2. Data collection and reduction
The observations were performed using the CEA/SAp mid-IR camera CAMIRAS (Lagage et al. 1993), equipped with a Boeing 128x128 pixels BIB detector which is sensitive up to 28 µm. The AGN in NGC1068 was observed with CAMIRAS attached to the ESO/La Silla 3.6m telescope on 1998 November 8-10 and to the CFHT/Hawaii 3.6m telescope on 1999 August 1. During these two runs seeing and weather conditions - humidity and amount of atmospheric precipitable water - were particularly favorable to observing in the mid-IR and extremely stable.
The orientation of the array on the sky was carefully determined at the start of each observing run. The pixel size was (20 µm window, ESO/La Silla) and (11 µm window, CFHT/Hawaii).
In the 20 µm window, we used a filter centered at 20.5 µm and with a bandpass (FWHM) = 1.11 µm. This filter is free of any important line contribution. In the 11 µm window, we used a filter centered at 11.2 µm and with a bandpass = 0.44 µm: although the 11.3 µm PaH line emission falls within this filter, one should notice that the spectroscopic results from the ISO satellite do not show its presence in the AGN of NGC 1068 (aperture of ). Therefore we are confident that these two filters reflect mostly the mid-IR continuum emission of the AGN. To avoid saturation of the detector by the high ambient photon background, the image elementary integration times were chosen to be 9.1 msec and 19.1 msec, respectively at 20.5 and 11.2 µm. Standard chopping and nodding techniques were applied with a chopping throw of . Elementary images were coadded in real time during 3 sec chopping cycles. At 20.5 µm a total integration time of 22.5 min was spent on the source observed at an airmass less than 1.3. At 11.2 µm, the total integration on source was of 65 min and the mean airmass of 1.04. A shift-and-add procedure was applied to the final images through each filter. The final signal-to-noise ratio at the peak emission on the images is 265 at 20.5 µm and 980 at 11.2 µm.
The nearby reference star 4 Eri was frequently monitored to serve as a PSF and flux standard at 20.5 µm, while Ceti was used at 11.2 µm (van der Bliek et al. 1996). The precision achieved in flux measurements is +/- 5% and +/- 10%, respectively at 11.2 µm and 20.5 µm.
Because final images are limited by the seeing and above all by the 3.6m telescope diffraction-pattern (FWHM of and at 20.5 µm and 11.2 µm respectively), applying a deconvolution procedure is mandatory. We use the Entropy Method developed by Pantin & Starck (1996), based upon wavelet decomposition. The Shannon theorem (1948) prescription allows to persue the deconvolution down to twice the pixel size, in this case down to , and iterations are stopped according to the residual map, whose properties should be consistent with data noise characteristics.
© European Southern Observatory (ESO) 2000
Online publication: December 5, 2000