Observations
November 1995
Work at NOAO began with an observing run of 5 nights in November,
1995, on the Coude Feed Telescope. Spectra from 390-520 nm, including
the H-alpha, H-beta, and H-gamma absorption lines, were obtained for
the magnitude 3.7, F6IV star Alpha Triangulum at a dispersion of
0.45A/pixel. The spectra also covered the Mgb lines at 518 nm and the
Ca II K line. While the nights were spectroscopic rather than
photometric, we were able to obtain several hours of continuous
observations each night. The Coude Feed Telescope has an aperture of
only 0.9 meters, too small to obtain the necessary signal to assure
the detection of acoustic oscillations. The data are useful, however,
for developing data reduction and analysis procedures. The
observations were obtained at a cadence of about 68 seconds, with an
integration time of 45 seconds.
Analysis of the timing data suggested that improvements were needed to
assure that the cadence remained constant, and to maximize the
observing efficiency. We estimate that the maximum per-pixel
signal-to-noise ratio we can expect in a 45 second exposure at this
magnitude is about 600/1.
Following this run, new ICE ccd control code was prepared by Rob
Seaman of the IRAF group to allow us to obtain spectra in uniform time
steps based on a heliocentric clock to eliminate variations in the
timing due to the motion of the Earth in its orbit.
Our data from the November run show a peak in the power spectrum at a
frequency of about 660 micro-Hertz (corresponding to periods near 25
minutes). This is the expected frequency for Alpha Triangulum, but
the signal is very weak and may be due simply to noise.
February 1996
In February, 1996, we had observing runs on the 2.1-m with the Goldcam
CCD spectrograph (Feb. 2-8 UT) and on the Coude Feed Telescope
(Feb. 2-6 UT), again with the coude spectrograph. We observed the
F5IV-V star Procyon with both telescopes. The same cadence was used
(20s) but the Coude Feed spectra were offset by 10s from the spectra
obtained with the 2.1-m. The first two nights of the runs were
cloudy, but we obtained approximately 4.5 nights worth of data with
the 2.1m telescope and 2.5 nights with the Coude Feed. Both data sets
have similar spectral resolution to the Alpha Triangulum data, about
half an Angstrom per pixel. Reductions of the spectra are underway.
March 1996
In March, 1996, we returned to the Coude Feed Telescope to attempt to
confirm the detection of oscillations in the G0IV star Eta Boo by
Kjeldsen et al. Approximately 3 of the 5 nights (March 8-12 UT) were
clear, and data reductions will be complete soon.
During the first part of the night on this run, we also attempted to
obtain spectra of Procyon and Sirius using a "ramp" readout mode of
the CCD. This mode allows us to improve the observing efficiency with
more or less continuous readout of the spectrum. These observations
are being reduced to determine if the signal-to-noise ration of the
spectra is adequate to permit precise measurements of the variation of
the hydrogen lines.
Sam Barden has also run Monte Carlo simulations
(click here for the figure) which show how many nights of data will
be needed to achieve a 90% probability of detection oscillations in
Procyon at the expected amplitude and frequency. With a 20 second
cadence, approximately 20 clear nights (160 hours) of observations are
needed. Reducing the cadence to 5 seconds would allow a 90%
confidence detection in only 4 nights.
May 1996
An important test of the equivalent width method consists in applying
it to integrated solar light. This has been done by Kjeldsen et
al. (1995), but they did not detect an obvious excess in the power
spectrum at 3 mHz at their noise level of 5.1 ppm. We have recently
repeated this experiment with a fast CCD camera that provides spectra
at very high signal-to-noise levels. On May 25, 1996, we obtained a
four-hour sequence of spectra at a sampling rate of 15 seconds with
the stellar spectrograph at the McMath-Pierce telescope on Kitt Peak
using the integrated light feed and one camera of the Zurich Imaging
Stokes Polarimeter (ZIMPOL) I. 128 spectra were averaged in real-time
at a rate of 10 spectra per second for every sampling period. After
averaging along the spectral lines, the signal-to-noise ratio per
0.027 nm wavelength bin was about 20,000. The power spectrum showed a
peak in the equivalent width of Hbeta at the expected frequency of 3.1
mHz. However, this result could not be confirmed in later
observations.
December 1996
6 continuous days of solar integrated light were observed with ZIMPOL
in down-town Tucson. There is a slight excess of power around 3 mHz
and some of the peaks correspond to the frequencies associated with
low-order p mode oscillations. However, this result is tentative and
needs to be confirmed with a more rigorous data analysis.
Last change: September 29, 1997
