2. WFPC2 direct imaging of J03.13
Following the ground-based identification of two resolved point-like images for the quasar J03.13 (see Paper I), we proposed to image this interesting system with the WFPC2 planetary camera (PC1) through two wideband filters in order to search for possible structure of the QSO images at 0.1 angular scales and also to possibly set additional constraints on the lensing model. The good dynamic range of the WFPC2 was essential to achieve these goals. Based on the photometry reported for this system in Paper I, we chose the ADC channel with 14 e/ADU gain and integration times of 160 (resp. 400 and once 300) sec., to avoid the saturation of the brightest QSO image through the F555W (resp. F814W) broadband filters. Given the two orbits allocated to this HST direct imagery program (ID 5958), our strategy has been to obtain 5 (resp. 6) such PC1 exposures with the F555W (resp. F814W) filters. Each of these exposures has been offset by in order to optimally dither the QSO images and possibly detect faint structures at small angular scales. In the usual nomenclature for WFPC2 data, the dataset names obtained on November 28, 1995 (18h32m - 20h34m UT) with the F555W (resp. F814W) filters range from U2YK0101T to U2YK0105T (resp. from U2YK0106T to U2YK010BT). Please note that the integration time for dataset U2YK0107T is 300 sec., instead of 400 sec. for all remaining F814W CCD frames.
Composite F555W and F814W images of J03.13 made after proper recentering and coaddition of the single PC1 frames are shown on the left panels of Figure 1.
On these composite and on each single CCD frames, J03.13 appears to be clearly resolved as two point-like components. No trace of a third compact image or of a faint intervening galaxy is visible. The somewhat extended core and knotty-like structure seen for each single image component are essentially due to the complex shape of the combined "HST + WFPC2 + filters" point spread function (PSF).
Due to approximate equatorial coordinates used for J03.13, the target has not been properly centered on the PC1 detector; it fell very near to one of its edges. For this location, we were not able to identify an adequately observed PSF using the WFPC2 PSF library search tool available at STScI. We therefore decided to proceed as follows. A set of approximately 100 simulated PSFs has been computed by means of TinyTim (Krist 1997) for different values of the focus (Zernicke parameter Z4) and of the jitter of the telescope. These numerical PSFs have then been fitted to the images of J03.13 with an automatic procedure fully described in Ostensen et al. (1997). For each CCD frame, optimal values for the Z4 and jitter parameters have been selected on the basis of the minimum for the residuals. Note that due to the orbital breathing of the telescope, excursions of up to 7 microns have been found for the focus (Z4) during the F814W observations of J03.13. A PSF was subsequently constructed with a pixel replication factor of 10 using TinyTim. An iterative procedure has then been applied to this 10x10 oversampled TinyTim PSF to address the problem of fitting the undersampled HST PSF peaks after proper recentering, rebinning and application of an appropriate pixel response function (Kernel). A more detailed description of this method is to be found in Remy et al. 1997 (in preparation). PSFs were constructed for the two resolved image components. Finally, these PSFs were used with the PLUCY application program (Hook and Lucy 1994) to deconvolve each single CCD frame of J03.13. The combined results of those deconvolutions are presented for the F555W (resp. F814W) filters in the upper (resp. lower) right panels of Figure 1. From these results, we can rule out the presence of a third component fainter than A by up to 5.2 mag. at angular separations 0.13 . Given that TinyTim PSFs are not more accurate than a few tenths of a percent, the very faint residuals seen near J03.13 A and B on the deconvolved images turn out to be non significant. Using the PLUCY deconvolution algorithm, the magnitude difference between the two unresolved components is found to be 2.14 0.03 and 2.16 0.03 for the F555W and F814W filters, respectively. We also used the optimal TinyTim PSFs selected above and the automatic photometric fitting technique described in Ostensen et al. (1997) to decompose all images of J03.13 with two point-like components, their relative positions and brightnesses being the only free parameters. After subtraction of the best fitted double PSFs, no significant residuals are seen. Adopting a scale of 0.04553 /pixel for the PC1 detector, we derive an average angular separation of 0.849 0.001 between J03.13 A and B, and magnitude differences of 2.14 0.01 and 2.13 0.01 for the F555W and F814W observations, respectively. These values compare rather well with those inferred from the PLUCY deconvolutions and with those reported in Paper I. We can also certify that the derived angular separation between J03.13 A and B does not vary with wavelength, contrary to a possible trend that was suggestive from Table 3 in Paper I.
Following standard HST photometric procedures (Whitmore 1997) and relying upon the values of the PHOTFLAM keyword appearing in the header of the WFPC2 frames, we have derived the integrated magnitudes of J03.13 in the STMAG system to be 16.5 and 17.3 0.1 through the F555W and F814W filters, respectively. Because the images of J03.13 are located at approximately 4 from the X CCD axis of the PC1, no CTE correction has been applied. Making use of the general relations between the STMAG F555W and F814W magnitudes and the Johnson V and Kron-Cousins I ones (Holtzman et al. 1995), the standard magnitudes of J03.13 are derived to be V = 17.3 and I = 16.8 0.1. We conclude that within the observational uncertainties, J03.13 A and B have not varied photometrically between March 1993 (see Paper I) and November 1995.
As already mentioned, the pointing of HST towards J03.13 was only approximate. The quasar was found to be at 13 south from the center of the PC1 frames, corresponding to the approximate coordinates previously published by Maza et al. (1993). Using the "metric" task in IRAF and from these slightly offset WFPC2 observations, we have derived the J2000 coordinates of J03.13 A and B to be 10h17m23.84s, 46 58.4 and 10h17m23.78s, 46 58.4 , respectively. These more accurate coordinates were mandatory in order to properly center the individual components of J03.13 within the standard 0.5 circular FOS aperture.
© European Southern Observatory (ESO) 1997
Online publication: April 6, 1998