During the Operational Phase a limited number of sources were observed repeatedly throughout the mission for cross-calibration purposes. The responsibility for instrument calibration rested with the Instrument Dedicated Teams (IDTs) located at VILSPA, and was supported by staff at the home institutes of the instrument teams. Demands from the different instruments and coordination of cross-calibration observations were handled by a dedicated `Calibration Liaison' team.

As a result of these cross-calibration observations, the overall agreement among the major instrument modes was assessed in several occasions during the mission. The observations confirmed that the internal consistency in terms of absolute flux calibration was in general comparable to pre-launch expectations or better and led to the discovery of some calibration problems in the few cases where larger discrepancies than expected were found.

During Post-Operations, an improved cross-calibration plan was set up, based on the use of the whole ISO Data Archive for cross-calibration purposes, extending the analysis to a much larger number of sources. This way it was possible to study specific calibration topics in detail on dedicated subsets as well as to perform statistical analysis of the results obtained after specific calibration improvements were implemented in the pipeline in many cases.

In addition, it was decided to change the global IDC strategy for cross-calibration, from the rather `instrument-oriented' approach which worked quite efficiently during the Operational Phase (but turned out to be inappropriate for Post-Operations) to a more `calibration-topic' oriented organisation. For this, up to eight different Working Groups were established, each one addressing a specific calibration aspect (glitches, beam profiles and pointing, transients, line profiles, polarisation, spectral matching, general photometric calibration and interactive analysis S/W). They were formed by members of the IDC and of the different NDCs, with all instruments represented, plus external contributors from other projects (ASTRO-F, SIRTF, Herschel,...) in many cases.

The progress made on each calibration topic under analysis was reflected every six months during Post-Operations in a series of (Cross-)Calibration Status Reports posted at the IDC web site where information was included on the latest achievements in the calibration of all four ISO instruments with special emphasis on cross-calibration aspects. At the end of the Post-Operations Phase (end of 2001) each of these Working Groups generated a Final Report, containing the final state-of-the-art of ISO calibration regarding each of the calibration problems addressed. All of them have been put together in the ISO (Cross-)Calibration Final Report, [59] available at:

This organisation turned out to be an ideal platform to exchange knowledge across the boundaries of the ISO instruments. The information gathered and the conclusions derived from the investigations carried out by members of these Working Groups on the many different calibration problems affecting the ISO instruments constitute an extraordinary source of information for other future space projects. Most of these results were presented to the astronomical community at The Calibration Legacy of the ISO Mission, [120] conference, held at VILSPA on February 5-9, 2001.

6.1.1 Definition of cross-calibration sources

As we have already mentioned, only a small set of astronomical sources was used during the Operational Phase for cross-calibration purposes. It was composed of a few bright stars (like HR 7310, HR 7341, HR 6436, $\gamma$ Dra or $\alpha$ Boo), planetary nebulae (NGC 7027 and NGC 6543), galaxies (Mrk 279), asteroids (Cybele, Hygiea) and planets (Neptune). For most of them, observations exist in IDA taken throughout the whole ISO mission with more than one instrument and in many cases covering a wide range of wavelengths.

During Post-Operations, and as part of the new cross-calibration strategy above described, the concept was extended to any source satisfying at least one of the following conditions:

A more detailed description of the selection criteria used to consider a given observation in IDA as suitable for cross-calibration purposes can be found in the ISO (Cross)-Calibration Final Report, [59].

6.1.2 Limitations and caveats

Because of the limited common dynamic range between the ISO instruments, not all kind of internal cross-calibration comparisons were possible. In this respect, the cross-calibration of LWS with PHT in the long wavelength range was especially difficult, since the dynamic range overlap between PHT and LWS was very limited.

When comparing results derived from different instruments/detectors we also need to consider the differences in beam and filter profiles as well as in spatial and spectral resolution. This applies both to internal and external cross-calibration.

In addition, in the case of external cross-calibration, results obtained from facilities other than ISO may be based on different absolute calibration systems: different zero-points, different set of astronomical standards, different models used for the same standards, can introduce an undesired bias which has to be taken into account.

Last, but not least, some of the sources included for analysis may have been observed slightly off-axis, or show unexpected variability or extended emission not previously reported in the literature and, thus, can introduce some noise into the system. This makes cross-calibration even more complicated since individual spectra must be carefully checked one by one before being validated (see Figure 6.1).

**Figure 6.1:** Example of the discontinuities observed in SWS spectra as a consequence of a bad pointing (top) compared to the discontinuities due to the extended nature of the source observed (bottom). Note the always positive flux jumps towards longer wavelengths reflecting the progressive increase of the SWS apertures in this latter case.
$\resizebox {9.0cm}{!}{\includegraphics{nml_cyg.ps}}$ $\resizebox {9.0cm}{!}{\includegraphics{w3_extended.ps}}$

On the other hand, the accuracy of the absolute flux calibration is always limited by the accuracy of the available models. Stars are too faint in the long wavelength range of ISO (beyond 100 $\mu$ m) to be used as standards, and asteroids and planets have to be used instead. While stellar models are believed to be accurate to $\pm$ 5% or better, the uncertainties in asteroid and planet model spectral energy distributions are larger (10-15% level). In the case of asteroids their characterisation is additionally complicated by their variability, which limits the accuracy of the absolute flux calibration of ISO at long wavelengths.

In the following we will present a summary of the main results obtained from a number of internal and external cross-calibration analysis as well as from some consistency checks performed against available models. A more detailed description of how these comparisons were made can be found in the ISO (Cross)-Calibration Final Report, [59].

Note that we intentionally exclude from this analysis internal cross-checks made by each instrument team within their own instrument detectors/bands/AOTs. These are addressed in the instrument specific volumes of this Handbook and will not be discussed here.

6.1 Cross-Calibration Strategy

6.1.1 Definition of cross-calibration sources

6.1.2 Limitations and caveats