2.1. Flux Calibration

Four main effects limit the flux calibration accuracy:

1

The Spectral Energy Distributions (SED's) of the astronomical sources used for flux calibration are only known to between 4 and 10%.

2

Pointing, see section 5.5.

3

Memory effects, see section 5.3.

4

Dark current subtraction.

In the photometric calibration of the SWS detectors, a flux scale is set for converting calibration sources. Most of these calibrators are non-variable stellar point sources of K1-5 III, whose absolute flux levels are given by careful selection of observed photometry and spectrophotometry with traceable calibrations (e.g., [Cohen, M. et al.\1995] ), and non-LTE model atmospheres (e.g., [van der Bliek, N.S.1997] ).

During the conversion of the signal from an observer's science observation, responsivity drifts in the detection system are taken into account by comparing signals from the science target to those of the astronomical calibrator at particular ``key'' wavelengths (or passband), and to those of an internal calibration source. One key passband is used for each AOT band, whose central wavelength is chosen on the basis of detector responses and spectral liabilities of the astronomical calibrators. The internal measurements are done in every observation, and establish linearity between external science and calibration target signals. This linearity can be ensured over windows of approximately 200 revolutions. Over longer periods, linearity is difficult to maintain due to separate drift properties of the internal calibrator and detectors. The wavelength dependent responsivity variations are accounted for with relative spectral response functions determined prior to launch, modified inflight for point-source diffraction losses and other low order instrument or telescope effects.

Since the responsivities measured from the dedicated calibration observations, combined, and stored in response tables, are taken from many scans of astronomical sources with a wide range of spectral properties and flux levels, any science target can be flux calibrated over its observed wavelength region. The accuracy to which this can be accomplished, however, is limited by sensitivities, short-term drifts (hourly timescales), and possible memory or hysteresis effects. These impose obstacles for giving a straightforward snapshot of the current photometric accuracies that is at the same time not unrealistic given the complexity of SWS. The current photometric accuracies would best be represented in a series of statistical time-, wavelength-, and signal-dependent RMS curves. These will be made available soon, but an idea of what has been achieved inflight for the grating at the time of the OLP 6 delivery (April 97) is shown in table 1. The FP flux calibration accuracy is tied to LW grating sections.

The quoted accuracies are 1-sigma RMS deviations from expected levels of cross-calibrated flux densities of standard stars and representative science targets. Note that the maximum uncertainties refer primarily to AOT band ends where higher order differences between pre-launch and inflight relative spectral responses continue to be characterized, and remembering that the relative spectral response variation is defined to be unity at each key wavelength.

Flux reproducibility was checked by observing the standard star HR6705 twelve times during various orbits. Fluxes between the various measurements agreed to within 6% for band 1, and 12 to 15 % for bands 2 to 4.