Fabry-Pérot observations with the LWS were implemented as a series of mini-scans. For each mini-scan, the grating was set to a fixed position and the selected interferometer was scanned over a range of gap values. At the end of the interferometer scan, the grating was moved to a different position and the next mini-scan began. Each mini-scan therefore sampled the grating response profile, which must be removed from the data. Removal of the grating response profile has proved difficult because of a non-repeatability of unknown origin in the grating position (Section 5.10). For high-resolution observations this non-repeatability corresponds to an uncertainty in the wavelength at which the grating profile was centred. Before OLP Version 8 the standard pipeline processing did not take into account this uncertainty: it just assumes that the grating response profile was centred at exactly the wavelength specified by the grating wavelength calibration. When the grating profile is removed from the Fabry-Pérot data, the result can be a spectrum that is significantly skewed. The unpredictable nature of the grating setting, combined with the fact that there is no independent means of determining the grating position, means we are forced to infer the grating position directly from the data.
For this reason, from OLP Version 8 onwards, the underlying grating
profile is not
removed in the standard pipeline processing. For L04 data, where the FP
scans were quite short, the effect should be very small. For L03, however,
there may be some effect on the slope of the continuum, depending on the
extent of the FP scan at each grating position.
The user is recommended to refer to FP_PROC, the LWS
Interactive Analysis
(LIA) FP processing tool which removes the grating profile. For L04
spectra
the user can remove the skewness interactively, thereby deducing the actual
grating position. For L03 observations, the grating
transmission profile is removed from each mini-scan separately and can be
shifted for each one until it matches best with its neighbours. This can
either be done interactively for each mini-scan or via an automatic least
squares routine which minimises the difference between overlap on adjacent
mini-scans.
New grating profiles have been derived, which are better characterised away from the maximum and thus have more accurate wings than the profiles used in the OLP calibration files. They allow to remove the grating shape across the entire mini-scans.