5.5 Background Spectra, Single Observation

When source crowding is not a concern, local background spectra are constructed by removing the point sources from the data (§7.12) and then searching around each source for the smallest circular region that encompasses at least a minimum number of background counts specified by the observer. As in the source spectrum, the BACKSCAL keyword in the single-observation background spectrum is computed by estimating the integral of the exposure map over the background extraction region.

When crowded sources significantly contaminate each other's extraction regions, more accurate local background spectra may be constructed from unmasked data by modeling the spatial distribution of the point source flux and then carefully constructing background regions which seek to subtract background components arising from neighboring sources (§7.14).