Astron. Astrophys. 363, 869-886 (2000)

2. Observations

We observed the emission of the J=1-0 line of ¹²CO in NGC 3593 using the IRAM Plateau de Bure interferometer between May 1997 and July 1998. The antennas were equipped with dual-band SIS receivers yielding SSB receiver temperatures around 40 K at the observing frequency. The system temperature oscillated between 350 K in winter and 500 K in summer. The spectral correlator was centered at 115.029 GHz (i.e., the transition rest frequency corrected for the galaxy's redshift derived from (HI)630kms^-1) with five units covering a total bandwidth of 480 MHz. The three central units (each, 80 MHz wide) provided a nominal frequency resolution of 0.625 MHz (1.63 km/s), whereas the remaining two units (each, 160 MHz wide), sampled the low and high frequency ends of the correlator bandpass with a frequency resolution of 2.5 MHz (6.52 km/s). The correlator was regularly calibrated by a noise source inserted in the IF system.

Visibilities were obtained using on-source integration times of 20 minutes framed by short ( min) phase and amplitude calibrations on the nearby quasars 1116+128, 3C273 for all observations up to July 1997, and, subsequently, on 1055+018, 1156+295. The data were phase calibrated in the antenna-based mode. On average, the residual atmospheric phase jitter was less than , consistent with a seeing disk of - size and with a % loss of efficiency. The flux of the primary calibrators was determined from IRAM measurements and taken as an input to derive the absolute flux density scale in our map, estimated to be accurate to 10%. The bandpass calibration was carried out using 3C273 and is accurate to better than 5%.

Observations were carried out to produce combined maps from two adjacent fields (each 42" wide and labelled as E and W) covering the nuclear CO emission of the NGC 3593 disk as follows: when observing field E(W), the phase tracking center of the interferometer was shifted by 10" (1/4 the primary beam of the antennas) to the east (west) of the mosaic center, located at and 49´06". The position angle (PA) of the galaxy disk is close to 90^o (PA=92^o; de Vaucouleurs et al. 1991), therefore and axes run nearly parallel to the major and minor axes respectively. The mosaic assures a coverage of 3.7 kpc (2.5 kpc) across the major (minor) axis, assuming the distance to be D=12.4 Mpc (Wiklind & Henkel 1992). The point source sensitivity derived from emission free-channels of 2.5 MHz width, varies between 3.1 mJy in field E and 3.5 mJy in field W.

The image reconstruction was done using standard IRAM/GAG software. Individual maps (256192 pixels with a 0.4" sampling) were created for E-W fields. The calibrated visibilities of each field were uv-shifted to the phase tracking center and reduced to dirty maps using natural weighting and no taper. The two maps were then combined to a single dirty mosaic, deconvolved and corrected for primary beam attenuation using the algorithm described by Gueth et al. (1995). The maps were cleaned with the Clark (1980) method to an average noise level of 3.0 mJy and restored by a 4.0 gaussian beam (with PA=22^o). While the difference in the synthesized beam shape due to the inhomogeneous uv-coverage and weighting in each subfield was taken into account in the image deconvolution process, the synthesized beam was deliberately fixed to the beam of the western subfield in the final image restoration process. Though the primary beam of each field was truncated at the 20% level to reduce effects from pointing errors in the image reconstruction process, the sensitivity gain in the central region is close to according to an rms noise level of 3.4 mJy/beam in the cleaned maps. No continuum was detected towards NGC 3593, down to an rms noise level of 0.45 mJy/beam in a 140 MHz wide band centered at 115.180 GHz. The conversion factor between Jy/beam and K is 7.8 mK/(mJy/beam).

2.1. Additional data

To better understand gas content and dust extinction in NGC 3593, we have acquired several additional images from the literature and from the HST archive. Corsini et al. (1998) kindly furnished us a digital version of their H image, and Moriondo et al. (1998a) their J- and K-band images. Because we wanted to probe the central regions of NGC 3593, we have also retrieved from the HST public archive the data and the calibration files for NGC 3593, a galaxy included in the study of Böker et al. (1999). NGC 3593 was one of 94 galaxies observed in the near-infrared continuum (with the F160W broadband filter) and in Pa (with the narrow band filter F187N), using the NIC3 camera on board of the HST mission. Following Böker et al.'s first steps, we re-reduced the two images with the best calibration files, and used the algorithm of van Marel (see Böker et al.) to eliminate the so-called pedestal effect.

In order to get an unbiased picture in Pa, we needed to sensitively subtract the continuum in the F187N image. Böker et al. (1999) made a first-order continuum subtraction of the F187N map by assuming the average ratio of intensities F187N/F160W to be constant to first order for the fainter pixels. However, application of their method to NGC 3593 seems to overestimate the continuum flux inside r10" since many of the central pixels become negative. The originally negative areas seem to be characterized by a color (defined as m-m) which is roughly 0.1 mag bluer than that estimated from the global fit on the fainter pixels. The strong color variations probably arise because the Pa emission is relatively weak inside the ring (r10"). When we locally estimate the continuum contribution to the F187N image and subtract it, we obtain a positive everywhere image with a rather symmetrical elliptical ring of ionized gas (see Fig. 4a).

© European Southern Observatory (ESO) 2000

Online publication: December 5, 2000
helpdesk.link@springer.de