«STUDENT RESEARCH PAPERS SUMMER 2011 VOLUME 22 REU DIRECTOR UMESH GARG, PH.D. REU Student Research Papers – Summer 2011 University of Notre Dame – ...»
If it was clear that the maximum corresponded to an Once we determined that there was no signiﬁcant eclipse and not simply noise, we then used IDL’s cur- change in the period, we used this information to sor program to hand-select the maximum and store better identify more eclipses. We started by plotting the time and magnitude into arrays. We used dates the area around the ﬁrst period and used the cursor where we had several subsequent eclipses to derive an program to record each eclipse by moving through initial guess by averaging out the distance between the plot in increments of our original period estithem. We further reﬁned this estimate by select- mate.This allowed us to increase from our original ing one well-surveyed eclipse and deﬁned that as the sixty-seven eclipse data point to one-hundred and zeroth eclipse. Every other eclipse’s date was then forty We then used these new data point to make a ﬁsubtracted from this and then divided by the initial nal adjustment on the period using the same method guess of the period and rounded to the nearest in- described above. Once we had the period deﬁned, we teger. Then gave the epoch, i.e. how many eclipses next decided to determine the average length of the away from our zeroth eclipse each eclipse was. By eclipse.
2.2 Pre-eclipse magnitude change In many of the orbits the star experience decrease the VATT telescope in Arizona. Once this was done, in luminosity before the primary (For an example see we determined what eﬀect this decrease had on the Fig.6). We ﬁrst recorded the date and magnitude of light-curve of the system, particularly it’s eﬀect on this decrease using the same methods as we did for the width of the eclipse. After analyzing the data the eclipse. We did this both for the AAVSO data and identifying the cause of any outliers, we then as well as data obtained during two separate runs of plotted the results and looked for trends.
To Table of Contents
2.3 Spectral analysis After the analysis of the AAVSO data, we per- we used IDL to calculate the total ﬂux in the U, B, formed spectral analysis of the star. Using data ob- V, and R bands (binned according to the Johnsontained from KPNO, we analyzed the emission spec- Morgan / Cousins system). Next, we then used the tra over two separate nights using the splot feature equation: −2.5 ∗ LOG(f lux) − 12 to turn the ﬂux of IRAF. In particular, we were interested in the into something resembling magnitude, with any elequivalent width (a measurement of the integral ﬂux ements with ﬂux less than 0 being set to 999. Ficompared to background) of the Hα, Hβ, and He II nally, we looked at the color changes of the hot-spot lines. After saving the equivalent widths into arrays,by comparing the U-B, B-V, B-R, and V-R magniwe used “hselect” to obtain the time of observation tudes. Noticing a possible rise in the B-R data just plotted their changes over time. before eclipse, we took the parts of the B-R magnitude and created a best-ﬁt line, which we then subAfter reading the ﬂuxes and wavelengths into ar- tracted from the data in order to better highlight rays (See Fig.1 for an example of a spectrum in IDL), this trend.
Figure 1: An example of a spectra showing H alpha (6562), H Beta(4860), and He II (4684) lines 3 Results / Analysis
3.1 Period and Magnitude changes over time Our initial investigation of the period, using (See Fig.2). Using a linear ﬁt on our ﬁrst set of
eclipse estimates obtained visually using the cursor eclipses gave us an estimate for the heliocentric Jucommand, revealed no signiﬁcant change in period lian day(HJD) of eclipse:
To Table of Contents
3.2 Pre-eclipse Dimming As mentioned earlier in this paper, the system We next decided to see if there was any correlation experiences a signiﬁcant change in magnitude unre- between the magnitude of this dip and the width of lated to the eclipse. However, this pre-eclipse dim- the eclipse. First, we determined an average width of ming does not appear in every eclipse in the data the eclipse of 0.00401 days(5.776 minutes). Then, we set, including some of the VATT data. We surmise selected certain eclipses with well deﬁned lightcurves that this is due to the use of diﬀerent ﬁlters for dif- and plotted the eclipse width vs. the magnitude of ferent observations as the dip appears to be heavily the dip (See Fig.4). From this graph, we determined wavelength dependent. In particular, this dip is very that there was no signiﬁcant correlation between the strong in blue wavelengths while it is unnoticeable in magnitude of the dip and the width of the eclipse.
red (See Fig.6). We hypothesize that this dimming Additionally, we plotted the diﬀerence between the is caused by the accretion stream. We believe that timing of the dip and the rise in magnitude out of the dip occurs when the hot-spot is blocked from the eclipse and determined there was no signiﬁcant the view of earth by the stream, thus diminishing correlation (See Fig.5) the total emission.
3.3 Emission Spectra and Color Variation 3.3.1 Emission / Absorption of Hα, Hβ and He II Using spectroscopic data from KPNO, we ana- the dip and absorption happened around the same lyzed the emission spectrum of the hot-spot using time while ours occur at diﬀerent points. We hythe equivalent widths of Hα, Hβ and He II. Fig.8 dis- pothesize that the accretion stream is causing this plays the data over two separate nights (the second eﬀect by blocking the emission of the hot-spot, and of the three nights was too cloudy to observe) while the fact that this star has the dip and absorption at Fig.9 shows the same data over-plotted and phased. diﬀerent times is due to the hot-spot being on the There are several diﬀerences of note between the two other side of the dwarf than the accretion stream’s nights. original direction unlike the star in the other study (See Fig.7).
The equivalent width of all the lines appears to rapidly increase after eclipse and continue to get stronger until rapidly dropping before the next Of particular note, several of the eclipses (espeeclipse. The third night in contrast has roughly the cially on the third night of observation) show a massame equivalent width throughout the time between sive increase in the equivalent widths of all lines.
eclipses. Additionally, the data on the third night While the large equivalent width can be partially atshows a leveling oﬀ and then extremely rapid de- tributed to the low total ﬂux, the peaks are clearly crease right before full eclipse that is not present on discernible. Additionally, the repeated existence of the ﬁrst night. Furthermore, both nights have times this peak in all three emissions makes it highly unwhere there exists both emission and absorption. likely to be caused by cosmic ray hits or ﬂuctuations.
The absorption tends to happen just after eclipse. We hypothesize that this increase in ﬂux is caused A previous study by Schmidt, G. et al. also noted by the accretion stream not fully being eclipsed and absorption and pre-eclipse dips, however in that case emerging around the star.
To Table of ContentsFigure 7: Representation of another polar by Schmidt et al.
3.3.2 Color Changes
In order to determine the color of the hot-spot, same data over-plotted in terms of the fraction of the we analyzed the diﬀerence in magnitude in several period. The third night data seems to show a redbands. The data that showed the least amount of dening of the hot-spot just before eclipse, however, noise was the B-band and the R-Band. Fig.10 show there is too little data and the data is too noisy in the B-R data for both nights while Fig.11 shows that order to make any deﬁnitive conclusions.
4 Conclusions We have produced a more accurate measure of the period of CSS 081231 over a time-span of several months. In addition, the small error on our period estimate puts bounds on variation on that period. In particular,we have shown that neither the timing nor the magnitude change of the dip aﬀects the period or base magnitude signiﬁcantly. As the dip is thought to be caused by the accretion stream, the evidence suggest that the amount of matter being accreted does not substantially eﬀect the orbit of the stars.
Additionally, the diﬀerence in timing of the dip and the absorption lines in contrast to other polars, as well as the sharp peak during eclipse indicates that the shape of the stream may play an important role in determining the emission from the star.
To Table of ContentsAbstract: Since early 2009, the Kepler spacecraft has continuously monitored over 150,000 stars in a 115-square degree field of view located in the constellations Cygnus and Lyra. Though the goal of the Kepler mission is to find extra-solar planets, another use of the Kepler data includes studying the behavior of normal and variable stars in the field. Every month Kepler downloads a fullfield image consisting of 84 charge-coupled device channels. In this research, images acquired at the beginning of the mission have been subtracted with later images to search for transient astronomical events, such as novae and supernovae, and variable stars not yet identified from this field. Approximately 1000 variables have been found, with a number of these being possible transients. These candidate variables and transients are compared with known variable star catalogs to determine which are newly discovered objects. A fraction of these new objects will be selected by Kepler for continuous study in Cycle 3.
Keywords: Kepler, transient, variable, supernova,
Launched in March 2009 to estimate the number of Earth-sized planets within and near the habitable zone of solar-type stars, National Aeronautics and Space Administration’s (NASA’s) Kepler Mission tracks over 150,000 stellar objects continuously in a field of view within the constellations Cygnus and Lyra. NASA finds these planets by searching for transits, astronomical events that take place when one celestial body, such as an Earth-sized planet, crosses in front of - or what appears to be in front of - another celestial body, or in this case, a star. Currently, those working on the Kepler Mission have identified 997 stars containing a sum of 1235 planetary candidates (Borucki, W. J., et al. 2011).
The array of CCD imagers has a wide field of view - 115 square degrees - specifically so that Kepler keeps a constant eye on an area with a high stellar density of dwarf stars. The photometer
in Borucki, W. J., et al. (2010) and Borucki, W. J., for the Kepler Mission (2010). There are 42 charge-coupled devices (CCDs) that are 1024 rows by 2200 columns. With two amplifiers per CCD, Kepler runs 84 channels; furthermore, the CCDs are mounted as pairs, meaning there are 21 modules. This can be seen in Fig. 1.
Figure 1: This image 1 displays the 42 CCDs paired into 21 modules in an image of the sky.
The data collected by Kepler is recorded in segments of quarter-years, as it is necessary for the spacecraft to rotate consistently and sequentially 90 degrees in order to illuminate the solar arrays and keep power running on the spacecraft. Due to the quarterly rotation of the spacecraft, the images of the stars are recorded on a different set of detectors (Borucki, W. J., et al.
The original image can be seen at http://kepler.nasa.gov/multimedia/Images/photogallery/?ImageID=12.
Software Bisque is credited with the information given on the webpage.
it will move to channel 53, 81, 29, and then back to channel 1 for Quarters 3, 4, 5, and 6, respectively. Also note that during Quarter 4, two CCDs - or four channels - were damaged in module 3: channels 5-8. These channels were not used in our data.
A second application of the Kepler Mission data is to detect and observe variable stars and transient astronomical events, or transients. Common transients include novae and supernovae, the latter being an important and desired find. We have an active program with Kepler to look at ~100 galaxies in the Kepler field of view in order to search for supernovae.
The purpose of this paper is to describe the process and results of searching through the Kepler Mission data to find possible variable stars, transients, and ultimately, supernovae. If a supernova is discovered in the data, we can access the images that recorded the beginning of the explosion. In certain kinds of type Ia supernovae, it is expected that a shock emission can occur less than a day after the explosion (Hayden, B. T., et al. 2010). This shock emission is due to some of the ejecta of the supernova smashing into the companion star. Simulations of this shock have been made and analyzed, though very few actual shocks have been observed. By looking through the Kepler data, we have the possibility of discovering a supernova with this characteristic in order to better study and understand the event.
Utilizing the link to.fits full frame images (FFIs) provided by NASA’s Kepler Data Analysis webpage, we accessed the monthly image downloads from Quarters 1 through 9. Due to the
oriented quarters; in this manner, we ensure that as few stars as possible leave the frame or change CCDs. This process makes it manageable to align and combine or subtract images for comparison. We selected the orientation that Quarters 1, 5, and 9 share, as it had the most available combined data.
Eight images for the Kepler Quarter 1 observation were recorded during the late part of April