The twilight flat-fields sequencer allows the user to efficiently acquire the maximum number of twilight flat-fields within a limited amount of time. As soon as the proper exposure time is determined by the user with the flux command for a given filter, the sequence can be started. Upon sky brightness evolution laws established for CFH12K for each of the different filters, the subsequent exposure times will be adjusted so that the flux level on all the images of the sequence will be uniform.

The CFH12K readout time is one minute (including pre-exposure setup, detector cleanup and full mosaic readout), the write to disk time is not included since it is a background task: a new exposure can be started while the previous is still being written to disk.

Note that the sky brightness evolution laws are for nautical twilight only (astronomical twilight is defined as the sun being 18 degrees below the horizon, nautical twilight as the sun being 12 degrees below the horizon). The total duration of twilight varies slightly from summer to winter but remains very close to 81 minutes (start/end of astronomical twilights time to sunrise/sunset time). To find out when to start the twilight flat-field sequence, refer to the sky calendar at the following address for the current day of observation: http://www.cfht.hawaii.edu/Temporal/SkyCalendars/ and find the fields "twi. end" and "twi. beg" in the Sun columns.

It is possible to take exposures through various filters for about 30 minutes, hence with a 25-minute brightness limit from start of nautical twilight, this sets a 55 minute limit from the astronomical twilight to start/stop acquiring flat-fields.

At dusk:
-> subtract 55 minutes from "twi. end" in the sky calendar to find the time where you can start acquiring twilight flat-fields (sky still pretty bright, good for the narrow band filters, Z & I filters).
-> subtract 25 minutes from "twi. end" to find the time where you should stop acquiring twilight flat-fields (the brightness law changes slope then and the sky is "dark"). It is then time to go to the field of interest and start a focus sequence.

At dawn:
-> add 25 minutes to "twi. beg" to find the time where you can start acquiring twilight flat-fields (sky still pretty dark, good for the B & V filters).
-> add 55 minutes to "twi. beg" to find the time where you should stop acquiring twilight flat-fields (the sky is too bright).

Example: September 20, 2000:
"twi. end"=19:33 -> start flat-fields at 18:38 and stop them at 19:08.
"twi. beg"=04:57 -> start flat-fields at 05:22 and stop them at 05:52.

If the reported sky flux in the image does not seem appropriate, the observer can break the sequence and restart it with a revised exposure time.

Note that you can conduct scientific observations without much impact for 20 minutes before the end of astronomical twilight at dusk, and after the beginning of astronomical twilight at dawn as listed in the Sky Calendar table. This still leaves, for example, about 5 minutes to return the telescope to zenith and be ready for the dawn twilight flat-fields.

The exposure time in the sequence will increase during dusk flat-fields, and decrease during dawn flat-fields. At the end of the sequence, the user can pick a new filter if the the sky is not yet too dark or too bright and start again by adjusting the initial exposure time first with the flux function. Since the daytime and early twilight sky is darker towards the red wavelengths, always start at dusk with longer wavelength filters (i.e. Z, I, R in that order) and continue with shorter wavelength filters (i.e. V, B in that order) and vice-versa at dawn (B, V then R, I, Z in that order). The narrow band filter twilight flat-fields (Ha, HaOFF, TiO, CN) should be taken when the sky is very bright (still, the sun has to be below the horizon since CFH12K on CFHT is very sensitive). It is possible to take up to 7 exposures for each of two filters at either dusk or dawn.

For the bands I & Z where the night sky will produce interference patterns on the CCDs (fringes), it is crucial to take twilight flat-fields when the sky is still very bright such that the contribution from the OH lines (which cause the fringing) is still largely dominated by the twilight sky continuum.

Typical flat-field flux levels suitable for the different filters are: B=5,000 ADUs, V=10,000 ADUs, R=15,000 ADUs, I=20,000 ADUs. The CFH12K detectors are linear to within 1% up to 55,000 ADUs. Exposure times less than 2 seconds are not advised because the shutter ballistic could affect the large scale structures of the flat-field. Read the important note in the flux command description regarding the large discrepancies in the sky level depending which CCD is looked at.

In the early phase of dawn twilights, the flux command is not appropriate since the exposure time would be too long. It is advised to refer to the time the flat-fields were taken at dusk after the Sun set and tune an exposure time in a symmetrical way vs. the time the Sun will rise. For example, if the Sun set at 6:20PM and you took a 120 seconds V frame at 6:50PM, you can predict that with a Sun rising at 6:00AM, you can take a 120 seconds exposure at 5:30 AM. After the very first exposure, you have information on the sky level of that exposure and you can then abort or break the sequence if needed to tune the exposure time up or down.

After taking a sequence in a filter and moving to another filter, a rule of thumb is to start the sequence with the exposure time of the last exposure taken with the previous filter. This usually gives a good match of flux ratio with the recommended values for the various filters (as long as your last exposure was properly exposed, otherwise the flux ought to be used.

There are different ways one can run the telescope when taking twilight flat-fields: leaving the telescope parked at zenith presents the disadvantage that numerous stripes due to drifting objects will be present along the X axis of the mosaic, making it harder to analyze the sky background. If the telescope tracks, one will get the objects at the same place on the different exposures. The solution is to have the telescope tracking and having the telescope return to zenith between each exposure during the readout, hence the field is different in each exposure. The twilight flat-fields sequencer has been programmed to automatically execute the slew back to zenith. When the sequence begins, a confirmation window will appear and ask you to remind the Observing Assistant to allow these slews.

To keep the contribution from the objects in the field low, the z focus must be set to a value close to the nominal value for that particular filter. Large doughnuts due to bad focusing are harder to erase when combining the frames.

Telescope activities (slews, z focus change) do not impact the camera readout, i.e. the signal is not affected. The flat-field sequence parameters are: the initial exposure time, the number of exposures to take during the sequence, and the filter. The system automatically determines if this is a "dusk" or "dawn" twilight sequence. A "Start sequence" button controls the sequence execution.
CLI= tf12k 13 7 auto B

This tool only works for the B, V, R, I and Ha filters so far.

To run the sequence from the GUI:

• Adjust the number of exposures you wish to take ("Number of exposures" field).
• Have the exposure time properly set based on the feedback from the "flux" command.
• Launch the sequence with the "Start sequence" button.
• Monitor the flux on the exposures as they get written to disk.
• Break the sequence if necessary to reduce or increase the new exposure time.

The CFH12K Focus Interface form provides an interface to focus the camera. The sequence consists in taking a set of exposures at different prime focus bonnette z (vertical axis) values and offsetting the telescope between each one. At the end of the sequence, the camera is read out (only the top half of each CCD is read, i.e. the area closer to the center of the field). This is done in multi-exposure mode where the shutter opens several times on a single frame. To differentiate the first from the last exposure, a double offset (10" vs. 5") is applied for the last exposure (this is illustrated at the top of the form). Depending on which CCD you look at, the last exposure might be up or down (top for CCD[00,01,02,03,04,05], bottom for CCD[06,07,08,09,10,11]).

The parameters defining the focus sequence are the number of exposures to be taken at a given exposure time (defined in the window, 10 seconds is a minimum to ensure that the PSF is uniformly convolved by the seeing), the central z value around which the sequence will evolve and the step in z. Steps of 0.2 are handy to define what the correct z value is about, steps of 0.1 are enough to define a precise focus for seeing higher than 0.8". Steps of 0.05 are strongly recommended if seeing is better than 0.8". In the course of the night, as focus changes with temperature and other conditions, short sequences of five positions which can take less than two minutes are of great benefit to ensure the best quality of your data.
CLI= focus 15 1.6 7 0.1 (focus sequence)

For example, a sequence of five 15-second exposures around z=1.3 with 0.1 steps will take the exposures at z = 1.1, 1.2, 1.3, 1.4, 1.5. The sequence will always start with the lower z by first going a bit lower to avoid any hysteresis effect on the bonnette vertical motion. The Start button will start the sequence (asking for confirmation first), the Abort button on the main GUI will abort it (asking for confirmation first).

As the focus sequence runs, the actual z value set for each exposure will be displayed in the lower feedback window.

The focus file will be written to disk with a "x" extension (refer to the main window for the current filename). Make sure you are looking at a star and not a compact distant galaxy. Also make sure you are not using a saturated star or one close to saturation: don't select one with a peak intensity above 40,000 ADUs to be conservative. It is safe to analyze several stars across the image. The file can be analyzed with various tools:

• Skycat (launch with "skycat &" from an xterm window). Load the image (single CCD) with the File->Open menu. Open the "Pick Window" from the "View" menu at the top of the main window. Select zoom "x1" for the main display and click 3 times on the "Z" button of the Pick window to dodge a bug in this tool. Then simply click on the "Pick Object" button at the bottom of the Pick window and go click on the object you want measured. The user has to reproduce the procedure of clicking on the "Pick Object" button and then going and clicking on the new object for each new measurement. The feedback from Skycat on the Pick window provides the FWHM in pixels (sampling at prime focus is 0.206" per pixel) along the major and the minor axes of the ellipse fitting the object. The best focus is found when "M" and "m" are within 0.3 pixels of each other AND "M" is minimal within the entire sequence and the peak intensity is maximal within the entire sequence (the Peak value is part of the feedback on the "Pick" window). See the CFH12K Focus Shapes figure for an example on how to interpret the evolution of the PSF across a focus sequence. Skycat has the advantage of being very fast vs. the other tools.
• Custom tool: documentation can be found on this page.
• IRAF with the "imexamine" tool. Go to the "/h/cfh12k/iraf" directory and launch IRAF with the "cl" command. Start "ximtool" with the command "!ximtool &" at the IRAF prompt. With the MEF format, the user has to give the extension number he wants displayed ("display 565770o[03]"). In SPLIT mode, the "lookat" command allows a transparent handling of the multiple file format ("lookat 565770o 02"). Notice that the MEF format works with the extension number (01 to 12) while the SPLIT format works with the CCD number (00 to 11): the correspondence is simply an offset of 1. The "lookat" script will also put you directly in the "imexamine" mode. To exit "imexamine", type the "q" keystroke on the keyboard with the cursor positioned on the image display. Use the Moffat profile fitting by typing the "r" keystroke on the keyboard while the cursor is put on the object you want measured. You get a display of the best fit: the less scatter you see around the fit the better (see this for an illustration). Also, the numbers at the bottom of the graphic window give you an estimate of the FWHM in pixels (sampling at prime focus is 0.206" per pixel). To access the information on the shape, you have to type the "e" keystroke on the keyboard to obtain an isophotal representation of the object (the CFH12K Focus Shapes figure uses it). With these two functions, you can derive the best focus position: the star image has to be round AND the FWHM must be the lowest AND the peak intensity is maximal within the entire sequence (the peak intensity is given with the Moffat profile function). This is a powerful tool though two different functions must be used, making this procedure rather time consuming (IRAF is also slow at loading the image).

The best z value can be entered in the focus interface. This value is sent to the telescope control system with the Set new z button (a confirm window will pop up). When moving to a new z position, the bonnette will always go to a lower z first before going to the selected value to avoid any hysteresis effect on the bonnette vertical motion.
CLI= pfocus 2.3 (set z value)

Some information is provided when the pfocus command is entered with no argument:

Syntax = 'pfocus z' - set prime focus z [mm] (-1.0 < z < 6.0)
Approx. z value: B=4.50 V=1.75 R=1.60 I=3.30 Z=2.55
Approx. z value: Ha=2.10 HaOFF=2.20 TiO=2.10 CN=2.50
Current prime focus bonnette z = 1.965 mm

Also, when observing in a given filter, if image degradation is noticed in the course of a set of exposures, the focus can be adjusted by small amounts (0.05 steps) to get back to the optimal position. To determine which way to go (up or down), refer to the previous focus sequence in the filter being used and look at the orientation of the star image ellipse above and below the best focus position. This angle changes by 90 degrees, hence it is easy to notice. Make sure that you compare the data from a SAME CCD both on the focus AND the science exposures. Recall that the last focus exposure with the higher z has a double offset (towards higher z) which can be either at the bottom of the sequence on the screen or at the top depending on which CCD you look at (top for CCD[00,01,02,03,04,05], bottom for CCD[06,07,08,09,10,11]). For example, the best focus position was found to be the central position of a 5-position sequence. The 2 last exposures (with the higher z) show that the ellipse has a direct angle of 160 degrees (see also the illustration below for a different scenario). If the science exposure exhibits the same effect then the z focus must be reduced by 0.05 units (never do more than 0.1; if this seems to be the case, then it is better to restart a complete focus sequence). If you send the command from the CFH12K session, it won't be executed immediately: the other pending commands will be done first (like finishing the current exposure). As a consequence for this particular situation, you need to ask the Observing Assistant piloting the telescope to send a passthrough command to the prime focus bonnette. Ask for the new value you wish, then monitor the various status windows for the new z value. This can be done during an exposure (within the first minutes of the exposure otherwise it would be quite useless). Beware that you'll have to wait for the end of the following exposure to evaluate the effect of the correction: when working with short exposure times this is quite efficient but it can be dangerous when using long exposure times. In any case, it is crucial to evaluate the image quality of the first image produced after such change.

The "CFH12K Focus Shapes" figure shows the Point Spread Function (PSF) behavior around the optimal focus position. Notice the rapid image degradation beyond 0.2 mm of the optimal focus position. "M" is the FWHM along the major axis of the ellipse fitting the star, "m" is the FWHM along the minor axis of the ellipse fitting the star. "Angle" is the direct angle of the major axis of the ellipse fitting the star. Notice here that the best focus indicates a FWHM of 0.5 arcsecond (I-band)! Beware that under average seeing conditions, this "rotation" of the PSF is harder to detect since the effect is blurred within the seeing disk.

When taking standards in good seeing conditions, the observer may need to defocus the camera slightly such that the brightest photometric standards don't saturate on a 4-second exposure: adding a 0.2 step on the optimal value for the initial filter will do (adding more would push into the domain where the PSF starts to degrade dramatically). This offset will be then also present on the other filters. For example, if the focus is optimal at 3.30 in the I band and the seeing in 0.5", use the command pfocus (or the focus GUI) to set the z stage to 3.50. Take all your photometric standards exposures by just selecting the new filters: for all of them the focus will be off by +0.2 mm. After the sequence, reduce the focus by 0.2 mm to be sure you're back at the best focus for the science exposures.

Within the GUI, as well as on a command line, the observer can select a dithering pattern. The system will automatically take exposures and offset the telescope between each. For the seven choices made available on this form (there are actually up to 21 positions available with this "dp12k" command from the CLI and script modes) all exposures are within a disk of 15 arcseconds on the sky. This is sufficient to fill the gaps between the CCDs (about 6 arcseconds) and jump over the largest cosmetic defects. It ensures also that the wide-field corrector optical distortion won't affect the data such that they can be stacked together without ressampling.

Beyond the 15-arcsecond disk, in normal subarcsecond seeing conditions, the observer will be forced to re-sample his data to correct for the optical distortion before being able to stack them (not doing so would result in terrible image quality degradation across the field of view). Also with large offsets between exposures, some significant portions of the field will be covered by different CCDs: this is an issue when combining the data since the CCDs in the mosaic have slightly different color equations: it will affect the photometry of the objects.

When used as a CLI command (or from a script), a second and third argument to "dp12k" allows the user to start the sequence at any given position within the dithering pattern and take a given numbers of exposures from that starting point (up to the number of exposures selected for that dithering pattern). This is handy when a sequence has been aborted and needs to be restarted (after an imposed focus sequence for example), or if the sequence needs to be extended to ensure that the desired signal-to-noise ratio will be reached after the transmission monitor showed that there is some absorption due to some cirrus or that the seeing degraded in the course of the sequence. It could also happen that the sequence needs to be restarted because a single night of observing was not enough to cover that object. In that case (or more generally anytime after a slew of the telescope was executed) the telescope needs to be perfectly recentered on the initial position. The "snap" and "offset" commands are of great interest to properly go quickly through this procedure (the FITS scientific frame previously obtained on the initial position must be the reference, not its coordinates which are affected by some pointing errors). But proper respect of Telescope Control System coordinates (guiding box position, X&Y coordinated of the guide probe) should allow recentering within 2 arcseconds.

From 2 to 4 positions, the disk is not centered on the initial position: the first position of the pattern is the initial position of the telescope and lies on the 15" diameter circle (position label "a" for DP2, position label "1" for DP3, position label "A" for DP4). The telescope returns to the initial position at the end of the sequence. For 5 positions and up, the disk is centered on the initial position of the telescope (position label "Ref" for DP5-21) and the system returns the telescope to that central position at the end of the sequence.

The dp12k script (/h/cfh12k/scripts/dp12k) can also be used as a reference by the observer to build his own scripts (with larger offsets for example if they are willing to ressample all their data to correct for the geometrical distortion and cope with mixing their signal from CCDs with different responses). Here is the help message for the "dp12k" tool ("Ref" refers to the first exposure of the sequence = the initial position of the telescope at the beginning of the sequence):

"FITS file format" Split/MEF FITS format selection

Open the FITS file format selection window. If the Multi-Extension FITS format is selected (FITS=MEF), for each image a unique FITS file is created with a ".fits" extension. If the MEF mode is off (FITS=SPLIT), the 12 individual CCDs will be saved in independent standard FITS files along with the primary header in an individual subdirectory. In the main image directory, there will also be a binned by 8 image of the entire mosaic.

Note that as shown in the drawing, the name of the CCDs does not match the extension name in the MEF file. Refer to the map on this page when looking at a particular detector within the mosaic.

The binned by 8 image has a "-B8.fits" extension. This binned image organizes the CCDs as shown on the drawing with the North towards the bottom and the East towards the right [Reminder: the CFH12K camera can not be rotated]. The relative geometry of the mosaic readout outputs is taken into account so that the image represents the observed field properly. World coordinates are set up correctly to allow the observer to travel in the CFH12K field and locate his objects of interest. Note that the binned resolution is very low (1.6 arcseconds/pixels!) and small objects won't be visible, but this quick-look is intended to be a tool for field recognition, centering, and checking. The high binning factor has the great advantage of enhancing the faint large scale features that would be invisible on the non-binned image. Use this image to detect any contamination problem (nearby bright star light scattering, vignetting by the guide probe, etc.). The binned by 8 image is automatically displayed at the end of each new exposure (that image is named current.fits).

It is easy to go from one format to another with the use of the split2mef and mef2split scripts. For example:

SPLIT to MEF: split2mef 495401b -> will create 495401b.fits (it will not remove the Directory 495401b and files in it though).

MEF to SPLIT: mef2split 495401b -> will create the subDirectory 495401b with all the files inside (it will not remove the file 495401b.fits).

You can also convert a whole set of images at once. They need first to be organized in a single sub-directory: use splits2mef and mefs2split scripts. For example:

SPLIT to MEF: splits2mef /h/cfh12k/images/Night161200 -> will create MEF files from all the SPLIT files found in the Night161200 directory (it will not remove the initial files though so beware of the disk space used!).

MEF to SPLIT: mefs2split /h/cfh12k/images/Night161200 -> will create the SPLIT images from all the MEF files in the Night161200 directory (it will not remove the initial files though so beware of the disk space used!).

Scripts can be executed from any window opened on the session host Xterm or Rxvt on mahina, or from the Director window using the "!" character prior the script name. They can be as simple as the following examples or more complex as some of the scripts that can be found in the "/h/cfh12k/scripts" directory. Scripts can call other scripts. The user should build his own scripts in a Directory separate from the "/h/cfh12k/bin" directory to avoid confusion (in the existing directory "/h/cfh12k/scripts" for example). In the following examples, the scripts can be executed from the Unix prompt with the following command:

source scriptname (from a terminal)
!source scriptname (from Director)

(the script can also be made "executable" with the "chmod +x scriptname" command and then the "source" command prior the script name is not needed)
chmod +x scriptname (after editing the script)
scriptname (from a terminal)
!scriptname (from Director)

WARNING: if you want to abort a script, first interrupt it with Ctrl C in its running window and then abort the exposure if necessary. Never abort an exposure before killing the script, it will confuse the system (use either the GUI or the CLI interface to send the break, stop, or abort command).

Scripts can combine the following commands: (commands "flux", "snap", "tf12k", "focus", "mef" are not appropriate for use within a script of course)

clicmd go n

-> take "n" exposures ("clicmd go" for a single exposure).

clicmd etype t

-> set "t" exposure type (o=object, f=flat, d=dark, b=bias).

clicmd etime n

-> set exposure time to "n" seconds.

clicmd raster full

-> set raster readout mode (full, full bin2, full bin4).

clicmd filter 0 B

-> select filter 0 (here with B filter at that position), positions are: 0,1,2,3. Refer to the current setup of the CFH12K in the main GUI window to indicate proper name of each filters (the menu will list them top down from 0 to 3).

clicmd fits observer CFH12K aficionado

-> set the FITS OBSERVER header for the next exposures.

clicmd fits object NGC 3486 B band

-> set the FITS OBJECT header for the next exposures.

clicmd fits comment This is the comment

-> set the FITS COMMENT header for the next exposures.

offset a b

-> offset the telescope by "a" arcseconds East and "b" arcseconds North.

pfocus z

-> set the z of the bonnette (-1.0 < z < +6.0).

dp12k n (2,3,4,5,6,7,8)

-> Take a set of n dithered exposures.