Technical considerations to prepare MegaCam observations

Table of contents:

  • Introduction
  • 1. Geometry
    • 1.1 Mosaic organization
    • 1.2 Wide-Field Corrector optical distortion and dithering patterns
    • 1.3 Evaluation of the image quality across the field of view
  • 2. Photometry
    • 2.1 Filter set
    • 2.2 Telescope and MegaPrime/MegaCam optics
    • 2.3 CCD quantum efficiency
    • 2.4 Narrow band filters: CFH12K filters forcing a reduced field of view
    • 2.5 AB Magnitude equation
    • 2.6 Transition from AB to Vega magnitude system
    • 2.7 Sky brightness
    • 2.8 Fringes in the i' and z' bands
    • 2.9 SNLS based photometric calibration
  • 3. Observing
    • 3.1 Immunity to bright stars scattered light


This page provides information of interest to the observer preparing MegaCam observations. Like CFH12K before it, MegaCam is operated in Queued Service Observing mode. However, in the case of MegaCam, classical observing is not offered to observers. Observers will receive their MegaCam data after pre-processing (and astrometric and photometric calibration on a per CCD basis) by the CFHT Elixir analysis system, though they may request raw data if necessary.

This page follows the structure of the various sections found in the "General Summary" table and provides further details related to the impact they may have on an observing strategy.

Starting 2015A, Megacam has a new set of broad and narrow band filters with a larger foot print than the old broad band filters, therefore allowing to observe with all 40 CCDs (see below).

The old filters (u*,g',r',i',z') are renamed with a trailing S (for "Small"), while the new filters are simply named u,g,r,i,z..

1 Geometry

1.1 Mosaic organization

Although the camera is read through all 80 amplifiers (two per detector), the data is re-organized at CFHT to splice amplifiers from the same detector into a single FITS image (a process we call "splicing"). Figure 1 shows the organization of the readout amplifiers within the focal plane. The final extension name is given on the CCD (00 to 35). After the splicing process (done on the fly - the raw data are the spliced data) all detectors have their pixel {X=1, Y=1} near the A amplifier, i.e. as if all the detectors in the mosaic had been read from the A amplifier. Note that the physical gaps in the mosaic are not represented on this figure.

Figure 1. MegaCam readout layout. North is at the top, East to the left.

1.2 Wide-Field Corrector optical distortion

There is a significant optical distortion beyond a 20 arcminutes radius and one should be aware than creating large monolithic images from a stack of dithered exposures will require a global resampling. The Terapix Data Center will provide this service to the CFHT community (the tool developed for that very function is called "swarp").

Nevertheless, depending on the scientific goals, recovering the medium (13 arcsec wide, or 70 pixels) and large (80 arcsec wide, or 425 pixels) gaps in the mosaic is not always a necessity. For that reason, three types of dithering pattern are offered in the QSO interface:

  • Large Dithering Pattern (ellipse 30'' x 180''):
    LDP1 -> LDP21 = "Dither Large Gaps"
    This in the pattern of choice to cover the largest gaps in the mosaic with 4 (and up) exposures. The pattern, as shown on Figure 2, actually becomes a very elongated ellipse as the largest vertical gaps are 6 times wider than the horizontal gaps. After proper resampling and stacking, single monolithic 1 by 1 square degree images can be created.
  • Medium Dithering Pattern (disk diameter 30''):
    DP1 -> DP21 with scaling factor 1.5 = "Dither Small Gaps"
    This is the pattern to pick if only three large stripes of sky with a one arcminute gap between them is fine to conduct the scientific program. Resampling is still a must-do here.
  • Small Dithering Pattern (disk diameter 20''):
    DP1 -> DP21 with scaling factor 1.0 = "Dither CCD"
    This is the pattern to pick if all 36 CCDs are to be processed separately. Global resampling to stack all the individual chips is not necessary as the optical distortion remains minimal at the scale of a single CCD with the applied offsets.

Figure 2. Distribution of optimally dithered exposures. (the disk becomes an ellipse for the large pattern).

1.3 Evaluation of the image quality across the field of view

The new wide-field corrector was expected to cause an image degradation from center to edge of less than 0.1 arcsecond. One should be aware than when requesting a given image quality, it is assumed as being the image quality from the center region of the mosaic. Hence, if 1.0 arcsecond is required over the whole field of view, one should request an image quality of 0.9".

We have defined image quality estimators to quantify this effect on the observed image. They rely on ratio of image quality measurements between various regions (shown on Figure 3) of the mosaic.

Figure 3. Regions of image quality analysis.

Some results are proposed here for an image taken in the 2003B semester optical configuration (300 seconds exposure in the r' band on a star field):
  IMAGE:      712976
  IQ MAP & CLASS TO MEDIAN 1-9 | *0 |+/-1|+/-2|+/-3|+/-4|+/-5|+/-6|+/-7|+/-8|+/-9|
  IQ difference in arcsec ---> <0.02<0.06<0.11<0.17<0.22<0.28<0.34<0.39<0.41< more
  |0.81|0.77|0.74|0.74|0.72|0.73|0.76|0.82|0.88|    |+1|*0|*0|-1|-1|-1|*0|+2|+3|
  |0.80|0.75|0.69|0.67|0.67|0.68|0.70|0.74|0.80|    |+1|*0|-2|-2|-2|-2|-2|-1|+1|
  |0.83|0.77|0.71|0.68|0.68|0.70|0.72|0.76|0.82|    |+2|*0|-1|-2|-2|-2|-1|*0|+2|
  |0.90|0.84|0.80|0.77|0.77|0.77|0.80|0.85|0.93|    |+3|+2|+1|*0|*0|*0|+1|+2|+4|
  CENTER     = 0.69"
  OUTER RING = 0.79"
  RADIUS R = 1.144 [ center / (top+left+bottom+right) ]
  PISTON X = 1.018 [ right / left ]
  PISTON Y = 0.937 [ top / bottom ]
MegaCam is focused (and soon autofocused) for the position that optimizes the image quality over the entire mosaic (optimal focus at the center would cause greater degradation on the edges).

2 Photometry

2.1 Filter set

MegaPrime currently has five ("old") broad-band filters, constructed by SAGEM: u*, g', r', i', z', now renamed uS, gS, rS, iS, zS . Except for u* these filters were designed to match the SDSS filters as closely as possible. Since Mauna Kea has signficantly less UV extinction than the SDSS Apache Point (3000 m) site, the u* filter is designed to take advantage of the improved UV capabilities of CFHT + MegaPrime at Mauna Kea. For this reason, we use the name u* for this filter to highlight the difference between this filter and the SDSS u' filter.

Starting 2015A, these (old) filters are getting replaced by a new set of broad and narrow band filters , simply named u, g, r, i, z for the new broad band filters, and Ha, HaOFF, OIII, OIIIOFF and CaHK for the narrow band filters. Due to their larger foot print, these filters allow a full (unvignetted) illumination of the 40 CCDs of the mosaic . The plot below (Figure 4) shows the (mean) measured transmission curves of the 5 new SDSS-like filters (solid lines) as measured by their respective manufactures, Asahi for u, Materion for the others. The old filter mean transmissions (as measured at CFHT) are shown in dashed lines for comparison.

The new filter transmissions are shown in solid, colored lines. Each filter was scanned 17 times by the manufacturer at different places, the mean transmission is reported in the figure. Note that the new z filter has a red cutoff and a redder blue cutoff, which explains the lower zero point compared to the old z filter. The full manufacturer scans are available here.

Figure 4. MegaCam filter set transmission and average CCD quantum efficiency. Dashed lines are for the old (S) filter set, while solid lines are for the new filter set. QE is in black, dot-dashed.

The following table gives basic filter charateristics for the broad band filters, old and new. The measured mean bandwidth and mean transmission over that bandwidth are derived from scans performed at CFHT for each filter.

MegaPrime (old and new) Broad Band Filters Characteristics
Filter uS gS rS iS zS u g r i z
Central wavelength (nm) 375 487 630 770 n/a 355 475 640 776 925
Wavelength range (nm) at 50% 337 - 411 416 - 559 569 - 693 684 - 843 827 - ... 311 - 397 398 - 552 566 - 714 699 - 854 849 - 1002
Bandwidth (nm) 74 143 124 159 n/a 86 154 148 155 153
Mean transmission (%) 68.5 78.4 82.1 90.7 90.0 89.3 94.4 97.0 96.2 96.2

Each old filter was scanned 10 times at a range of radii and the resulting measurement have been averaged to generate the curves below. Newer data from actual filters scans are available in ASCII format here. The original scans by the manufacturer (SAGEM), which differ slightly from the CFHT scans *but which are believed to be more accurate*, can be found here.

The full manufacturer scans of the current filters are available here.

2.2 Telescope and MegaPrime/MegaCam optics

The CFHT is equiped with a 3.58 diameter mirror and MegaPrime causes an obstruction which is larger than its central hole. After adding the footprint of the primary focus and the four spider branches supporting it, the final collecting area of the primary mirror is 80,216 square centimeters (each spider branch is 5 cm wide, over the area of the primary mirror, 3 are 111.5 cm long and one is 81.5 long, the prime focus can be modeled by two circles of 67.5 cm radius each with a central bridge between their centers of 30 cm).

The reflectivity model for the primary mirror is of a freshly coated, bare aluminum. The curve is shown on figure 5 and the data used to generate that plot are available here.

The transmission model of the Wide-Field Corrector and the camera window is derived from combining all the transmission and diffusion effects of the various glass and coatings. The curve is shown on figure 5 and the data used to generate that plot are available here.

Figure 5. Primary mirror and MegaPrime+MegaCam optics (minus filters) transmission.

2.3 CCD quantum efficiency

The following table is the average measurement of the quantum efficiency (QE) over the 40 CCDs populating the MegaCam focal plane. The dispersion at each measured wavelength is also provided and shows that there are variations from chip to chip; in consequence these numbers must be considered only to describe an average behavior of the detectors.

The measurements were performed individually on each chip at CEA.

MegaCam CCDs average QE
Wavelength (nm) 3504004505005506006507007508008509009501000
QE (%) 48.978.385.784.782.880.877.871.861.949.437.624.812.04.3
Standard dev. (%)

The data from this table are available in ASCII format here.

A more precise table with a point every 10 nm is available here. Note that these are results from a simulation for a given set of CCD characteristics that match best the table of real measurements provided above.

2.4 Narrow band filters: old CFH12K filters (forcing a reduced field of view) and new, full mosaic filters

A set of CFH12K narrow band filters (HaS, HaOFFS, TiOS, CNS, HbS, HbOFFS, OIII) are still available. Their characteristics are summarized in the table below.

MegaPrime's CFH12K Narrow-Band Filters Set
Central wavelength (nm) 658.4 645.3 777.7 812.0 487.4 478.8 504.8
Bandwidth (nm) 7.6 9.0 18.4 16.1 8.1 7.8 8.6
Peak transmission (%) 96 94 92 95 88 78 90

These narrow band filters cover 14 chips at the center of the mosaic (10 to 16 and 19 to 25) though the external ones will have their external sides strongly vignetted. The resulting fielf of view is 42 by 28 square arcminutes. More information on this filter set can be found here.

Starting with 2015A, a new set of narrow band filters, covering all 40 chips in the mosaic, are available for observations. Due to the lack of slots in the instrument jukebox, users are strongly encouraged to use these new filters in place of old CFH12K ones when possible.

Note that the ETC is now available also for the new filters, however, due to the current lack of statistics, values for sky background and airmass extinction terms were deduced from their broad band counterparts. Photometric calibration is not yet available for these filters, however, all necessary standard star observations have been done, which will allow a photometric calibration to be computed for all data taken with these filters (starting from 2015A) in the near future. A summary of their properties are given in the Table below. Detailed transmission data are available here.

MegaPrime's new, full mosaic Narrow-Band Filters Set
Central wavelength (nm) 659.0 671.8 500.6 510.5 395.1
Bandwidth (nm) 10.4 10.7 9.9 9.5 9.6
Peak transmission (%) 93 95 91 93 91

2.5 AB Magnitude equation

AB instrumental magnitude equation
In electrons m ~ Zp[e-/sec] - 2.5log(Gain[e-/ADU]) - k(airmass - 1)
In ADUs m ~ Zp[ADU/sec] - k(airmass - 1) (Elixir data)
Gain (e- / ADU) 1.62 (CCD00)
Filters uS, gS, rS, iS, zS, u, g, r, i, z
Zero points (e- / sec) at 1 airmass (Zp) 25.74, 27.00, 26.50, 26.38, 25.34, 25.74, 27.11, 26.74, 26.22, 25.02
Zero points (ADU / sec) at 1 airmass (Zp) 25.22, 26.48, 25.98, 25.86, 24.82, 25.22, 26.59, 26.22, 25.70, 24.50 (Elixir data)
Airmass term (k) 0.350, 0.150, 0.100, 0.040, 0.030, 0.350, 0.150, 0.100, 0.040, 0.030

The quantum efficiency and the gain vary slightly from CCD to CCD though they are both fairly uniform over the whole mosaic (average gain is 1.67 e-/ADU with a 0.2 dispersion) and the dispersion on QE is presented in the table in 2.3.

The zero points advertised on the MegaPrime pages come from photometric frames processed by Elixir. The flat-fielding step takes care of making the zero point uniform over the entire 36 CCDs (40 for new filters) of the mosaic by applying a multiplicative factor scaled on CCD00, e.g. the response is normalized to CCD00. The result is a flat looking image with a unifom sky level but with a detection limit changing slightly from CCD to CCD due to their intrinsic differences in quantum efficiency and read noise.

When applying a zero point to some Elixir processed data, one should use the gain of 1.62 by default, hence apply the equation "In ADUs" provided in the table above.

2.6 Transition from AB to Vega magnitude system

The table below allows a translation from MegaCam AB to Vega Magnitudes.

AB to Vega magnitude system offsets
Filter uS gS rS iS zS u g r i z
Offset -0.346 +0.092 -0.171 -0.401 -0.554 -0.621 +0.087 -0.181 -0.392 -0.526

2.7 Sky brightness

Sky brightness in the various MegaCam filters was derived from the 2003A semester data, and specially from engineering nights where the sky brightness had been precisely monitored in respect to the airmass, and then later in respect of the Moon phase.

The measurements were derived from data obtained exclusively in photometric conditions. The combination of cirrus and the Moon in the sky is basically impossible to predict, but the effect will be less in i and z than in u, g and r.

It is remarkable that the sky retains the same brightness at zenith in the i and z band no matter what the phase of the Moon is (we tested between 0% and 70%). This demonstrate that MegaPrime is a prime grey time instrument. However the sky brightness in these two bands depend greatly on the airmass.

2.8 Fringes in the i and z bands

The fringing level in the i band is reasonably low (6%) and is fairly well corrected by Elixir, though we noticed that we may now be facing residuals due to variations in the sky spectrum during the course of a run or even a night. The fringing level in the z band is high: 15%.

MegaCam is not a great instrument to observe in the z band as the CCDs sensitivity is low at those frequencies and the sky background is very high resulting in very strong fringing.

2.9 SNLS based photometric calibration

Starting with semester 2015A, the elixir zero points are based on a comparison between observations of the SNLS deep fields obtained during each run, with the catalog of absolutely calibrated SNLS "tertiary standards" (see Betoule et al. 2013). Specific color terms have been computed to relate the new MegaCam wide band filters photometric systems to the old wide band systems using synthetic magnitudes on stellar libraries, and successfully compared to empirical measurements of these color terms (courtesy S. Gwyn.). The latter are implemented in elixir, and the photometric equations and keywords described below are included in the FITS headers of the processed data. In the following, variables in capitals correspond to FITS keywords.

If we define the instrumental magnitude as

 m = -2.5*LOG(DN)+2.5*LOG(EXPTIME), 
where DN is the total count number of a source (in ADU), and is the corresponding exposure time in seconds, then the calibrated (AB, SNLS) magnitude m_snls, based on MegaCam old broad band filter ugriz set, is related to this instrumental magnitude, in a given band, by the following equation:
  m_snls = m + PHOT_C + PHOT_K*(AIRMASS-1) +PHOT_X*(m_snls(PHOT_C1)-m_snls(PHOT_C2))
where PHOT_C is the value of the zero point in that band (current value for the camera run), PHOT_K is the extinction coefficient in that band, AIRMASS is the observation airmass, PHOT_X is the slope of the color correction, and PHOT_C1 and PHOT_C2 are the (SNLS, old Megacam broad band) bands used to define the color term. This equation assumes that these color terms are known.

If one just wants the AB magnitude in the filter used for the observation, then PHOT_X can be put to zero in the equation above. Of course, if the filter used is one of the old broad band filters (used to define the SNLS photometric system), then the corresponding PHOT_X are naturally equal to zero. The following table gives the relevant information for the SNLS photometric system, and also for the SDSS 2.5m photometric system for reference. The linear color terms PHOT_X (SNLS) have been adjusted to color-color diagrams obtained from matching catalogs obtained on the SNLS deep fields, and by matching catalogs to the SDSS catalog for PHOT_X (SDSS) (courtesy S. Gwyn, see this page ).

Photometric calibration FITS keywords and values for the old and new broad band filters
Filter uS gS rS iS zS u g r i z
PHOT_K -0.35 -0.15 -0.10 -0.04 -0.03 -0.35 -0.15 -0.10 -0.04 -0.03
PHOT_X (SNLS) 0.0 0.0 0.0 0.0 0.0 -0.232 -0.087 +0.065 +0.068 +0.065
PHOT_C1 (SNLS) - - - - - uS gS gS rS iS
PHOT_C2 (SNLS) - - - - - gS rS rS iS zS
PHOT_X (SDSS) +0.241 +0.153 +0.024 +0.003 -0.074 +0.086 +0.067 +0.087 +0.089 +0.045
PHOT_C1 (SDSS) u_sdss g_sdss g_sdss r_sdss i_sdss u_sdss g_sdss g_sdss r_sdss i_sdss
PHOT_C2 (SDSS) g_sdss r_sdss r_sdss i_sdss z_sdss g_sdss r_sdss r_sdss i_sdss z_sdss

These zero point estimates have been compared to other estimates based on the observation of spectrophotometric (CALSPEC) HST standards and their corresponding synthetic magnitudes, and have been found to be in reasonable agreement. The same HST standards and synthetic magnitudes are used to give a rough estimate of zero points for the new narrow band filters. In addition, the SNLS code (courtesy Marc Betoule) to produce superflat (photometric grid) solutions has been adapted to the new focal plane layout with 40 CCDs, and was used to produce photometric grid corrections on all new wideband filters, as well as the narrow band filters as soon as data become available. See "MegaCam photometric calibration" for details.

3 Observing

3.1 Immunity to bright stars scattered light

There are three possible ways for contamination by bright stars:
  • Nearby off-field bright stars:
    As shown on the "Off-field bright stars immunity" page, this kind of contamination is not an issue anymore with MegaCam.
  • Reflection off the wire bonds on the focal plane:
    There was a concern that the exposed golden pads in the middle of the mosaic (the part where the CCD signal are connected from the substrate to the connectors) would cause some problem. No such thing as they sit below the sensitive surface and the light can not in consequence cause a contamination. A very bright star was moved around in that area and no contamination could be noticed.
  • Reflective halos:
    They are caused by the low level reflections from the CCD surface back to the optics and then back to the CCD. If the star is very bright, then it ends up becoming an obvious feature. They are shown on Figure 6. They can not be avoided and pretty much ruin any signal within the area of the halo, typically 7 arcminutes in diameter. While this effect was most prominent on the CFH12K in the B filter, here it is present in all filters, a result of the different anti-reflection coating. The example on Figure 6 shows a uSgSrS "true-color" image of the NGC 2244 cluster, the bright star on the upper right is V=6.0 mag (USNO) and the one on the lower left is V=6.8 mag (USNO).

Figure 6. Reflection halos in MegaPrime due to the optics (NGC 2244 cluster).

Preliminary preparation of the pointings can be conducted using the "MegaPrime Field Mapping" tool.