The release of the final version of the ISOPHOT Off-Line Processing software and the generation of the ISO Legacy Archive by this software closed the main period of the ISOPHOT calibration. In most observing modes, however, the ultimate photometric accuracy has not been reached yet, and the calibration work is going on in the framework of ESA's ISO Active Archive Phase project. The KISAG at Konkoly Observatory, Budapest, is funded by ESA's PRODEX programme in order to contribute to these calibration efforts. Here we describe our strategy for the future work.

So far the calibration efforts concentrated on the main instrumental problems which affect several or all observing modes. Also the solutions for the problems and the proposed correction schemes were as general as possible, in order to implement them easier in the automatized off-line processing. In many respects, however, this general approach - which was very successful in establishing the basic calibration of ISOPHOT - has reached its limitations. One possibility for further improvements would be to understand the specific problems of individual observing modes and work out dedicated correction algorithms which are not necessarily applicable to any other mode. This dispersion of the calibration process may rise difficult questions concerning documentation, exchange of information, and a homogeneous update of the Legacy Archive, but it also promises significant improvements in the final photometric accuracies. In the KISAG we follow this approach and concentrate on the calibration of a few individual observing modes. We assume that the main calibration steps implemented in the final version of the Off-Line Processing software (ramp linearization, reset interval correction, dark current subtraction, signal linearization, FCS calibration tables) are good enough, and we do not plan to re-analyse them. We do investigate, on the other hand, new methods for better elimination of cosmic glitches, transient correction, and a better extraction of the source's flux from a measurement. We started already working on point source photometry in minimap mode (AOT 22), and will continue with absolute photometry of point and extended sources (AOT P05, P25), and with chopped photometry (AOT P03, P22).

So far the correction algorithms for the general instrumental effects emerged from the analysis of specifically designed calibration observations. For the additional corrections we want to develop no dedicated measurements are available. One has to rely on science observations from the Archive performed in that observing mode. On the basis of ideas already discussed in several papers and internal calibration reports (Schulz 1998, Fajardo-Acosta et al. 1999, Laureijs & de Muizon 2000) we plan the analysis of a selected ISOPHOT observing mode (or sub-mode) as follows:

1. Perform an Archive query and select all - scientific as well as calibration - observations performed in the selected observing mode;
2. Identify all objects which could be used as standards, i.e. accurate prediction of their flux density is possible (see "Secondary standards for ISOPHOT");
3. Process the identified standard measurements in a homogeneous way using the best data reduction methods available;
4. Search for systematic trends in the [Measured-Predicted] residual flux densities (the most obvious plot is a correlation with the brightness of the source, but all possible relationships with the main measurement parameters are worth to check);
5. Try to understand the reason behind the observed trend (transient effect, beam profile, flatfield etc.) and invent new data processing methods to eliminate it, OR fit the observed trend and develop an empirical correction formula to correct for the systematic discrepancies;
6. Repeat (3)-(5) until no obvious trend can be detected. The remaining scatter defines the final measurement uncertainty;
7. Document the new data reduction steps as well as the empirical fits, so that other people having data in the same observing mode could profit from the results.

When the final accuracy of an observing mode turned to be significantly better than the one currently available in the Legacy Archive, one could consider a re-calibration of the Archive data.