The spectral and spatial resolution of the COS/FUV instrument depend on Y-position on the detector. Since different lifetime positions occupy different Y positions, the spectral and spatial resolution of COS/FUV depends on lifetime position. Line Spread Functions (LSF), which describe the one-dimensional line profile of an infinitely narrow spectral line from a point source, and Cross-dispersion Spread Functions (CDSF), which describe the cross-dispersion profile of a point source on the detector, are now available on the COS website (see http://www.stsci.edu/hst/cos/performance/spectral_resolution).

As for LP2, the LSFs and CDSFs at LP3 were computed using Code V and an optical model of the COS/FUV instrument. This model includes the mid-frequency wavefront errors (MFWFE) of the optical telescope assembly (OTA). The main adjustments between LP2 and LP3 in the Code V model are the field angle and focus values. For a full description of the model at LP2, see COS ISR 2013-07 by Roman-Duval et al. In addition, the LSFs and CDSFs at LP3 were computed on a finer wavelength sampling (1 Angstrom) than had been previously used for other lifetime positions. The LSF and CDSF models at LP2 were also re-computed on the same wavelength sampling of 1 Angstrom as LP3. The LP2 LSFs and CDSFs were updated accordingly on the COS website.

Each column of the LSF tables represents a line profile at a given wavelength as a function of X pixel along dispersion. Similarly, each column of the CDSF tables represents the cross-dispersion profile at a given wavelength and as a function of Y pixel on the detector. The LSFs and CDSFs are provided for each grating and cenwave combination.

The LP3 LSFs and CDSFs were validated using observations of SMC star AzV 75 and taken as part of program 13931 ("Third COS FUV Lifetime Position: Wavelength and Resolution Calibration (LCAL2)", PI Roman-Duval). The archival STIS E140M spectrum of AzV 75 was convolved with the model LSFs for each cenwave. The resulting convolved spectra and corresponding line profiles were compared to and found to be in excellent agreement with the COS spectra from program 13931.

We note that the LSFs available for LP1 on the COS website and described in COS ISR 2010-09 by Ghavamian et al. were computed using a different, simpler ray-trace model. The COS instrument team is working on generating LP1 LSFs and CDSFs using the same Code V model as for LP2 and LP3, but the LP1 models have not been tested at this point.