In the most general case, the SPI image reconstruction is an intractable problem as it would require solving for one background value for each combination of detector, pointing, and energy bin in addition to solving for the source parameters. This problem has more unknowns than measurements. Clearly, some assumptions on the background behavior have to be introduced.
For all point source analyses, we recommend to use the default flat field option. In this case, pre-defined flat-field spectra are used. They are rebinned to match the bin size of your analysis, and the program will select and use the appropriate flat-field spectrum for your analysis automatically. With this option, the detector to detector ratios are kept fixed (from that of the empty field), but one background coefficient is derived through the imaging reconstruction process. In fact, one coefficient is derived for each time interval, and the length of the intervals is provided as a number of consecutive pointing through the next parameter. We use a value of 5 in the cookbook example. The background is then assumed to be constant within intervals of 5 pointings. Note that in general, much longer time intervals could be used, in the range 20 to 100 pointings.
A good background modeling alternative is provided by either the GEDSAT template or the GEDSAT model, which are based on the count-rate of the events saturating the Germanium detector electronic. In this case, un-select the flat-field option and select the corresponding background option. For the GEDSAT model set the ``Number of Models to use'' to 1, and select the GEDSAT model. In the image reconstruction process, we will assume that the real detector background varies in the same way as this model, which will be scaled to the actual data to determine the real background. GEDSAT is a good model for most point-source analyses.
It is recommended to read further advises from the user manuals of the spi_obs_back and spiros programs before using more sophisticated background models. See also the documents provided by the SPI team [11] for difficult cases.