Instrument calibration is vital, and rather complex, for a tilting filter instrument of such wide spectral range as ours.
A tilting filter instrument requires careful determination of the instrument function (i.e. the detailed shape of the monochromatic instrument response) at many different spectral positions. This has even been mentioned as the "principal disadvantage" of this kind of instrument (Meriwether 1984), and calibration data management could become quite difficult unless aided by computerized data acquisition.
In contrast, grating spectrometers have (approximately) constant bandwidth over the spectral range in which they are used, so that determination of the instrument function at a fixed wavelength may be sufficient, and only the determination of the spectral variation of instrument sensitivity (spectral response) requires measurements at different wavelengths.
Our instrument is calibrated as a whole, not its optical components individually (Scheer 1987; Reisin 1987). This has the advantage (among others) that calibration errors do not add up, and the effects of component alignment are automatically included.
The calibration process can be broken down into several steps:
1. Determine the shape of the instrument function at many different wavelengths. This requires suitable discrete monochromatic line sources, or a tunable source like a monochromator. Using a monochromator (our case) requires correction for monochromator bandwidth, unless very-high-resolution devices are used.
2. Determine the variation of spectral response at different wavelengths. This requires a continuum source of known spectral shape. In the near infrared, a common incandescent lamp is a very good approximation to a black body close to its spectral maximum, and the spectral slope is only weakly sensitive to the filament temperature of the lamp.
3. Wavelength calibration. This requires different spectral lines of known wavelength, over the spectral range. We take advantage of the fortuitious fact that simple Neon indicator lamps have several calibration lines, or groups of lines, in our spectral range, and that different makes of these lamps behave quite similarly (relative line intensities may vary). Some Neon lamps also have an Argon line (at 852.37823 nm vacuum wavelength) in the gap between two Neon lines, so that a total of seven reference positions are available for calibration. This eliminates the need for costly calibration lamps. Since the wavelength calibration of our instrument is strongly temperature dependent, it must be done over as wide a range of ambient temperatures as encountered in normal operation.
4. Absolute intensity calibration. We can omit this step. It would require a blackbody with an aperture that fills the entire field-of-view of the spectrometer. In the near infrared, this would be very difficult to achieve with reasonable precision (working at the short-wavelength slope of a relatively cold blackbody, where temperature sensitivity may be prohibitively large). With a hot blackbody (very hot, to make temperature sensitivity small), controlled attenuation over many orders of magnitude would be needed. Fortunately, for determining rotational temperatures, such an absolute calibration is not needed, and for observations of intensity variations, only instrument stability must be assured.
As a diffusing screen that is uniformly illuminated to fill the field-of-view, we simply use white paper (not a commercial MgO screen). Spectral neutrality of the paper is verified separately by using an optical setup involving two reflections off the same brand of paper (if there were any spectral modulation, it would show more clearly, with the double reflection).
For computation of synthetic spectra, the continuous functions describing the instrument at intermediate settings are obtained from the discrete set of calibration data by least-squares fitting.
Apart from the laboratory calibration that is a prerequisite for successful airglow data acquisition, housekeeping information collected automatically in the field adds useful data on instrument stability and can be used to extend the wavelength calibration over the ambient temperature range actually encountered.
J. W. Meriwether, Jr., "Ground-Based Measurements of Mesosphere Temperatures by Optical Means", in Handbook for MAP (Middle Atmosphere Program) Vol. 13, R.A. Vincent ed., Urbana Illinois 1984.
E. R. Reisin, "Medición espectroscópica de temperaturas atmosféricas en la zona de la mesopausa", Tesis de Licenciatura en Física, FCEyN, Universidad de Buenos Aires, November 1987 electronic edition (in Spanish, PDF file, 1945 kB)
J. Scheer, "Programmable tilting filter spectrometer for studying gravity waves in the upper atmosphere", Applied Optics 26, 3077, 1987.