Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 335, 522-532 (1998)

Previous Section Next Section Title Page Table of Contents

Appendix A: instrumental setup and data reduction

ISOCAM observations were obtained using the 3" pixel scale in the microscanning mode, in which frames are obtained by shifting and adding images taken on slightly different positions on the sky. The microscan rasters had [FORMULA] points, with a step of 3" between each consecutive pair of points. In turn, the images obtained at each raster position were the coaddition of a number of individual short exposures: in this way, basic parameters defining an observation are the integration time of each individual exposure and the number of exposures at each raster point. These parameters were adjusted for each field and filter separately, taking as a reference the expected flux of the faintest sources of interest in each frame, extrapolated by means of a fit to the measured ground-based fluxes at shorter wavelengths. The background sky level at each filter was roughly estimated by using the flux per unit column density and per unit solid angle integrated in the interval from 2 to 15 µm in [FORMULA] Ophiuchi, as derived by Boulanger & Pérault 1988. Hydrogen column densities were taken from Loren 1989. Then, the spectral energy distribution of interstellar emission found by Sellgren et al. 1985 for NGC 2023 was upscaled to the expected [FORMULA] Ophiuchi values by multiplying by the ratio of estimated integrated fluxes between both regions.

The combination of exposure times and frames per microscan position required to reach a given signal-to-noise ratio was chosen based on the output of the ISOCAM Time Estimator available at the ISO Proposal Data Entry Center in ESTEC. In many cases, the minimum required on-target times for a desired signal-to-noise ratio were obtained for a very low number of individual exposures at each raster position, using an accordingly long time per exposure. This would have been likely to result in a relatively large fraction of pixels discarded because of cosmic ray glitches, thus degrading the quality of the final data. To avoid this, we imposed a minimum number of 12 exposures per raster position.

A very important issue in flux-calibrating ISOCAM observations is the stabilization of the detector: the lower the flux step when the illumination of a pixel is changed, the longer its stabilization time becomes. In staring-mode observations, this can be overcome by taking a number of stabilization exposures before the ones to be actually used for flux measurement purposes; the number of such exposures needed to achieve a 90 % of the stabilized count rate in the pixels illuminated by a source is given by the ISOCAM time estimator as well. The situation changes when microscanning is used, as then there is a flux step in the pixels near the source position every time that the telescope moves to a new raster position. Introducing the necessary stabilization frames at the beginning of each raster point would then imply a prohibitive investment of time. Fortunately, the transient behaviour of the ISOCAM arrays is at present well understood and modelable (Abergel et al. 1996), and a number of algorithms have been developed aimed at providing the necessary corrections leading to a correct flux measurement (Siebenmorgen et al. 1996).

We used an independent Astronomical Observation Template (AOT) for each observation at each filter. Given the low fluxes expected for our sources, a large number of stabilization images (calculated using the ISOCAM Time Estimator as mentioned above) with the same integration time as the science images was taken prior to the first exposure on the first raster position. The observations were then reduced using "CIA", the ISOCAM Interactive Analysis software (Delaney et al. 1996). The reduction procedure of each data cube (i.e., the set of individual images for each field and filter) can be outlined as follows (descriptions of the different steps, algorithms and nomenclature can be found in the ISOCAM Data User Manual, Siebenmorgen et al. 1996):

  • Slicing of the data cube into its individual images.

  • Dark frame subtraction, using the Cal-G dark frame.

  • Cosmic ray glitch removal, using the multi-resolution median transform method.

  • Stabilization correction, using the Saclay transient model fitting.

  • Normalization to unit exposure time.

  • Flat field division, using the Cal-G flat field frame.

  • Raster reconstruction.

The processed frame obtained in this way was then transformed into a FITS file, with each pixel value given in ADU s-1 normalized to unit gain. A gain factor of 2 was used for all the exposures, as recommended in the ISOCAM Users' Manual, and given the low flux levels of our targets. The pixel values were then transformed to flux units by using the ADU-to-mJy conversion factors for each filter given in the ISOCAM user's manual. No correction for deviations from the [FORMULA] behaviour assumed in the ADU-to-mJy conversion factor were applied, which seems acceptable in view of the spectral energy distributions in the ISOCAM domain presented in Fig. 1.

Digital photometry was finally performed, and magnitudes were derived, in the way described in Sect. 2.2. As a consistency check of the data reduction and analysis procedure, it is worth noting that the agreement between the ground-based [FORMULA] measurements and the magnitudes derived in the SW1 filter is in general very good.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: June 18, 1998