Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 362, 310-324 (2000)

Previous Section Next Section Title Page Table of Contents

2. Observations

2.1. ISOCAM observations and data reduction

The observations were obtained with ISOCAM in September 1996 using a 6"[FORMULA]6" pixel field of view for the 32[FORMULA]32 element mid-infrared camera, covering a 3´ field of view for each array pointing (see Cesarsky et al. 1996for a complete description). The observations made with the broad filters were performed as square 3[FORMULA]3-step raster maps with a shift (i.e. overlap) of 16 pixels between successive positions giving a final total field of view of 7.8´[FORMULA]7.8´. The integration time was 2.1 seconds (time for one exposure). A number of exposures varying from 7 to 17 according to the filter were eliminated at the beginning of each raster map. An additional 15 exposures were taken for each raster position for all the filters in order to insure better stability of the detectors, which carry remanents of their previous illumination history. The integration times were thus 283.5 seconds per filter. The filters were LW2 (5.0-8.0 µm), LW3 (12.0-18.0 µm), LW4 (5.5-6.5 µm), LW6 (7.0-8.5 µm), LW7 (8.5-10.7 µm), LW8 (10.7-12.0 µm) and LW10 (8.0-15.0 µm). The raw data were then processed in the usual way using the CIA software 1. A library dark current was subtracted form the broad filter data and a flat-field correction was made with a flat field constructed from the data themselves. The new transient correction described by Coulais & Abergel (1999) does not give reliable results for the bright point-like sources. We thus treated the broad band ISOCAM images with the software built by Starck et al. (1999, inversion method). Corrections for field distortion have been applied to filter images before combining them in each raster. A second-order flat-field correction was finally used to match the levels on contiguous edges of the elementary maps of the rasters. This correction is only of a few percent and affects the photometry in a negligible way. The zodiacal light background used for the raster maps was the lowest emission level in each broad band image. This eliminates efficiently the zodiacal light which is distributed uniformly, but not if very extended emission is present.

Full scans of the two CVFs in the long-wavelength channel of the camera have been performed by decreasing wavelengths in April 1997. The total covered wavelength range was 5.15 to 16.5 µm. Each wavelength was observed 12 times in each scan leg, with an elementary integration time per measurement of 2.1 second. The total observing time was 4500 seconds, almost entirely used on-source. To correct the raw data for the dark current we used the dark model developed by Biviano et al. (1999) that takes into account the variation of the dark current inside a single revolution and among all the revolutions. Then, the data was deglitched and corrected for the transient response of the detector. Once again we applied two transient methods, the inversion method (Starck et al., 1999) and that which uses the Fouks-Schubert equations (Coulais & Abergel 1999). The latter method gives unreliable results in the short channel of the CVF, with negative flux for most of the pixels (before zodiacal light subtraction). This is probably due to an overestimation of the dark current which should also be corrected for the transient response. Thus to be coherent with the adopted raster map transient correction, we present here the CVF observations corrected with the inversion method. Flat fielding was done using dedicated CVF zodiacal measurements that take into account the stray-light due to the mirror and reflections between the CVFs and the detector (Boulanger, private communication).

The background of the maps is dominated by zodiacal emission that must be subtracted. The whole field of the CVF observations also collects extended emission of the N66 region, preventing us from measuring and subtracting the zodiacal light using the classical methods. In order to correct for zodiacal emission, we used the redundancy between the raster maps and the CVF observations in the following way:

  1. We verified that the background emission of the raster maps is consistent with the COBE-DIRBE zodiacal light measurements scaled for the single orbit in which all the maps were made (Reach, private communication). Due to the small size of the observed field and to the high ecliptic latitude we can safely assume that the zodiacal emission is uniform over the field.

  2. We measured in the raster maps, using Point-Spread-Function (PSF) fitting, the fluxes of point-like sources that are also seen with the CVF, and the backgrounds around them. For all these sources, we then obtained a measure of the sum of the zodiacal background and of the diffuse SMC background in the 7 broad-band filters.

  3. We built the equivalent of the broad-band filter images from the CVF data cube using the transmission curves of the filters given in the ISOCAM cookbook, and made the measurements of fluxes and backgrounds for the same point sources. The source fluxes provide a mutual calibration between CVF and filters. For each filter the background measurements are the sum of the diffuse emission (N66 and SMC) and of the zodiacal light. Between the two sets of observations (broad band filter and CVF) the only component that could change is the zodiacal light, since the solar elongation of the field changed between the filter and CVF observations. Thus, the ratio between the two background levels gives the variation of the zodiacal light between the two sets of data. We have found this ratio to be quite constant, equal to 1[FORMULA]0.3 for each filter except for the two shortest-wavelengths filters for which the zodiacal emission is very low and our determination uncertain. We have thus obtained a spectrum of the zodiacal light which is approximately the same for the filters and the CVF observations.

  4. The intensity of this spectrum is approximately half of that of the CVF zodiacal light spectrum published by Reach et al. (1996), due to the high latitude of the SMC. We have thus subtracted this spectrum multiplied by 0.5 from our CVF data. The final result is a position-wavelength data cube with zodiacal light subtracted from which one can extract spectra at given positions or monochromatic maps.

There is a slight position shift between the observations made with the Short-Wavelength CVF ([FORMULA]m) and those made with the Long-Wavelength CVF ([FORMULA]m) due to a slightly displacement of the SW-CVF and LW-CVF. Its effect is unimportant outside regions like the brightest emission peaks. In these cases we sometimes had to interpolate pixels or use the spectrum of the Short-Wavelength CVF of the adjacent pixel that matched the level of the Long-Wavelength CVF at 9 µm in order to produce a reasonable spectrum.

It should be emphasized that the reduction of ISOCAM data is not yet in its final stage. We estimate that the uncertainties in the intensities of the ISOCAM data presented here are [FORMULA]30 [FORMULA]. The main source of uncertainty for both broad band and CVF data is the not complete correction of the detector transient response. For the CVF data an additional uncertainty arises from reflections between the CVF and the detector (Okumura 2000).

Finally, the coordinates given by the satellite were affected by errors of the order of 10" both for filter and CVF observations. This is due to the lens jitter ([FORMULA]1-2 pixels) and not to the satellite. Fortunately several stars are visible in the filter observations, and allowed the recentering of the images on the Digital Sky Survey (DSS) images (Fig. 5). The CVF frames were recentered on the filter maps using the star at 0h 59m 27s, -72o 09´ 55" (marked as 755 on Fig. 2 and Fig. 5) which is detected in the 5 µm continuum map built from the CVF (Fig. 2). Fig. 1 shows the general outline of the filter observations as a map in the LW2 filter (6.75 µm) superimposed on an H[FORMULA] image.

[FIGURE] Fig. 1. Map of N 66 in the LW2 filter (6.75 µm) (grey scale) superimposed on an H[FORMULA] image (contours) from the survey of Le Coarer et al. (1993) kindly communicated by Margarita Rosado. Coordinates are J2000. The LW2 image has been smoothed to the resolution of the H[FORMULA] one, 9". The SE-NW bar is more marked in the IR image than in H[FORMULA]. Notice the spur in mid-IR emission extending to the NE of the bar, with no clear correspondence in H[FORMULA].

[FIGURE] Fig. 2. CVF map of N 66 in the 5 µm continuum (contours) superimposed on the DSS image (grey). Coordinates are J2000. Star N 346-755 in the catalog of Massey et al. (1989) is marked. It has been used to recenter the CVF images on the DSS image.

2.2. CO observations and data reduction

Observations of the CO(2-1) emission line at 230 GHz were obtained as part of the ESO-SEST Key program: CO in the Magellanic Clouds. The SEST telescope, located at La Silla Observatory (Chile) has a 15m diameter and a FWHM beam at 230GHz of 22". The backend used for the observations was a 2000 channel acousto-optical spectrometer (AOS) with a total bandwidth of 86 MHz and a channel width of 0.043 MHz. At the frequency of the 12CO(2-1) line the velocity range is 112 km s-1 and the velocity resolution is 0.056 km s-1. The observations were done in the position-switch mode, with a reference position far from the known CO emission zones. The receiver was a SIS mixer with system temperature of about 500 K. Intensity values were calibrated using the chopper wheel technique. The intensities are given in [FORMULA] (Kutner & Ulich 1981) and they take into account the correction for atmospheric attenuation and rearward spillover. To convert [FORMULA] to main-beam temperatures Tmb one has to divide [FORMULA] by [FORMULA]=0.60 at 230 GHz. Pointing was checked periodically on the SiO maser R Dor, and a calibration CO spectra towards the SMC source LIRS 49 (Rubio et al. 1996) was taken every day since R Dor could only be observed after the SMC.

Initially, N 66 was observed in the CO(1-0) emission line in a coarse grid with a 40"[FORMULA]40" spacing centered at [FORMULA](J2000)= 0h 59m 07.5s, [FORMULA](J2000)= -72o 10´ 26". This grid includes the entire field studied with ISO. Emission was found at the offset position (80", -80") and a fully-sampled map was done in the CO(1-0) line around this region (see Rubio et al. 1996). To improve the spatial resolution and derive the physical properties of the molecular cloud, N 66 was later fully mapped at 10" spacings, about half the HPBW of the SEST, in the CO(2-1) emission line. A 10 [FORMULA] 11 grid was made, each position observed with an integration time of 240 seconds giving an r.m.s. noise of 0.1K per channel. The 230 GHz observations were made with half of the high resolution backend, the other half being connected to the 115 GHz receiver. The spectra were smoothed and a linear baseline was removed.

A contour map of the CO(2-1) emission integrated over the velocity interval from 154 to 166 km s-1 is shown on Fig. 12, superimposed on the LW2 (5.0-8.0 µm) image. More recently, we discovered faint CO emission associated with the peaks shown on Fig. 5. This will be discussed in a further paper (Rubio et al. 2000.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: October 30, 19100