Baseline removal

The EKH method guarantees that the mean of every scan should be zero (the transputers remove the mean on-line). In the absence of spikes the data would not need baseline removal but in some cases a large spike can adjust the mean of the scan and the baseline should be recalculated after spike removal (with despike2).

For the ``Emerson II'' method the situation is more complicated since the mean of each scan is now not guaranteed to be zero (and in fact the transputers do not attempt to remove a baseline in this case). scan_rlb must be run in order to remove the baseline (each bolometer sees a slightly different background). For data where the scans are long enough to be off-source LINEAR baseline removal can be used. For more complicated source structure MEDIAN is worth a try although extremely complicated regions (e.g. OMC-1) may cause problems.

In order to overcome this problem it is also possible to specify specific scans that can be used for calculating the offset level since the DC level appears to be fairly constant during an integration. When the SECTION baseline removal method is selected a SCUBA section (§) can be used to specify exposure (scan) numbers or actual positions in the data stream. Usually the first and last exposures are used since these are most likely to be `off-source'. The appendix on scan_rlb contains some examples on the use of SCUBA sections to select baseline regions.