Slitmask Design and Observation Technique Notes
The DEEP2 DEIMOS pipeline is NOT a fully generalized DEIMOS data pipeline. The code was developed with the goal of reducing DEEP2 observations. For this reason, the pipeline has a couple of noteworthy limitations. Also, there are certain slitmask design choices that can be made to optimize the power of the pipeline. In these terse notes, we will try to explain these pipeline limitations and slitmask design recommendations.
Recall that slitmasks must be submitted to the UCO-Lick shops at least two weeks prior to your observing run. At present, slitmaks are NOT milled at Keck. Instead, all masks are milled at the UCO Lick shops and a 2-week lead time is needed. For more information on the milling process and protocol, refer to the Keck website.
DITHERING BETWEEN FRAMES
The DEEP2 pipeline is NOT able to handle dithering between science frames. That is, the code assumes that the positions of objects do not vary from exposure to exposure. Due to the very minimal fringing in the DEIMOS CCDs and the very short slit sizes employed by DEEP2 (typically slit lengths of ~5"), there is a reduced need and ability to dither.
If dithering is necessary, then the science frames could be reduced individually using the pipeline and combined by the user. However, reductions of a mask with only a single science frame would suffer from poor cosmic-ray rejection. Reasonably robust cosmic-ray rejection can be obtained with 2 science frames. For this reason it is recommended that if one wishes to dither and wishes to use the pipeline, then pairs of science frames should be taken at each point in the dither pattern and these frames should be taken back-to-back so as to minimize temporal variations in the sky. One last thing to consider in choosing to dither or not is that the b-spline sky-modeling employed by the pipeline works better if the slits are slightly tilted. The vertical slits required for dithering would not be ideal (see dicussion below regarding tilted slitlets).
When choosing the orientation or position angle (PA) of your slitlets, it is best if the slitlets are NOT oriented at the same PA as the mask. That is, it is best if the slitlet PA on the sky is not the same as the mask PA on the sky. If a slitlet is aligned vertically, then the 2-d bspline sky model that the pipeline uses in sky-subtraction becomes poorly sampled. For vertical slits, the 2-d wavelength solution will be constant in the spatial direction. That is, for a slit with pixel dimensions 4096x100 there will only be 4096 unique lambda values. While if a slit is slightly tilted with respect to the mask PA then there is a small spatial gradient in the 2-d lambda solution so that the wavelength sampling is higher. Thus, to achieve optimal sky-subtraction, it is recommended that slits be tilted by at least 5 degrees relative to the mask PA. Note that for shorter slits (< 5" in length) where there is less real-estate from which to construct the sky model the importance of tilting the slits is greater.
It is recommended that slit widths be set at ~1". Accumulation of dust on the slitmask causes the typical slit function to be poorly defined at the edges of the slit. For very narrow slitlets this problem becomes more prominent as the dust accumulation accounts for a larger portion of the slit real-estate.
The DEEP2 DEIMOS pipeline was not designed for very long slits. For this reason, it is recommended that slits be restricted to < 30-60". For longer slits, errors in determinations of the slitlet tilts can cause wavelength solution errors that result in poor sky-modeling and sky-subtraction. Additionally, to achieve good sky-subtraction, it is also recommended that the slit lengths by long enough so that there is > 2.5" of slit real-estate away from the object from which to model the sky emission.