GHRS Instrument Handbook
A discussion of calibrations and a summary of the GHRS calibration plan for Cycle 5 is presented in Chapter 5. You should not need to measure sensitivity for yourself because the GHRS is highly stable and we believe we have performed sensitivity measures as well as they can be done.Wavelengths are another matter. The default wavelength scale from routine data processing is good to 1 km s-1 rms or better for the medium-resolution and echelle gratings, more than adequate for most needs. However, the wavelength zero point can drift because of temperature changes and the effects of the Earth's magnetic field. Because of these latter effects, we recommend that individual exposures not exceed 5 to 10 minutes in length. That prevents any loss of resolution from smearing.
If the zero point of the wavelength scale of your spectrum is important to your science, you may wish to obtain a wavecal as part of your program. A wavecal is an ACCUM with a TARGET of WAVE specified using aperture SC2. A wavecal should always precede the science observations to which it will apply. Whether or not you ask for a wavecal, you will get an observation called a SPYBAL each time you use a new grating for the first time and about once every other orbit thereafter. A SPYBAL (SPectrum Y BALance) is a wavecal obtained at a fixed wavelength for each grating, and it is used to center the spectrum on the diodes in the direction perpendicular to dispersion. Thus the SPYBAL will not be centered at the same wavelength you are observing, but it is useful for determining a correction to the wavelength zero point, and the accuracy achieved is nearly as good as if a wavecal had been used.
Even if you obtain a wavecal or use a SPYBAL, it will apply only to the ACCUM immediately following it in time. This is because of the drifts mentioned above. For fairly bright objects it is possible to cross-correlate individual spectra to determine offsets and so to create a final spectrum with little or no loss degradation in the wavelength zero point. Such a technique will not work for objects whose radial velocities vary, of course, and so individual wavecals may be necessary. They may be needed for faint objects too.
You may wish other calibrations that are not described here. If they are not part of our calibration program (see Chapter 5) then you will probably need to obtain them yourself and will need to request the time to do so. Please consult us if you have questions.