On 2005-Apr-04, a number of new photometric throughput (pht) and time-dependent sensitivity (tds) reference files were delivered that provide significant improvements to the absolute flux calibration for 1D extractions of STIS low-dispersion, PRISM, and echelle modes.
Throughputs for G140L and G230L have increased by about 2% at most wavelengths (thereby giving extracted fluxes that are lower by about 2%), with somewhat larger changes at the shortest wavelengths of each grating and in the vicinity of the Lyman-alpha line.
Throughputs for CCD gratings have not changed with this release. Once corrections for time-dependent-sensitivity (TDS) changes and charge transfer inefficiencies (CTI) are made, the absolute flux accuracy for all STIS low-dispersion observations done using the 52X2 aperture should have relative accuracies of 3% or better and an absolute accuracy of better than 5% (see discussion in Bohlin et al. 2000, AJ 120,437).
For E140H and E140M the new sensitivities are typically 3 to 20% higher, with substantially larger changes at the very shortest wavelengths near 1140 Angstroms. E230H and E230M sensitivities are changed by less than 10% at most wavelengths. TDS corrections are now included for all echelle observations, and for echelle observations using primary wavelength settings and the 0.2X0.2 aperture, typical flux accuracy should be ~5%, although isolated individual observations may occasionally be sufficiently miscentered in the 0.2X0.2 aperture to have a more noticeable flux loss.
The new NUV MAMA PRISM throughput curves differ from the previous ones by up to 15%, and extracted x1d PRISM fluxes now also include TDS corrections. The absolute accuracy of these observations should now be about 5%. A new and more accurate photometric correction table (pct file) which corrects the flux for the vertical extent of the chosen extraction region has also been delivered. An error in the previous prism pct file, which could cause errors in 1D-extracted fluxes of up to 25% for slitless PRISM data, has been corrected.
Note that there are a number of issues for the calibration of
extracted fluxes for point source spectra that have yet to be fully
- For the CCD gratings there is some vignetting as a function of position along the slit that is not yet taken into account. This can cause wavelength dependent flux errors of up to 5% for spectra not placed near the central rows of the detector.
- The current calibration software assumes that the grating and aperture throughputs can be determined separately and then multiplied in any combination. This is apparently not true for the G430L and G750L gratings, which contain extra Lyot stops that appear to make the grating throughputs aperture dependent. For the 52X0.2 and 52x0.2E1 apertures, this can lead to flux errors of as much as 10%, and can create discontinuities of 2 to 6% in the overlap regions between different gratings.
- At the shortest wavelengths of each first-order grating, the sensitivity function varies rapidly near the blocking filter cutoff, and small wavelength misalignments can cause flux errors of order 10%.
- The secondary wavelength settings for the echelles still use pre-launch estimates for the blaze function and do not yet include corrections for shifts of the blaze function with respect to the wavelength scale between different observations. These effects can result in significant distortion of the relative fluxes within each spectral order. However, this new pht and tds file delivery will yield improved mean fluxes even for these secondary settings.
- The implemented sensitivity curves for the first-order medium resolution settings are of varying quality and have not yet been fully updated using all available standard star observations. While most first order medium resolution central wavelength settings are accurate to 3% or better, a few show anomalies at the 5 to 10% level, including a few that may mimic spectral features (e.g., the G230MB 2135 setting has a spurious 20 Angstrom wide 5% deep dip in the calibrated fluxes). Those wavelengths defined as secondary settings and E140M observations taken after the 1999-Mar shift of FUV-MAMA spectra to a position with lower dark current are especially in need of improvement.
We hope to address most of the above issues in the next few months and deliver a final flux calibration for 1D spectral extractions by the end of the year.