to calculate the distance to each system, the absolute visual magnitude, and the color, to create a color-magnitude diagram of our sample, shown in figure 1, and described fur- ther in the results section. 2.1. Light Curve Detrending For each star in this sample, we used data from the halo photometry routine described in the introduction as our cor- rected beginning flux time series. Data from campaigns 11 and 12 was split into two distinct observing periods, which we joined together. Then, each lightcurve was examined in- dividually, and edited to remove systematic effects such as Figure 1. Color-Magnitude Diagram of the stars observed in this clear steep drop-offs followed by near instantaneous correc- study, with the Sun shown in purple. tion due to thruster fires, or an overall ramping effect through VARIABILITY OF BRIGHT ECLIPTIC STARS 3 2.2. Frequency Analysis is the result of individual convection cells reaching a stel- After detrending each star’s time series, a Lomb-Scargle lar surface in close proximity. As the hot plasma rises in a periodogram was calculated from the corrected flux (cor- convection cell and reaches the surface, it cools and begins rected flux visible in figure 2 frame e, periodogram in frame to descend back into the stellar interior, streaming along the f).