These data were obtained in 2000 and early 2001. Observations of other standard stars (GD71, GD246, and GD659) were not used because the effective areas derived from their spectra are inconsistent with those derived from the above observations. The deviations are consistent with what one would expect from modest errors in the stellar model parameters. When revised models are computed, these observations may be included in the FUSE flux calibration.

The count-rate spectra from individual exposures in a single observation are carefully aligned using interstellar absorption features and then combined. If multiple observations are available for a given star and aperture, the spectra from all observations are combined to produce a final spectrum. All narrow features in the spectrum (ISM or photospheric absorption lines and uncorrected detector defects) are removed by replacing the affected pixels by a polynomial fit to the surrounding continuum. This ``cleaned'' spectrum is then smoothed by a 20-pixel boxcar and divided by the model spectrum (with units of ph/cm²/s/Å) to produce a new effective-area curve (in units of cm²).

For MDRS and HIRS observations, where effective slit throughputs may vary from one exposure to the next because of time-variable channel alignment, the exposures are grouped into two classes: those with fluxes exceeding 90% of the peak flux seen, and those with fluxes exceeding half of the peak flux. The exposures in the second group are renormalized to match the mean of the first group in order to produce an effective ``photometric'' flux estimate.

The LWRS effective areas are quite consistent from one star to the next, except for the prominent worm feature in LiF 1B. The situation is more complex for the MDRS aperture. In most segments there is a wavelength region where the effective areas are consistent and one or more where they differ. Shadows from the detector grid wires are seen in the raw data for each of the MDRS spectra, though not as prominently as for LWRS LiF 1B. Additional MDRS calibration data will be required to understand the range of variation and to determine if it can be corrected in future versions of the pipeline.

The worm seen in the LWRS LiF 1B effective area curves is removed by ``bridging'' the depression in each individual curve with a low-order polynomial, then computing a smooth correction to rescale the effective area. Application of the resulting effective-area curve to a spectrum containing the worm should always result in a depression in the observer's spectrum. This should be less confusing than the previous calibrations, which would often yield a broad bump alongside the depression whenever the worm's position in a data set differed from that in the calibration exposures.

The effective-area curves from the various stars are averaged, then smoothed three times in succession with a 40-pixel boxcar. The intent of the smoothing is to remove features on scales comparable to pixel shifts applied by the pipeline (Doppler shifts, astigmatism corrections, etc.), but to retain the larger-scale features that are recognizable in each of the individually-derived effective-area curves. The final effective-area curves for the LWRS aperture are plotted in this figure. Curves for other apertures can be found in the section entitled Comparison with CalFUSE v1.8.7.

The final effective-area curve is converted into an inverse-sensitivity curve with units of ergs/cm²/count/Å. Multiplying the observed count-rate spectrum by the inverse sensitivity gives the flux in ergs/s/cm²/Å.

     Correction Factors for LWRS Spectra

Date         LiF 1A  LiF 1B    LiF 2A  LiF 2B

2000.01.01   1.0000  1.0000    1.0000  1.0000
2001.01.23   1.0293  1.1412    1.1122  1.0453 *
2001.01.25   1.0118  1.1074    1.1005  1.0526
2001.11.21   1.0494  1.1914    1.1603  0.9824
2002.02.25   1.1001  1.3245    1.2107  1.0353
2002.06.19   1.1152  1.3738**  1.3014  1.0458
2003.01.04   1.1868  1.4466    1.3299  1.1003
2003.04.03   1.1776  1.4459    1.3050  1.0641

             SiC 1A  SiC 1B    SiC 2A  SiC 2B

2000.01.01   1.0000  1.0000    1.0000  1.0000
2001.01.23   1.0586  1.0207    1.0256  1.0684 *
2001.01.25   1.0458  0.9739    1.0205  1.0687
2001.11.21   1.0750  1.0174    1.0760  1.0690
2002.02.25   1.1666  1.0582    1.1213  1.0951
2002.06.19   1.2351  1.1120    1.1850  1.0829
2003.01.04   1.3151  1.1931    1.2543  1.1924
2003.04.03   1.3317  1.2144    1.2484  1.1649

* The detector high voltage was raised between the 2001.01.23 and 2001.01.25 observations. For more information about the dates and effects of detector voltage changes, see Time-Dependent FUSE Calibration Effects.

** The LiF 1B correction factor for June 19, 2002, was revised upward (from 1.1736) in August of 2003 based on a new analysis that carefully excluded the worm-affected region of the spectrum. The flux-calibration file reflecting this new correction factor is called flux1b913.fit.

Here's a plot of the same information, displayed in terms of relative sensitivity (which is 1 / correction factor):

These correction factors are uncertain by a few percent, especially for segment LiF 1B, because of confusion arising from the worm.

To correct the intensity of a spectrum processed with the version 008 flux calibration, simply multiply the spectrum by the appropriate correction factor. These factors have been incorporated into versions 009-015 of the FUSE flux calibration files, which are available on the FUSE FTP site. If an observation was obtained between the effective dates of two flux-calibration files, CalFUSE interpolates between them. Otherwise, it simply adopts the most recent flux-calibration file.

Note: We do not attempt to correct spectra obtained through the MDRS and HIRS apertures for changes in the instrument sensitivity. The low throughput of these apertures, combined with the likelihood that their spectra are non-photometric, makes it difficult to obtain meaningful correction factors for these channels. We therefore employ a scale factor of 1.0 for all MDRS and HIRS observations.

The greatest uncertainties in a line or continuum flux derived from a FUSE spectrum are due to systematic effects not included in the error bars of the inverse-sensitivity curve. An estimate of the error of our flux calibration can be obtained by comparing the effective-area curves derived from various white-dwarf stars (see plots, above). Differences among the curves reflect errors in both the model atmospheres and the stellar parameters upon which they are based. In most channels, the scatter in the derived effective areas is between 2 and 4%.

Other systematic uncertainties reflect various detector flat-field effects, and their relative importance depends upon the science that one is trying to do. Errors from the moiré pattern dominate the uncertainties in the fluxes of narrow emission lines, unless the observation was obtained using an FP-split (or the equivalent was achieved via grating and mirror motions). For broad features, the moiré is unimportant, but larger-scale flat-field features are. Both of these effects are discussed in The FUSE Observer's Guide. Finally, when fitting a spectral energy distribution, the greatest uncertainty comes from the worm, which may depress the observed flux over tens of Ångstroms by 5% or more.