Astron. Astrophys. 319, 617-629 (1997)

2. Observations and reduction

2.1. The instrumentation

Comet P/ST was observed with the 2-m-RCC telescope of the Bulgarian National Astronomical Observatory, Rozhen (BNAO). The focal reducer (FR) of the Max-Planck-Institut for Aeronomy (Jockers,1992) was used with interference filters centered at 426 nm, 620nm, and 642 nm for the 0-8-0 transition of H₂ O , the (2-0) A transition of CO , and for the continuum, respectively. The FR converts the initial focal ratio of f/8 into f/2.8. The interference filters are placed in the parallel beam behind the collimator. Their transmission curves are plotted in Fig. 1, together with portions of comet P/ST spectra, obtained on Nov 17, 1992 by S. Wagner at the 3.5 m telescope of the German-Spanish Astronomical Centre, Calar Alto (M. Küppers, private communication). The dotted line represents a spectrum in the coma of the comet and the dashed one a spectrum about 10⁵ km tailward of the nucleus. The CO images are contaminated by a C₃ emission which is confined close to the nucleus. The possible contamination by CH has disappeared in the tailward spectrum because of the relatively short scale length of this ion. More critical is the contamination of the H₂ O image by CO emission. If not accounted for, it will cause an overestimate of the H₂ O content. In addition, close to the nucleus the H₂ O image might be contaminated by C₂ (a blend with CO at 619 nm). The most extended contaminant, C₂, has a scale length of 6.6 km (A'Hearn et al. 1995) . We therefore will avoid a region of this extent around the nucleus in the discussion of our results.

Fig. 1. Transmission curves (solid lines) of the CO , H2 O , and the continuum filter are shown in the upper, middle, and lower panel. The dotted lines are from a spectrum of comet P/Swift-Tuttle's coma and the dashed line from a tail spectrum.

A camera with an EEV P86000/T CCD chip was employed to record the data. The detector comprises 576 385 pixels with size 22 22 . The angular size of the imaged field is 7.0 4.6 arcmin with a resolution of 0.8 arcsec per pixel.

2.2. The observations

The observations were carried out in the night of November 25, 1992. A list of the obtained images is given in Table 1. The heliocentric distance of the comet was 1.0 AU, the geocentric distance was 1.3 AU. One pixel equals 756 km at the comet. The deprojected scale along the radius vector of the comet is 1002 km. The phase angle (Earth - comet - Sun) was .

Table 1. List of observations of comet P/Swift-Tuttle, Nov 25 1992

The spectrophotometric standard star Aql, not far from the location of the comet, was observed for the absolute calibration of the images immediately after the comet frames. Its flux was taken from Voloshina et al. (1982). Intermittent clouds prevented to obtain a sequence of standard stars at different zenith distances.

2.3. Data reduction

2.3.1. Bias, flat fields, and sky background

The raw images were bias subtracted and divided by flat fields, obtained during twilight through the same filters in fall of 1992. As the frame was filled almost entirely by the comet, the process of sky background estimation was somewhat complex. We selected boxes with different sizes in the lower right corner of the images (see Fig. 3). As expected, the histogram analysis of these boxes shows an increasing asymmetry (growing right wing) with increasing boxsize. A rather small box did not contain enough pixels for a statistically significant histogram. A good compromise was found for boxes containing approximately 8000 pixels and having histograms almost symmetrical relative to their maxima. The results of this procedure are summarized in Table 2. The last two columns contain numbers obtained by calculating mean values and standard deviations in several smaller boxes between the star trails in the lower right corner of the images. The numbers in column 2 were adopted as sky background count.

Table 2. Sky background estimation

2.3.2. Extinction correction

For the CO images the extinction coefficient was extracted from the science frames themselves. A value of was obtained by using the total signal in boxes of to pixel around the nucleus. We consider the obtained value of 0.35 as reliable for several reasons. The first one is the relative large difference between the airmasses at which the three blue images were obtained. Second is the fact that near to the nucleus the signal is dominated by continuum emission (see Fig. 2). The independence on box size precludes temporal flux changes of the comet. The third reason is that this value is equal to the mean extinction coefficient for BNAO at this wavelength. For the red spectral region a value of 0.15 was derived from the known ratio of the red to blue extinction coefficients for NAO Rozhen. In order to estimate the photometric error we have considered all standard stars observed with our instrument in a red filter and at nm (in total four stars) in October and November 1992 which were observed at similar or better sky conditions and allowed the determination of extinction. The airmasses were 1.01, 1.25, 2.23 and 2.36. The instrument response at nm has a standard deviation of % and % at 426 nm. The error in the flux ratio of both wavelength was %. A relative error of % for the fluxes and % for the flux ratio seems to be representative for our work.

Fig. 2. Upper panel: Signal contributions of H2 O and continuum in the on-line image at 620nm. Lower panel: The same for the CO emission in the on-line image at 426nm. The off-line image is multiplied with the corresponding continuum scaling factors. Cuts are shown parallel to the tail axis. They are averaged over distances from 1.5 to 3 km northward of nucleus.

2.3.3. Subtraction of dust continuum

In order to subtract the continuum from our plasma images we have to obtain a value for the continuum scaling factor, k, defined by the following relation:

If neutral color of the cometary dust is assumed, k must be equal to the response ratio of our instrumentation against solar radiation at the wavelengths of the off-line and on-line image. As the standard star used is of A-type we transform to solar fluxes by using the relation:

where S and F are signals and fluxes, respectively, indicates the corresponding on-line wavelength, and subscript `A ' stays for the values of the observed A-type star. The fluxes for the sun were taken from the Kurucz atlas (1985) and convolved with the transmission curves of our filters. The numerical values of the different terms appearing in Eq.  2 are presented in Table 3. The k - values in column "neutral" were obtained from the response ratio, , by performing atmospheric extinction correction for each particular image. The application of "neutral" k -values in Eq.  1 left some residual continuum contribution in the plasma frames. Close to the nucleus the difference images still showed a peak, typical for the continuum. There is theoretical and observational evidence (Bonev and Jockers, 1994) that the plasma distributions are rather flat around the nucleus, whereas the continuum is strongly peaked. Furthermore, the different spatial dependence of the dust tail as compared to the H₂ O and CO tails in November 1992 sets another restriction to the determination of the continuum scaling factors. The dust tail has a rather strong curvature toward north (see Fig. 3). Therefore, the choice of the scale factors is limited by the condition that in the upper left corner of the plasma images the ion contribution should be small but must be positive. Another restriction comes from the condition to cancel out the continuum far from the nucleus on its sunward side. Fig. 2 shows profiles parallel to the tail axis averaged over distances from 1.5 to 3 km northward of nucleus. The signal of the raw on-line image, the intensity scaled off-line image, and their difference, the pure emission, are shown. Evidently, on the sunward side the continuum contributes 90% and more to the total signal in the on-line images. All considered conditions were used as a complex empirical criterion to derive improved values for the continuum scaling factor, k. It turned out to be impossible to simultaneously cancel the overall dust tail and a jet structure close to the nucleus (see Fig. 3). Therefore, with the empirical method two continuum scaling factors were obtained, one for the central peak and one for the dust jet. The values are presented in Table 3 in columns "mean" and "jet", respectively. The continuum contribution in the plasma images was removed by using Eq. (1) with the mean continuum scaling factor (Table 3). The applied criterion allowed the determination of the scaling factors in relativelly narrow limits, about 5% for the CO , and 2% for the H₂ O image.

Table 3. Numerical values of the terms in Eq.  2 and the obtained scaling factors for continuum subtraction. See text and Eqs.  1 and 2 for more explanations.

Fig. 3. The continuum image obtained at 642 nm, converted to intensity. The grey levels represent log10(). The overplotted contours differ by 1 magnitude and the outermost one denotes an intensity of , equivalent to 19.86 magnitudes arcsec-2. The spatial scale is in units of 1000 km at the comet, perpendicular to the line of sight.

In order to increase the signal to noise ratio we removed the star trails from the raw images before performing the continuum subtraction. As the trails are vertical in our images the stars were removed by interactive horizontal interpolation.

2.3.4. Reduction to ion column densities and solar continuum

After continuum subtraction the ion images were calibrated for emission line intensities and then transformed to column densities. Taking into account that we have observed only the red portion of the 0-8-0 transition of H₂ O we have taken half of the resonance fluorescence efficiency factor (g-factor) given by Lutz (1987), i.e. 2 10^-3 photons s^-1 ion^-1. The A (2-0) emission of CO was transformed to column densities by using the g-factor provided by Magnani and A'Hearn (1986). During the observations the heliocentric radial velocity of P/ST was -8.4 km s^-1. Most probably further in the tail the ions are accelerated up to velocities, greater than the heliocentric radial velocity of the comet, making thus the influence of the Greenstein effect stronger than that of the Swings effect. Therefore we have taken the g-factor averaged over the values given by Magnani and A'Hearn (1986) for the velocity range 10 - 20 km s^-1, 3.7 photons s^-1 ion^-1.

Our H₂ O -filter is contaminated by the CO subband of the A (0-2) transition (see Fig. 1). Since we know the CO distribution we can, using the data given by Magnani and A'Hearn (1986), calculate the strength of this contamination and subtract it out. Magnani and A'Hearn do not explicitely tabulate the g-factor for the subband of the (0-2) transition but within the velocity range 0-50 km s^-1 it can be calculated from Tables 1 and 2 of this paper to photons s^-1 ion^-1. This leads to a CO contamination in the H₂ O filter of 17%. The H₂ O image (Fig. 5) was left uncorrected but in the more quantitative plots of Figs. 6 and 10 this correction was applied.

The continuum image was calibrated in terms of mean solar disk intensities, (Jockers et al., 1993). In these units, the continuum intensity, i, is related to the product of geometric albedo, p (Karttunen et al. 1987), phase function, , and filling factor, f, of the dust particles via the equation:

where r is the heliocentric distance of the comet and is the solar radius.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de