spacer link to MAST page spacer logo image spacer
link to STScI page

Photometric Reductions

The following section on photometric reductions was reprinted, with permission from the author, from the paper "The Copernicus Ultraviolet Spectral Atlas of Tau Scorpii" by John Rogerson and Walter Upson (Ap J Suppl. 35,37-110,1977). The corrections described for cosmic ray and charged particle background, guidance variations and scattered light however, are relevant to all Copernicus observations.

The variation of the instrument sensitivity with wavelength is not well known, and consequently no attempt has been made to normalize the observations to a constant sensitivity condition. This atlas should therefore not be used to estimate the general shape of the ultrtaviolet continuum of Tau Sco, but it is entirely suitable for investigation of the line spectrum.

The counts detected at each measurement point reflect the flux in the stellar spectrum but are also affected

  1. by a background due to cosmic rays and trapped ions from the lower Van Allen belt,
  2. by how well the guidance system keeps the star image centered on the spectrometer entrance slit, and
  3. by scattered light within the spectrometer.
Corrections for these three effects are considered below. The counts may also be affected by the strength and direction of the magnetic field in which the photomultiplers find themselves, but the behaivor of many repeated U2 counts at a fixed point in the spectrum suggests that this effect must be small. This is presumably also true for U1, which is physically identical and operated identically.

Cosmic Rays and Charged Particles

This source of background counts has been carefully studied by York and Miller (1974) using counts obtained when starlight was not entering the spectrometer. The cosmic ray flux depends mainly on the geomagnetic lattitude of the satellite, while the trapped particles are primarily concentrated in the South Atlantic Anomaly. In addition, some of the data photomultipliers have windows the fluoresce for some time following a heavy exposure to the charged particles, with the result that the background for these tubes at a particular time depends not only on the current particle flux but also on their recent exposure. Although the data for this atlas do not depend on these fluorescing tubes, the background counts for all tubes are estimated by the same procedure. Under these circumstances it is expedient to designate the position of the satellite by the longitude of the ascending node of the orbit and the time after pasage of the satellite through this node. Using this coordinate system ensures that each time the satellite reaches a certain coordinate pair, it will have had the same recent history of cosmic ray and trapped particle flux exposure. The study by York and Miller shows that the background counts are quite repeatable (within expected statistical flucuations) over time intervals that are long compared to the Tau Sco observing interval when using the above-described coordinate system. Aimed with this empirical result, a table was prepared giving the expected background count as a function of the satellite coordinates, and this table was then used in removing the background count from each measurement.

[Note: the Copernicus raw data files include background counts for most scans (see precautionary comments though). In cases where a background was for one reason or another not included, a program is available for deriving values as described above. Also, the table of background counts as a function of coordinates, is now available in FITS format. ]

Correcting for Guidance Variations

In spite of an excellant guidance system, changing external torques operating on the satellite cause the stellar image to drift about on the entrace slit of the spectrometer. The result is that the amount of stellar flux entering the spectrometer slowly changes, giving rise to variations in the observed spectrum which have no counterpart in the star's intrinsic spectrum. These variations may be as great as 10%. This effect can be corrected with the help of the signal from U2 which, as noted before, is held fixed so that its output signal provides information on the variation of stellar flux entering the spectrometer. Each U1 count is corrected for guidance drift by dividing it by the simultaneously acquired U2 count and multiplying it by the average of all U2 measurements made with U2 held fixed in position.

While the correction procedure is straightforward in theory, there are four operations that must be performed on the U2 counts before they can be used to correct for guidance variations. First, the cosmic ray and charged particle background must be removed. This is accomplished in the same manner as described above for U1. Second, while U2 is help fixed in the spectrometer, its wavelength is not quite fixed in the stars spectrum. This is caused by the changing Doppler shift mainly due to the orbital motion of the satellite. The componenet of the satellite velocity in the direction of the star is known for each measurement, and hence the instantaneous wavelength being observed by U2 can be calculated. In order to predict the U2 count variations due to the changing wavelength, the U1 spectrum in the neighborhood of the U2 position has been numerically degraded to the U2 spectral resolution (nominally 0.2 Å). This degraded U1 spectrum gives the local spectral variation of the U2 signal and allows the U2 counts to be corrected to what they would have been in the absense of a variable Doppler shift. The correction is generally quite small since the satellite valocity range of ± 7.5 km/sec is small compared to the U2 resolution of about 50 km/sec. Nevertheless, in a few monitoring positions the U2 slit was unintentionally fixed near the edge of a strong absorption feature, and the deduced corrections are not negligble. Third, the corrected U2 signal still contains a noise component which is mainly statistical. Since the guidance variations have a slow time scale (that of the satellite orbital period), we have smoothed the high frequency noise in the U2 signal, leaving intact the low-frequency guidance-induced variations. Fourth, a possible systematic error exists in the guidance correction procedure. The average U2 values for each monitoring position do not necessarily reflect the average quality of guidance; i.e., the total stellar flux passing the entrance slit, averaged over the time interval that U2 is in one monitoring position, may differ from that at another monitoring position. In order to fit together the various spectral segments, the averaged U2 signals must be corrected for variations in average guideance quality at each fixed position.

This correction was made by forming the ratio of U2 signals (corrected as above) near spacecraft midnight for consecutive U2 positions. This ratio ia assumed to be the correct intrinsic ratio since it is measured under approximately the same orbital configuration during a time (midnight) when guidance perturbations due to scattered light are at a minimum. This ratio was then used to correct the average U2 signals to the same quality of guidance.

The corrected and smoothed U2 signal finally was used to monitor and remove the guidance variations in the U1 signal. The separate segments then formed one continuous spectrum.

[Note: listing files are available which indicate the U1 and U2 scans for which separate monitoring scans were obtained.]

Scattered Light

Finally, the U1 counts must be corrected for scattered light within the spectrometer. This component of the observed counts may be estimated with the help of a number of interstellar absorption features which appear to be saturated. The residual signal in the saturated core is assumed to be due only to scattered light at the wavelength of the feature. The interstellar features that have been used for the scattered light analysis of the second-order spectrum are listed in Table 2.

While studing interstellar absorption at Lyman Alpha, Bohlin (1975) found the U1 scattered light to be linearly proportional to the strength of the local spectrum averaged over a 24 Å band centered on the wavelength of interest. Scattered light values, predicted according to Bohlin's method, were compared with the observed values given in Table 2 and while the dependence on the local average is quite evident, the detailed agreement was of inadequate quality for the purposes of this atlas. Attempts to modify Bohlins formulation to improve the detailed agreement over the whole second-order spectrum were only partially successful, though it was found that an 18 Å average is somewhat superior to the 24 Å average. See paper for rest of discussion on scattered light.