Subsections

• 4.9.1 Ghost images from point sources
• 4.9.2 Ghosts and straylight from extended sources

4.9 Ghosts and Straylight

Ghost images and straylight affect any kind of photometry or morphology studies (Okumura et al. 1998, [45]; Blommaert et al. 2001a, [12]). The ISOCAM ghost image is formed after 2 reflections: on the detector and on the filter. For the fixed filters these effects could be minimised by tilting the filters with respect to the beam path. For the CVF making a tilt to avoid ghost and straylight effects was not possible due to the physical structure of the CVF and the constraints on its mounting. Thus both effects were expected to cause some problems in CVF data.

4.9.1 Ghost images from point sources

In Figure 4.22 three independent observations of the same star, $\delta$ Draconis, done with the CVF in the 6 $^{\prime \prime }$ pfov, at a wavelength $\lambda$ = 7.358 $\mu$m are shown in a logarithmic intensity scale. In one of the observations, the star's image was put at the centre of the detector array, in the other two, the source was placed in the bottom left and right corners of the array.

Figure 4.22: Ghost images in three CVF observations of $\delta$ Draconis. Intensity scales are logarithmic.

Two types of ghost images can be seen. One has a ring-like appearance and is comoving with the source image. The other has a point-like shape and moves symmetrically with respect to the ISOCAM array centre, as the source moves. Other images and movies of these ghosts are available in the ISOCAM web pages at:

http://www.iso.vilspa.esa.es/ $\rightarrow$ ISO Explanatory Library $\rightarrow$ CAM.

The ratio between the integrated ghost signal and the signal from the star can reach up to 30%, for the comoving ghost, and 9% for the symmetric ghost. Although the latter ghost contains less flux, it is very apparent in an ISOCAM image as the brightest pixel reaches a brightness (about 4% of the point source flux) which is comparable to that of the comoving ghost. The spectral dependency of the ghost contribution can be seen in Figure 4.23, taken from Blommaert et al. 2001a, [12].

Figure 4.23: The summed relative intensity of the first and secondary ghosts normalised to the point source flux in the main beam at 2 positions on the detector as a function of wavelength.

In order to avoid systematic errors in the CVF photometry one must be aware of the ghost flux contribution. Since the SRF was derived without including the ghost flux in the estimation of the total flux from a star, for consistency observers should not include the ghost flux in the estimation of the total flux of a given source. For a discussion on the way this can be done see Blommaert et al. 2001a, [12]; 2001b, [13].

4.9.2 Ghosts and straylight from extended sources

Straylight from extended sources was investigated through observations of the zodiacal light with three CVF step positions (corresponding to the wavelengths of 7.7, 11.4 and 15$\mu$m), and in all pfov's. In Figure 4.24 we show the image obtained on the zodiacal light emission at 11.4$\mu$m with 12.0 $^{\prime \prime }$ pfov. Standard data reduction without flat-fielding was applied. In this configuration, the field mirror acts as a field stop, and only the central 16$\times$16 pixels of the detector array should be illuminated. However, one can see from the figure that the region of the detector not covered by the field mirror is not completely dark and the field mirror edges are not sharp. This is mainly due to the ghosts of the extended source and to a lesser extent to straylight. The total flux outside the field mirror region sums up to $\sim$25% of the flux within the region covered by the field mirror. Ghosts and straylight caused an important limitation to CVF images. In CVF images, real physical structures with an average per-pixel flux $\leq$10% of the background are hard to detect. Figure 4.25 shows an example of a CVF image of the `flat' zodiacal emission. Indeed, the ghosts and straylight pattern may lead to mis-interpretations of the CVF images, e.g. the ghost pattern could be incorrectly attributed to extended emission around a target star. Of course, the situation gets worse when bright sources outside the field of view also contribute as straylight. In Blommaert et al. 2001a, [12] a method is presented to correct zodiacal CVF spectra for straylight radiation. However, the method has not been tested on CVF observations of extended emission with contrasted sources.

Figure 4.24: An image of the zodiacal background taken with the CVF at 11.4$\mu$m, with 12.0 $^{\prime \prime }$ pfov. Flux units are ADU/G/s. Note that even outside the field stop, outside the central 16$\times$16 pixels, the detector shows illumination

Figure 4.25: The so-called `two-lung' shaped ghosts in a flat extended source