The Infrared Space Observatory (ISO) of the European Space Agency ceased observations on April 8 1998, 10 months after its expected end. This mission was a great success (Kessler 1999) and many of the resulting scientific discoveries are yet to come. Many of the scientific analyses of the data are still held up by data reduction problems. In particular for the camera on board ISO (ISOCAM - Cesarsky 1996), instrumental effects which are not well understood limit its sensitivity.
Much effort has been made to model the response of the ISOCAM array based on theoretical understanding of infrared detectors (Abergel 2000; Coulais & Abergel 2000 and references therein). Nevertheless, no overall model of the ISOCAM response to flux steps and glitches is available. ISOCAM data reduction thus relies on an empirical understanding of the detector. Recently, sophisticated data reduction techniques have been developed to take into account some aspects of ISOCAM response (Starck et al. 1999; Desert et al. 1999; Aussel et al. 1999; Altieri et al. 1999). These methods are close to being optimal for the detection of point sources. Nevertheless, in many observations, instrumental effects still prevent the study of faint extended emission.
The development of the data reduction method presented in this paper was motivated by the analysis of ISOCAM observations of diffuse and translucent interstellar clouds. Such clouds, when illuminated by the solar neighborhood radiation field, have mid-infrared brightness with very low contrast with respect to the zodiacal light (at most a few percent). To reveal these low signal-to-noise ratio clouds, variations in the detector response have to be corrected to an accuracy better than a fraction of 1%. At the present time, available data-processing techniques do not achieve such high sensitivity for extended emission. In this paper, we present a data processing method that makes it possible.
The algorithms presented here apply to observations made in raster mode with the Long Wavelength channel of ISOCAM (LW). It makes use of the fact that a given position in the sky has been observed several times by different camera pixels. Tests have been performed on extended emission observations presenting low and high contrasts. We have checked that our methods can be applied to most raster mode observations, and as a consequence concerns a significant fraction of ISOCAM observations (extra-galactic, galactic and even solar system observations).
To illustrate the data processing we use two different observations of the same field, obtained sequentially in exactly the same configuration. The amplitude of instrumental effects are not the same in both observations, giving us constraints on the validity of our method. In Sect. 2 we present the observations used to illustrate the data reduction chain. In Sect. 3 standard data reduction techniques are briefly presented. The main problems of the standard reduction are described in Sect. 4 and the new data processing approach to address these problems is detailed in Sect. 5. The performances of the overall method are discussed in Sect. 6.
GRB1 | GRB2 | |
Date | April 5 1997 | April 5 1997 |
ISO Revolution | 506 | 506 |
Time since activation | 2.12 (hrs) | 3.42 (hrs) |
Filter | LW10 | LW10 |
Wavelength |
![]() |
![]() |
Right Ascension J2000 | 14h 50m 18.2s | 14h 50m 18.1s |
Declination J2000 | ![]() |
![]() |
Total number of readouts | 907 | 907 |
Number of raster steps | 8 ![]() |
8 ![]() |
Readouts per positions | 12 | 12 |
Step size (pixels) | 8 | 8 |
Exposure time (seconds) | 5.04 | 5.04 |
Pixel size | 6'' | 6'' |
Comment | Strong LTT | Good quality |
Copyright The European Southern Observatory (ESO)