SIG switching is used and the signal and reference spectra are present in each integration. For each integration after doing the total power calibration each sampler will have 2 spectra with different SIG values. The SIG=T spectra will be know as the signal spectrum and the SIG=F spectra will be referred to as the reference spectrum. This is the simplest case. It is not necessary to store a reference spectrum in the calibration module for later use nor is it necessary to pause the pipeline processing while a reference spectrum is determined.
A reference spectrum is determined from the data. There are several cases. In all of these cases, the reference spectra necessary to reduce the rest of the imaging data will be saved in the calibration module for later use. In several of these cases it is impossible to proceed to the next step until after all of the image data is taken. A quick-look short-cut should be described for each case.
Targeted reference position. The antenna is pointed at a specific reference position. There are at least two versions of this case. A single reference position may be observed prior to the imaging data. The imaging data may be bracketed by reference position observations (how is the reference spectrum then determined - averaging, interpolating, some other way?).
Synthesized reference spectrum. The imaging data is used to arrive at a reference spectrum using regions of the imaging region assumed to be free of signal. We need to describe the ways that one might describe that region. Is a single reference spectrum (a simple average) sufficient or is something more complicated desired.