9.7 Fringes and Baseline Ripples

All SWS bands show fringes to some extent. They are a modulation, with wavelength, of the flux falling on the detectors and originate on parallel plane surfaces in the light path which act as Fabry-Pérot etalons. Plane parallel surfaces at a distance of a few mm up to a few cm are perfect for generating FP effects in the infrared wavelength range. It is assumed that etalons are formed inside filters, on plane mirrors and, most conspicuously in band 3, in the BIBIB detectors themselves. Combinations of all these effective etalons result in a sometimes wild fringe pattern. Every reflecting surface which is added to the light path can add N-1 extra fringe components excluding all the overtone options. Bands 1, 2 & 4 are only lightly affected by fringes, whereas they are more pronounced in band 3. An example of fringes in the band 3D spectrum of a star can be seen in Figure 9.9.

Figure 9.9: Example of residual fringes in band 3A. The (scaled and shifted) RSRF is displayed in red to show that the residual fringes are indeed leftovers of the RSRF. In green is a defringed version of the same scan, shifted up by 100 for clarity. The baseline ripples show as a slow modulation on the spectrum.

Most of the observed fringe patterns disappear in the calibration process where the observation is divided by a similarly fringed relative spectral response function (RSRF). Some residual fringing still remains in the calibrated spectrum. However, its exact pattern is presumably determined by the location of the source in the slit. The fringe pattern does not shift as a whole when moving from one spot within the slit to another. Instead different components of the fringe pattern shift by different amounts, resulting in a completely altered fringe pattern. See also Figure 8.7. Another cause of extra fringing is found in the combination of the size of the observed object and the speed at which its spectrum is scanned. This combination determines the smoothing in the spectrum. Ideally the RSRF should match exactly the smoothness of the observation, which unfortunately is almost never the case. The RSRF was observed at full spectral resolution during ground tests using a black body source which filled the slit completely. For SWS02 and SWS06, and for slow SWS01s, we have an RSRF where the fringes are not sharp enough and for fast SWS01s the RSRF is not smooth enough. In the pipeline processing we try to remedy this by smoothing or peaking up the RSRF according to the speed of the observation. The three reasons mentioned above (location, speed and extent) can cause a residual fringe pattern in some bands of up to 10%. Table 9.2 lists the main residuals and the bands in which they occur.

Fringes can have an unknown effect on both the continuous flux level and on line fluxes. For monochromatic light the amplitude of the fringe pattern can be much higher. So line emission might so to say sneak through the fringing FPs with much higher (or lower) amplitude than the surrounding continuum flux, depending on where on the fringe the line is located. No study has been carried out the assess this effect.^9.2

For continuum flux another effect is present. When the signal passes through the fringing FPs some of the flux is reflected backward and other is passing through, but in essence we loose flux. When the signal is divided by a perfect RSRF this lost signal is restored. Such a perfect state is not given to us. The signal is divided by a not quite adjusted RSRF and any of the following things might happen. When dividing by a too weak RSRF which is properly aligned in wavelength we still miss signal in the original fringes. When the RSRF has too strongly peaked fringes we actually end up with too much signal; the fringes invert. And finally when the amplitude of the fringes is perfect but the fringes are out of phase, we get a signal which is on average okay but still has shifted fringes. So in all cases we have residual fringes, but we do not know what was the cause. After removal of the residual fringes we could have either too little continuum flux or too much or we might be lucky. Indeed baseline ripples have been found when comparing bright calibration sources with stellar models (Van Malderen et al. 2002, [42]). These baseline ripples have an amplitude of a few percent and they are shaped quite like the envelope of the fringe pattern on the RSRF.

Tools exist in OSIA and the ISO Spectral Analysis Package (ISAP) to carry out defringing of AAR data. The OSIA tool FRINGES removes (residual) fringes in a completely automated way from SPD or AAR data.

For more information on fringes, how they are modelled and how well they can be removed, see Kester, Beintema & Lutz 2001, [21].