Februrary 2017 STAN
In this STAN we announce the release of an updated NUV DISPTAB as well as recommendations for optimal NUV extraction box heights.
Updated COS NUV Wavelength Dispersion Solution Reference File Delivered
Due to lack of suitable calibration and/or science data, the zero-points of the dispersion solutions had not been updated with on-orbit values after the Science Mission Observatory Verification phase in 2009 for all G185M settings, and for select cenwave/stripe combinations of the G225M, G285M, and G230L gratings listed in the table below. In early 2016, a temporary wavelength dispersion reference file (DISPTAB) was delivered in which the on-orbit zero-point, or d values, from the closest stripe were copied into the appropriate columns (see the June 2016 STAN for more information). This preliminary update improved the zero-point accuracy from ~1 Å (100 km/s) to about 0.3Å (30 km/s).
To improve the wavelength calibration further, we used newly acquired STIS data to update the G225M, G285M, and G230L settings listed in the table below. All the G185M cenwaves and stripes were also updated by using newly acquired COS data. We cross-correlated the COS data to the STIS data, which is used as a wavelength reference owing to its high wavelength accuracy (~10 km/s for medium-resolution gratings). We used the the resulting offset to update the a0 zero-point term in the dispersion solution, leaving the d zero-point term in the dispersion reference file as is.
This update now provides a wavelength zero-point accuracy of 15 km/s (0.17 Å) for the M-gratings, except G185M which has an improved accuracy of ~30 km/s (0.3 Å), and 175 km/s (1.8 Å) for G230L. Further results are discussed in COS ISR 2017-02, "Updates to the COS/NUV Dispersion Solution Zero-points", that will be released in the near future.
All cenwave and stripe combinations that were affected by this NUV Wavelength Dispersion Solution Reference File change are listed:
|1786||NUVA, NUVB, NUVC|
|1817||NUVA, NUVB, NUVC|
|1835||NUVA, NUVB, NUVC|
|1850||NUVA, NUVB, NUVC|
|1864||NUVA, NUVB, NUVC|
|1882||NUVA, NUVB, NUVC|
|1890||NUVA, NUVB, NUVC|
|1900||NUVA, NUVB, NUVC|
|1913||NUVA, NUVB, NUVC|
|1921||NUVA, NUVB, NUVC|
|1941||NUVA, NUVB, NUVC|
|1953||NUVA, NUVB, NUVC|
|1971||NUVA, NUVB, NUVC|
|1986||NUVA, NUVB, NUVC|
|2010||NUVA, NUVB, NUVC|
COS NUV Extraction Box Heights for Faint Targets
The COS team studied the effect of varying the HEIGHT parameter in the XTRACTAB reference file (COS Data Handbook Section 3.7.11) for COS/NUV data. An extraction height of 8 pixels is used when estimating the signal-to-noise ratio in the COS exposure time calculator (ETC), while CalCOS uses a height of 57 pixels to create the final spectra. This is because the flux calibration for bright sources is performed using a spectrum extracted with a box large enough to encompass all the flux, while for faint sources, the maximum possible S/N is obtained by minimizing the dark rate and thus reducing the extraction height.
We studied the effect of varying the HEIGHT parameter in the XTRACTAB for targets of different brightness levels to examine whether there are benefits to lowering the value of 57 pixels used by CalCOS. We found that in order to avoid having to derive flux calibrations for each possible extraction height value, the extraction height could only be reduced to ~37 - 43 pixels. This did not improve the S/N of faint sources significantly, while possibly having a negative effect on the flux accuracy of brighter targets. We recommend PIs use a smaller box height for faint sources (i.e., when the dark rate is similar to the source count rate), but do not reduce the height below 13 pixels to ensure no significant signal is lost. To do this, PIs will need to customize an XTRACTAB, edit the HEIGHT column, and recalibrate as outlined in the COS Data Handbook Section 3.6.
Lastly, we point out that the centering of the spectral stripes may cause some flux to be left out of the extraction region when the extraction height is reduced. Due to target centering uncertainties (sometimes up to 13 pixels in the cross-dispersion direction), the B_SPEC value listed in the XTRACTAB file may not be perfectly centered on the middle of the stripe. While this will not have a significant impact on the flux accuracy in spectra with a 57-pixel height, this effect may result in significant vignetting and subsequent loss of signal for small height values. Therefore, customizing the B_SPEC values may be necessary as well.
More information about this work will be published in ISR 2017-03, “Analysis of COS NUV Extraction Box Heights” in the near future.