Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 337, 207-215 (1998)

Previous Section Next Section Title Page Table of Contents

2. Observations and reductions

2.1. Photometry

Our photometric data were obtained on five epochs between September 1996 and February 1997. The log of the photometry is given in Table 1. The data were bias subtracted and flat fielded using tasks in IRAF 1. The transformations to standard magnitudes were obtained from observations of four Landolt (1992) standard fields and one E-field of Graham (1982) on the night of January 13, 1997. We solved for extinction and color corrections using the IRAF task PHOTCAL. Several local comparison stars were calibrated in the field. The supernova magnitudes of all other nights were measured relative to these stars. The comparison stars are marked in Fig. 1 and their magnitudes are given in Table 2.


Table 1. Supernova 1996N - log of photometry observations.


Table 2. Magnitudes of comparison stars.
Magnitudes for the comparison stars marked in Fig. 1. The uncertainties are about 0.05 magnitudes.

Obtaining magnitudes of a fading point source on a complex background is not a simple task. Normal aperture photometry is often not adequate (Turatto et al. 1993). We tried two different approaches: Point spread function (PSF) fitting and galaxy template subtraction.

For the PSF fitting we used DoPhot (Schechter et al. 1993). This software package obtains magnitudes by PSF fitting, using an empirical PSF obtained from one of the stars in the field. A problem with this method is the variability of the PSF across the field, seen in several of our images. This variability is probably due to the rather complex optics of the EFOSC-type instruments (Magain et al. 1992) and means that systematic errors can be introduced by forcing the fit of the empirical PSF star onto the supernova. Subtractions of the fitted PSFs do show rather large residuals.

To minimize these errors each magnitude was obtained using two different standard stars, number 1 and 2 in Fig. 1, for the empirical PSF. These stars are well exposed, relatively isolated and located at opposite sides of the supernova. The differences for the measured supernova magnitudes are actually rather small, with a standard deviation of less than 0.05 magnitudes in R.

To check the performance of DoPhot on the complex background we also extracted the magnitudes of a star close to the supernova (offset from SN 1996N by [FORMULA] W, [FORMULA] N), and obtained 19.21[FORMULA]0.04 in R. The error is the standard deviation of the magnitudes for the five epochs and gives a handle on the consistency of our measurements and a hint to the errors of the supernova magnitudes, at least for the earlier epochs.

As an independent check on some of the systematic errors we also tried the galaxy template subtraction method. On October 23.3, 1997, we obtained deep images of NGC 1398 in V, R and I with the ESO/NTT. Inspection of these images showed that the supernova had faded from visibility, the 3[FORMULA] upper limit being 23.0 magnitudes in R.

For each of our previous frames the NTT images were aligned and rescaled using standard IRAF tasks. Thereafter the sky was subtracted from the images and the NTT image was scaled to match the flux of the other image as measured in the comparison stars.

Before subtracting the images we had to assess the problem of different PSFs. We did this in two different ways both starting with the construction of a PSF with the IRAF/DAOPHOT task PSF, using three nearby stars.

One way to proceed is to deconvolve the PSF having larger FWHM with the other PSF using the Lucy-Richardson deconvolution task LUCY. Thereafter the image with the smaller PSF was convolved with the kernel from the above deconvolution. An alternative approach is to deconvolve both images with their respective PSF using a relatively large number of iterations. Thereafter the deconvolved images are smoothed with a Gaussian, resulting in images were all stars are nice Gaussians with preserved flux. The frames were then subtracted and the subtracted frame was measured with aperture photometry. The template method is, however, also likely to introduce errors. For example, we have used four different telescope systems with different color response. Also, the methods used for smoothing the images before subtracting are not optimal, in the first method residuals may remain due to differences in the variable PSF, whereas in the second, flux conservation of a point source on a diffuse background is not guaranteed. The success of these techniques can only be judged by careful inspection of the subtracted frames.

We believe that by using both of these very different techniques, PSF fitting and template subtraction, we have a good chance of avoiding any systematic errors that might be inherent in these methods. We did not see any systematic differences between the two methods and the measured differences per epoch were typically less than 0.1 magnitudes. The magnitudes we present in Table 3 are consistent with both of the above mentioned methods except for the latest epoch, where only the template subtraction method was used. The supernova was clearly visible in these images, but the signal to noise was too low for reliable PSF fitting.


Table 3. Supernova Magnitudes.

2.2. Spectroscopy

The log of our spectroscopic observations is shown in Table 4. With the exception of the January 1997 observation each spectrum is the combination of observations with two grisms to cover the blue and the red wavelength range. The slit was always oriented East-West. Since the supernova culminated close to zenith this was always close to the parallactic angle. Spectral reductions included bias subtraction and flat fielding, wavelength calibration with observations of arc lamps obtained immediately before or after the supernova observation, and flux calibration using standard stars (Hamuy et al. 1992). The typical accuracy of the wavelength calibration is between 0.5 and 1.0 Å, depending on the resolution provided by the grism. Some of the observations were obtained under non-photometric conditions, also slit losses might contribute to flux uncertainties. To establish the absolute fluxes for the spectra we convolved them with our simultaneous R photometry.


Table 4. SN 1996N - log of spectroscopic observations.
a Epochs in days past discovery, for the corresponding UT Date, see Table 1.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: August 6, 1998