Next: 6.15 Side Order Contamination Up: 6. Caveats and Unexpected Previous: 6.13 Removal of the

6.14 Use of `Non-prime Data' in FP Observations

During each L03 and L04 observation the LWS FP and grating settings were optimised for the detector whose bandpass filter included the wavelength range of interest. This was designated as the `prime' detector. However, all ten LWS detectors recorded data simultaneously in their own spectral ranges. The other nine detectors are known as `non-prime' and often recorded useful data that can complement the prime data. These data could not be processed within the old version of FP_PROC, which only dealt with prime detector data. This procedure has been updated to allow the use of the non-prime detectors, as described in detail in Polehampton 2001, [34].

**Figure 6.13:** *Example of the raw data (after wavelength calibration) recorded on a `non-prime' detector during an L03 observation. There are clearly useful mini-scans present in the data (in colour).*
$\resizebox {16cm}{!}{\includegraphics{polehamptone1.eps}}$

In the LIA routine FP_PROC all ten detectors appear in a menu bar at the top of the screen. This allows non-prime data to be selected and processed in the same way as prime data. In order to process all non-prime detectors the FP throughput calibration was extended to cover wavelengths outside each FP's nominal range. The detectors within each FP's nominal range are shown in Table 6.3.

**Table 6.3:** *Detectors within each FP's nominal range.*
FPS	SW1 SW2 SW3
FPL	SW4 SW5 LW1 LW2 LW3 LW4 LW5

**Figure 6.14:** *The grating profile shape (solid line) is shown with the limits used to define `useful' data. The red and green mini-scans are classed as `useful' whereas the blue mini-scan is not.*
$\resizebox {11cm}{!}{\includegraphics{useful_data_fig.ps}}$

As a general guide to know how good a mini-scan is (see Figure 6.14), we define it as `useful' if it had at least one data point located above 90% of the maximum transmission of the grating. Assuming that the grating profile was approximately Gaussian, the 90% level occurred at a distance from the profile centre of:

$\begin{displaymath} x^{2}=-{\rm ln}~0.9~\bigg(\frac{FWHM}{2(2{\rm ln}2)^{1/2}}\bigg)^{2} \end{displaymath}$

(6.4)

where is the full width at half maximum of the grating profile. This gives a cut-off distance of =0.195 from the profile centre. This limit ensures that most of the prime data are defined as having good signal-to-noise. This limit is nevertheless only meant to give an indication. Some mini-scans might not meet the `useful' data criteria and still be good to use, only with a slightly lower signal-to-noise ratio.

A good demonstration of the improvement that can be achieved using non-prime data is illustrated in Figure 6.15 showing a comparison of prime and non-prime data for the 53 $\mu$ m OH lines in Sgr B2 (from Polehampton 2002, [33]).

**Figure 6.15:** *Comparison of prime and co-added non-prime data for the 53 $\mu$ m OH lines in Sgr B2. Prime FPS observation is in green and co-added non-prime observations (FPL) are in black.*
$\resizebox {11cm}{!}{\includegraphics{53_fps_fpl.ps}}$

Next: 6.15 Side Order Contamination Up: 6. Caveats and Unexpected Previous: 6.13 Removal of the

ISO Handbook Volume III (LWS), Version 2.1, SAI/1999-057/Dc