3 The last phases with HST

3.1 STIS observations

In order to follow the supernova to very late times, and to get a better understanding of the supernova environment, we have obtained detailed imaging at four epochs using the Space Telescope Imaging Spectrograph (STIS) onboard the HST. A log of our HST observations is given in Table 4.

 

 

Table 4: Log of HST observations.

Date	MJD	Epoch	Exp. Time	Exp. Time
(UT)		(days)	CL (s)	LP (s)
11 June 00	1707	778	1240	1185
25 June 00	1721	792	5672	--
21 Nov. 00	1869	940	4000	1319
28 Aug. 01	2150	1221	5674	--

The observations on 2000 June 25, 2000 Nov. 21, and 2001 Aug. 28 were taken as part of the Supernova Intensive Study programme (P. I. Kirshner). The data from 2000 June 11 was taken as part of the Survey of Host Galaxies of Gamma-Ray Bursts (P. I. Holland, Holland et al. 2000). The total exposure times in the 50CCD (clear, CL) aperture were 1240, 5672, 4000, and 5674 s on the four epochs, respectively. The 50CCD aperture has a central wavelength of 5850 Å and a width of 4410 Å.

In addition to the CL frames we also obtained some images in the F20X50LP (long pass, LP) filter at two epochs. The total exposure times were 1185 and 1319 s, respectively. The F28X50LP aperture has a central wavelength of 7320 Å and a width of 2720 Å.

The 2000 June 11 images were taken using a four-point STIS-CCD-BOX dithering pattern with shifts of 2.5 pixels (= $0\farcs127$) between exposures. The other images were dithered using the same pattern with shifts of 5 pixels (= $0\farcs254$).

The data were preprocessed through the standard STIS pipeline and combined using the DITHER (v2.0) software (Fruchter & Hook 2002) as implemented in IRAF. We selected a final output scale of $0\farcs0254$/pixel, corresponding to exactly half the original size, and set the "pixfrac'' parameter to 0.6. Since there were only two images taken using the LP filter on 2000 Nov. 21 we had to increase "pixfrac'' to 1.0 for these data in order to fill all pixels in the output frame. This resulted in a reduction in the image quality of this data compared to the drizzled 2000 June 11 LP image. The six drizzled images were finally reregistered onto a common system.

3.2 Photometry

The supernova is located in a complex region of the host galaxy (Fig. 1). There are several objects within a radius of $1\arcsec$ from the supernova position, and SN 1998bw itself is located on top of an extended, filamentary structure. In order to perform photometry of the supernova without contamination from the underlying structure we again used the template subtraction method to subtract the 2001 Aug. 28 CL image from each of the other CL images. An example subtraction is given in Fig. 1.

We then constructed PSFs for each subtracted CL image using DAOPHOT II/ ALLSTAR (Stetson 1987; Stetson & Harris 1988) and fit these to the image of the supernova on each subtracted frame. We computed aperture corrections for each frame using several isolated stars. The PSF magnitudes were corrected to an aperture with a radius of $1\farcs108$, which Table 14.3 of the STIS Instrument Handbook suggests contains 100% of the flux from a point source. These corrections were about 0.1 mag at all epochs. The instrumental magnitudes were converted to the AB system using the zero points of Gardner et al. (2000).

In order to understand any systematic errors we performed a series of artificial star tests on each CL image. We added a set of stars with known magnitudes comparable to that of the supernova to locations on each frame where the underlying structure was similar to the underlying structure of SN 1998bw. We then subtracted the template image and performed photometry on these artificial stars in exactly the same way as for the supernova. The resulting median differences between the input and recovered magnitudes of the artificial stars were always below 0.2 mag. We regard this as a conservative error on the difference photometry.

No template image was taken using the LP filter. Therefore, no reliable magnitudes could be obtained in this band. For the 2000 June 11 image we estimated the supernova magnitude by pure PSF fitting on the drizzled image. The estimated magnitude is LP $_{{\rm PSF}}=25.02\pm0.07$. Note that there was an error in the LP aperture corrections used by Fynbo et al. (2000). That paper also contains some color information on the surrounding objects, although a correction of -0.51 mag should be added to all those LP magnitudes. The PSF magnitude for the supernova will clearly also include some of the light from the underlying background, which will result in overestimating the flux from the supernova. As mentioned above, the 2000 Nov. 21 LP frame is less useful. Visual inspection of that frame does show the supernova, just as in the CL frame from the same epoch, but we were not able to perform a reliable PSF estimate of its magnitude, since the remaining PSF residuals were always large. Therefore, only the CL magnitudes will be discussed below.