Searching for Open Cluster Variable Stars Using Short Duration Unfiltered Images

1
Searching for Open Cluster Variable Stars Using
Short Duration Unfiltered Images

• J.A. Wilkerson
• T.S. Brown
• B.K. Lacoul
• Luther College
• AAVSO 94th Annual Meeting
• October 15, 2005

2
Student Participation
Zebadiah Howes Buena Vista Univ. Travis
DeJong Dordt College Forrest Bishop Decorah High
School
Todd Brown Brajesh Lacoul Kari Frank Alex Nugent
Robyn Siedschlag Siri Thompson Matt
Fitzgerald Heather Lehmann Amalia Anderson Hilary
Teslow
Support Roy J. Carver Charitable Trust (Grant
00-50) Luther College R.J.
McElroy Trust/Iowa College Foundation
3
All images acquired with a 12 Meade LX200 and
Apogee AP6E camera
4
OUR DATA SETS
Cluster Dur. (s) n images Date
range of images NGC 129 11.0 9
14,470 11 Aug. 2003 ? 8 Sep. 2003 NGC
2682 (M67) 2.0 14 34,769
25 Feb. 2004 ? 26 Apr. 2004 NGC 6514 (M23)
3.5 25 45,668 19 June 2003 ?
8 Sep. 2003 NGC 6531 (M21) 3.5 21
29,987 26 June 2002 ? 8 Sep. 2002 NGC
6694 (M26) 9.0 20 27,940
24 June 2004 ? 9 Sep. 2004 NGC 6514 (M23)
2.5 29 60,000 23 June 2003 ?
31 Aug. 2005
All images are unfiltered
5
DATA PROCESSING

• CALIBRATION
• Dark Noise Correction
• Flat Fielding
• ALIGNMENT
• Use a single frame for entire data set
• STAR ID FLUX MEASUREMENT
• Aperture photometry for flux determination
• FRAME NORMALIZATION

6
Frame Normalization

• Identify four reference images from throughout
  the night
• Calculate average flux for each star in all four
  frames this is the reference flux
• Determine the flux of each star in the frame to
  be normalized this is the sample flux
• Calculate (reference flux/sample flux) for each
  star
• Normalization factor median of all ratios in (4)

7
Use Web Version of the BDA catalog for magnitudes
(Mermilliod and Paunzen , http//www.univie.ac.
at/webda//)
8
Variable Stars of Particular Interest

• Short Period (lt 1 day) Variables (e.g., b
  Cephei, d Scuti and W UMa)
• Longer Period Eclipsing Binaries
• Cataclysmic Variables
• Others have recently been surveying variables in
  open clusters. (e.g., Kafka and Honeycutt, AJ,
  126, 2003 Kafka et al, AJ 127, 2004 Zhang et
  al., MNRAS, 355, 2004 Paunzen et al., AA, 418,
  2004)
• Our clusters have the interesting feature of
  age-blind selection.

9
Sample Light Curves from Our Data
10
The Trouble is

• Unfiltered images result in large residuals
• A single normalization factor cant normalize
  every star in the field
• Red stars are overcorrected blue stars are
  undercorrected

11
A Way to Search for Variables
Work in flux/normalization space Flux(t)
CL(t)f(nf(t)) Flux(nf) CL(t(nf))f(nf)
12
Each normalization factor usually occurs at two
different times each night
The result is hooks in graphs of flux against
normalization factor
13
(No Transcript)
14
Using the hook to find variables

• Fit a second-order polynomial to flux as a
  function of normalization factor
• Define the WSVI statistic to measure the
  deviation of a stars flux from the polynomial
  fit using paired observations
• Find the mean WSVI for a subset of stars
• Measure each stars WSVI deviation from the mean
  of its subset

Based on a variability index developed by Welch
and Stetson (AJ, 105, 1993)
15
fjsi is the measured flux in image jsi.
f(j1)si is the measured flux in image
(j1)si. f,exp is the expected flux derived
from the polynomial fit. There are N total
images. The number of flux pairs used in
calculating WSVI is n and s is the spacing
between the images constituting each flux pair.
The standard error is calculated for each
successive set of twenty images from the scatter
in the flux values for those twenty images. Thus,
one standard error is used for images 1 to 20,
another for images 21 to 40, etc.
16
(No Transcript)
17
(No Transcript)
18
Three stars are known with periods between 0.3
and 0.5 days in our M67 field (Xin et al., Ch. J.
AA, 2, 2002 Sandquist and Shetrone, AJ, 125,
2002 van den Berg, AA, 382, 2002)
We looked at 3 nights of M67 data the three
known short period variables had WSVI deviation
ranks of 1, 2 and 8 for about 500 stars in field.
19
What we are doing now
1. Looking for phase and period consistency in
light curves
2. Developing WSVI technique for longer period
variations
3. Trying template fitting for light curve
reconstruction
20
Conclusions

• We have many previously unknown variable stars
  in each of our fields
• Our WSVI technique finds variable stars
• Further work needs to be done to understand
  the techniques robustness against false alarms
• Further work must be done to understand our
  ability to re-construct light curves from our
  data