Technical considerations to prepare WIRCam observations

Introduction

This page provides information of interest to the observer preparing WIRCam observations. Like Megacam, WIRCam is operated in Queued Service Observing mode. Classical observing is not offered to observers. Observers will receive their WIRCam data after pre-processing (and astrometric and photometric calibration on a per chip basis) by the CFHT Elixir analysis system, though they may request raw data if necessary.

Detector Controller

The WIRCam detectors are read out using two SDSU controllers (2 chips per controller). Each HAWAII-2RG array is read using 32 amplifiers. The total readout overhead was brought down to 4.8 seconds, BUT the overhead charged to PIs in their phase II proposal is 10sec (bear in mind that the telescope offsets (7sec - unguided, 14sec - guided) are not charged to PIs. This more than compensates for the charged readout overhead of 10sec).

FITS files

WIRCam images are stored as multi-extension FITS (MEF) files containing 4 extensions (1 per detector). Each extension may contain a cube of images (if you use multiple exposures at each telescope position or/and micro-dithering), or one single image (if you don't). The data are stored as 16-bit signed integers, giving a minimum size for a WIRCam file of slightly above 32 MB. We limit the maximum size to 2 GB. Recent versions of ds9 (>4.?.?) support viewing cubes within a MEF.

Micro-dithering

The WIRCam pixel scale is 0.3"/pixel, so that individual images are only Nyquist-sampled for seeings worse than
0.3 * 2 = 0.6" (i.e. half of the time, as the median CFHT seeing is ~0.65"). To provide well sampled images under good seeing conditions, WIRCam implements optional micro-dithering (or micro-stepping) of its images. In this mode the Image Stabilizer Unit (ISU) is used to offset successive images by 0.5 pixel on a 2x2 XY pattern. The individual images remain undersampled (for good seeing), but a composite image with finer sampling (0.15") can be constructed from the set of micro-stepped images, using data processing techniques such as simple interlacing, drizzling , or optimal Fourier-plane combination. In the ideal case where the 0.5 pixel offsets are executed exactly, all 3 techniques converge to simple interlacing. A number of misconceptions on micro-dithering have in the past appeared in observing proposals, so we will try here to clarify what it is, and perhaps more importantly what it is not.

Like the more familiar "macro-"dithering used in most optical and IR observing sequences, micro-dithering tries to evenly spread the fractional part of the offsets over the area of a pixel, as needed for good reconstruction (e.g. Lauer (1998)). By controlling the offsets more accurately, it achieves that particular goal much faster than feasible with the random fractional offsets produced by telescope motion: O(20) random offsets are needed to achieve as uniform a pixel coverage as a 2x2 micro-stepping pattern.

Because the offsets are so small, micro-dithering does no provide any measurement of the sky, so some amount of larger scale dithering (or telescope nodding) is always needed in addition. On the other hand micro-dithering is essentially instantaneous (faster than the array readout, and done in parallel), while larger scale dithering incurs the significant overhead of a telescope offset. If you are going to obtain multiple guided exposures at each offset position to reduce overheads, micro-dithering comes for free and there is then little reason not to use it.

Micro-dithering requires guiding, as it uses the guiding stars to control the offset produced by the ISU. Guiding needs to be optimal on more than 1 star so this is why we limit the use of micro-dithering to wide-band filters (Y,J,H,Ks).

Micro-dithering is not expected to improve images which are already Nyquist-sampled because the current seeing is bad. It therefore brings no gain if the seeing is worse than ~0.8". On the other hand it does not harm those images either, and could help in the (unlikely) event that the data are actually obtained with significantly better seeing than requested during Phase-2.

For a given exposure time, micro-dithered images and non-micro-dithered images have identical limiting magnitudes.

Optical distortion

The measured optical distortion of WIRCam is <0.8% (maximal in the corners of the field) or ~20 pixels. As a consequence, dithered full-field images need to be be interpolated to a common grid before stacking whenever they involve offsets larger than ~60 pixels (a 60 pixels offset gives a 0.5 pixel misalignment in the corners).

Filter set

The WIRCam filter collection currently includes 4 broad-band filters (Y, J, H, Ks) and 9 narrow-band filters (Low OH- 1, Low OH- 2, CH4 On, CH4 Off, W, H2 v=1-0 S(1), K Continuum, Bracket Gamma, CO).

NOTE: the WIRCam filter wheel has space for only 8 filters, and the set of installed filters will be fixed for a given semester. The filters which are offered for the next semester will be selected after the national Time Allocation Committees have evaluated the proposals, and conflicts if any will be arbitrated by the SAC. Consult the WIRCam filter table for quantitative information on the wavelengths and bandwidths of the filters.

Detector quantum efficiency

The quoted quantum efficiencies by Rockwell Scientific in J and K are between 70 and 85%. From on-sky tests, it appears that the four science chips have the same efficiency to within 10%. More can be learned on the throughput here.