COS Instrument Handbook for Cycle 24

help@stsci.edu

5.9	COS Spectroscopy of Extended Sources

COS spectra of extended objects can have significantly lower resolution than those of point sources, depending on the spatial distribution of the source. For example, measurements of Lyman-α airglow lines, which uniformly fill the COS aperture, show R = 1450 for G130M and R = 165 for G140L. Filled-aperture observations of SNR N132D confirm R ~ 1500 for both FUV M gratings (France et al. 2009, ApJL, 707, L27). In the NUV the situation is much worse, because a source that fills the aperture will lead to cross-contamination among the three spectral stripes on the MAMA detector.

A similar situation arises when multiple point sources fall within the aperture. COS was designed to resolve two point sources separated by 1 arcsec in the cross-dispersion direction, but on-orbit measurements of the XD profile reveal that such observations are possible only for selected grating and central-wavelength settings (Section 5.1.9). Note that light from a point source falling more than 0.4 arcsec from the center of the PSA will be attenuated (Section 8.8.2).

While the performance of COS when observing extended targets was never optimal, the extended target flux calibration was still reasonably accurate up to the COS/FUV move to Lifetime Position 3 (LP3) in February 2015. This is however not the case at LP3 due to its location, close to the gain-sagged Lifetime Position 1 (LP1) regions, and the new two-zone extraction algorithm required by the small LP1-LP3 separation (see February 2015 COS STAN). 

For extended targets larger than ~0.6” (FWHM), the spatial distribution of incoming light on the COS/FUV detector is significantly wider than that of a point source. As a result, the light distribution of extended targets may overlap low-gain regions at LP1, located 2.5” above LP3, and may incur significant flux loss.

In addition, the two-zone extraction algorithm implemented to support LP3 science observations extracts counts in 2D profiles that tightly follow the optical footprint of the light from a point source on the detector, thus reducing the number of dark counts and avoiding large chunks of spectra from being discarded when bad or LP1 sagged pixels overlap with the wings of those profiles. This algorithm therefore provides accurate calibration for point sources, but the flux calibration of extended targets (i.e., with FWHM larger than ~0.6”) may be significantly off because the spatial distribution of their flux on the detector is wider than that of a point source.

Combined with the aperture vignetting, these effects may result in an unreliable flux calibration for COS/FUV/LP3 observations of extended targets (larger than 0.6” FWHM).

Starting with Cycle 23 and continuing in Cycle 24, COS observers are required to define their targets as EXTENDED = “YES” or “NO” at the target description level. This new field is required for COS observations only. This change is intended to provide an indication as to whether the source extent might be large enough to affect the reliability of the default pipeline extraction which is currently optimized for point source targets. In the future this keyword may also be used to allow the pipeline to use an alternate extraction procedure better suited to extended sources. Both COS FUV and NUV observations of targets with FWHM larger than about 0.6 arcsec or radius larger than 0.35 arcsec should be considered as extended. “YES” should be selected for such targets and “NO” for targets of smaller angular extent. The field should be left blank if the target is not being observed by COS. If this new target-level description keyword is not provided, APT will trigger a warning message asking the PI to define it, and any proposals submitted without a value specified for all COS targets will be sent back to the PI to be corrected. See the COS STAN articles from February and March 2015 for further details. Note that the old exposure-level EXTENDED keyword, set via the “optional parameters” prior to Cycle 23, is still present in APT but has been deprecated. As a result, the exposure-level EXTENDED keyword should no longer be selected for any exposure in Cycle 23 or later.