5.4 Calibration sources and achieved accuracy

5.4.1 Spectroscopic calibration

5.4.1.1 Grating calibration

For version 6 of the pipeline, a new wavelength calibration was derived consisting of

• new detector angles, which differ from the previous ones by a small amount ranging from -.1 (SW1) to +.04 (LW5)
• new coefficients for the relationship grating angle vs lvdt which were derived using all calibration data from rev 93 to rev 335. They consist of 280 records of 6 fine structure lines (52 m, 57m, 63m, 88m, 145m, 158m) from 4 different sources (NGC6543, NGC6826, G298.228, IRAS15408-5356)

The size of the grating spectral element was determined and found to be equal to the size that was quoted in the LWS Observers Manual, i.e. 0.29 m for the SW detectors and 0.60 m for the LW detectors.

The sources and lines that were used are given in Table 5.2.

 

Table 5.2: The sources and emission lines that are used for the wavelength calibration of the grating and of both FPs. For the FPS calibration all lines were observed in three FP orders. For the FPL the lines were observed in two orders. Type indicates the source type (PPN: Proto Planetary Nebula; PN: Planetary Nebula).

Source	Type	Lines
Grating wavelength calibration
NGC6543	PN	52, 57, 63, 88 m
NGC6826	PN	52, 57, 63, 88 m
G298.228	HII region	52, 57, 63, 88, 145, 158 m
IRAS15408-5356	HII region	52, 57, 63, 88, 145, 158 m
FP wavelength calibration
NGC7027	PN	63.184 [OI], 157.741 [CII]
NGC3603 P1	HII region	51.814 [OIII], 57.317 [NIII], 157.741 [CII]
NGC6826	PN	51.814 [OIII], 88.356 [OIII], 157.741 [CII]
G0.6-0.6	HII region	51.814 [OIII], 57.317 [NIII], 63.184 [OI], 121.898 [NII]
		145.525 [OI], 157.741 [CII]

The accuracy for the grating wavelength calibration quoted here is based on several observations of NGC6543, and two targets that were not used for the derivation of the coefficients: NGC7027 and W Hya. The measured lines had intensities ranging between and  W/cm . 85% of the measured lines show a deviation from their expected wavelengths of less than 25% of a grating resolution element (0.07 m for the SW detectors and 0.15 m for the LW detectors). The maximum deviation that was measured was 0.27 m on LW4 or 45% of a resolution element. See also section 7.5 for some caveats on the grating wavelength calibration. The accuracy that is achieved for the wavelength with the grating is given in Table 5.3.

 

Table 5.3: Accuracy of the wavelengths for an LWS spectrum. These accuracies are based on actual measurements. 85% of all measurements were within these limits for the accuracy. Note that the accuracy for the FPs can be improved using the new calibration coefficients given in section 7.7.

mode	Accuracy
grating calibration	25% of a resol. element
	0.07 m for SW detectors
	0.15 m for LW detectors
FPS calibration	m RMS
	scatter in the fitted calibration
FPL calibration	m RMS
	scatter in the fitted calibration
FP absolute accuracy	40 km/s due to errors in FPS
	and FPL calibration

5.4.1.2 Fabry-Perot calibration

The RMS scatter in the fitted calibration of the Fabry-Perots gives a measure of the internal accuracy of the calibration. This was determined to be m for FPS and m for FPL. However, tests using several strong lines on a single source have shown a significant velocity difference between lines observed with FPS and FPL. The differences can be as large as 80 km/sec with FPL giving systematically higher velocities. This difference is due to an error in both the FPS and the FPL calibrations (see section 7.7).

The sources and lines that were used are given in Table 5.2. The accuracy that is achieved for the wavelength with the FPs is given in Table 5.3.

5.4.2 Photometric calibration

5.4.2.1 Grating calibration

The accuracy of the photometric calibration is determined by a number of factors:

• The accuracy of the model for Uranus that was used to calculate the absolute response is believed to be about 5%. However some spectral features that are present in the spectrum of Uranus were not present in the model, leading to spurious features in the spectra (see section 7.6).
• The measurement of Uranus that was used for the calibration had a signal to noise ratio of about between 180 (for the SW4 detector) and 520 (for the LW1 detector). Therefore no data product derived using the calibration on Uranus can have a S/N ratio larger than this, even if the S/N on the measurement is higher. New observations on Uranus are planned to improve the S/N ratio of the calibration.
• Transients or memory effects may have an influence on the photometric accuracy of the data. The extend of their influence is not clear at this time.
• Ramp (non) linearity will also influence the accuracy. It is believed that Derive-SPD is handlilng this reasonably well, but improvements are being investigated. The size of this effect is not clear at this time, and will depend also on the source.
• The dark background removal will, especially for faint sources be an important factor in the photometric accuracy. The size of this effect depends on the source strength and the spectral shape. For very faint sources the drift correction applied in AAL may result in negative fluxes (see section 7.13).
• Glitches also influence the photometric accuracy, since they have an effect on the responsivity of the detectors. The size of this effect depends on the time in the orbit when the observation was carried out.

All these factors together lead to a photometric accuracy for LWS grating mode spectra of 10% between repeated scans on the same detector (this is mainly due to the effect of responsivity changes), and 30% between adjacent detectors (mainly due to responsivity changes and dark backgroound removal problems).

5.4.2.2 Fabry-Perot calibration

For the Fabry-Perot mode, the photometric accuracy was determined by comparing the integrated line fluxes observed with the FP with the fluxes observed with the grating or line fluxes published in the literature. The sources and lines are given in Table 5.4. It was found that for strong lines accuarcy is typically better than 30%. For faint lines however, the FP fluxes can be off by almost a factor 2. This is mainly due to the removal of the darkcurrent and the responsivity drift correction which are known to be less accurate for low signal levels (see also section 7.13).

Table 5.4: The sources and lines in this table were used for the determination of the photometris accuarcy of the Fabry-Perot data.