NOTCam User's Guide

• Instrument overview
• Detector
• LN2 spilling
• Shutter monitoring - "the buzzer"
• Field orientation
• Flat fields
• Focusing
• Sky background level
• Individual exposure times - Number of averages
• Dithering strategies
• about NOTCam data aquistion
• about darks
• Overheads
• Observing scripts
• Calibrations
• Standard stars
• Data format
• Encountered problems
• NOTCam QuickLook reduction package
• Data reduction - flat fields

Check also these links:

• New Science Array (SWIR3) (detector in use since Dec-2007)
• First Science Array (SWIR2) (in use from Oct-2005 to Apr-2006)
• Engineering Grade Array (SWIR1) (used Jun2001 - Dec2005 and May2006 - Dec2007)
• NOTCam sensitivity (zeropoints, backgrounds)
  
  
• NOTCam Cookbook Guides you through the night at the telescope (NOTCam imaging and spectroscopy).
• NOTCam spectroscopy

Instrument Overview

In the future we expect to upgrade NOTCam with Wollaston prisms for imaging polarimetry and spectro-polarimetry.

On entering the front window of the dewar, light passes through the following components before impinging on the detector ( engineering grade array (SWIR1) , first science grade array (SWIR2) , new science grade array (SWIR3) ):

• the aperture wheel
  - At the telescope focus, carries slits and masks. There are 4×50-mm and 8×18-mm slots.
• the collimator
• the shutter
  - A rigid spinning disk.
• two filter wheels
  - Each with 16 slots of 25mm diameter allowing for the use of up to 30 filters.
• the pupil wheel
  - At a pupil plane, carries Lyot stops and other pupil masks. There are 16 slots.
• the grism wheel
  - Carries dispersing elements. There are 16 slots. Check NOTCam spectroscopy pages.
• the lens wheel
  - Carries two sets of imaging optics, one for each plate scale (the WF and HR cameras), and the pupil re-imager.

Since NOTCam is kept cold for long periods at a time, we do not change filters, slits, grisms etc. more often than absolutely necessary (see the current setup).

The original optical design is described here (pdf).

This entire assembly is mounted on an optical bench which is in thermal contact with an LN₂ tank. Additional cooling is provided by a pulse-tube refrigerator (PTR) from Iwatani. The optical bench is covered with a radiation shield and this and the LN₂ tank are surrounded by a superinsulating blanket. The final enclosure is the dewar skin which maintains a high-quality vacuum surrounding the instrument. More about Mechanics and Cryogenics.

The SWIR detector

Together with the upgrade from SWIR1 to SWIR2 in December 2007, there was also an upgrade of the NOTCam Array Electronics. The first science array, SWIR2, failed after 6 months and was replaced by a new science array, SWIR3, which has been used since December 2007. Each detector has (had) its particular characterics, described in the above links to the detector verification pages. An upgrade of the clock boards in January 2010 improved the performance of the currently installed SWIR3 detector in removing the shift-register glow and lowering the read noise.

The behaviour of the detector is continuously monitored in our Detector Quality Control.

Schematic drawing of the detector with the four quadrants marked in the numbering system used by Copenhagen University Observatory (CUO) for NOTCam. (Note that Rockwell/Teledyne numbers from 1 to 4, but in the same order.) The arrows show the direction of the fast readout and the corners in which it starts. All four quadrants are read out simultaneously.
Left: Valid from 2001-2005. The lower left corner of quadrant #1 is the location of pixel (x=1,y=1).
Right: Valid since Jan 2006. The lower left corner of quadrant #0 is the location of pixel (x=1,y=1).

LN₂ spilling

Since January 2003 NOTCam is running on the PTR also while mounted on the telescope, and the problem of spilling LN2 is no longer that critical in terms of temperature stability of the detector. For typical nights the detector temperature is as stable as +- 0.1 degree. Tracking a target at very low airmass for a considerable time causes some spilling and the detector temperature deviation will be larger, but in general this is not a problem. NOTCam is re-filled with LN2 in the mornings in order to make sure that any deviations due to spilling are corrected before the next night. See NOTCam filling (internal page with limited access).

In 2009 the vent tube got an extension fitted such that LN2 will spill from below the instrument to avoid dripping on sensitive parts.

Shutter monitoring - "the buzzer"

Field orientation

NOTCam orientation on the sky for different field-rotation values.

If you wish to rotate the image on the DS9 display to have North Up and East Left, you can click on Zoom and then tick the Align button, or alternatively, using the command line, you can toggle alignment on/off with the sticky command notcam.wsc-align yes/no.

field-r -90 gives South left and East up in the images displayed both on the BIAS DS9 (left) and the post-processing DS9 (right). Since Jan-2006 when changing data format to Multi-Extension-Fits (MEF) the stored images are flipped in X with respect to the images in the real time buffer (left DS9). Note that the image shown on the BIAS DS9 (left) is flipped in the display only in order to avoid confusion! (Its X values are not flipped and X increases from right to left! NB! You should not use the BIAS DS9 to measure positions!)

The command notcam.teloffset x y moves the target(!) a given number of arc seconds in the detector X,Y directions independently of the field-r orientation. For example, teloffset 10 10 moves the target 10'' to the right and 10'' up on both the BIAS DS9 and the post-processing DS9.

For example, for field-r -90, the telescope(!) moves 10 arc seconds:

• South with teloffset 10 0
• East with teloffset 0 -10
• North with teloffset -10 0
• West with teloffset 0 10

If you are using another field orientation and want to know exactly in which cardinal direction you are moving the telescope, you can do the following test. Set previous image for sky subtraction on the DS9 with setskysub 1 and take an exposure. Move the telescope 10'' in any direction using teloffset, and take another exposure. On the DS9 you will see the current position and the previous position (negative) and by clicking the Zoom and the Align buttons on the DS9 it should be evident in which direction the star has moved, meaning the telescope has moved in the opposite direction.

If you are preparing your observations and you want another field orientation it is advised to use our Pointing Script Generator where you can indicate your sky position angle in degrees and the corresponding telescope field-rotation is calculated.

We recommend differential twilight sky flats rather than dome (lamp ON-OFF) flats. Dome flats are not perfectly flat at the NOT. Thus, if weather allows in the evening, point to a selected blank field in the east about 10-15 min before sunset and then open the dome (i.e. select a field with RA a couple of hours more than current ST to make sure you avoid pointing towards the sun in the west). For the morning twilight point to the west (RA of blank field being 2-3 hours less than the current ST) about 30-40 min before sunrise and remember to close the sideports first. Also, close the dome completely before parking the building in the morning if you continue until after sunrise! Use the NOTCam Sequencer observing script:

The New Science Array (SWIR3) flatfield (WF camera, Ks-filter) is flat to 1% in 20x20 pixel areas and flat to 3% across the FOV, and looks about the same in all filters. Flat field corrected science frames are typically flat to 0.1%. See SWIR3 flat field.

The observing script notcam.skyflat t should take about 3 minutes to perform. This short duration ensures a sufficient sky gradient for each filter within 10-15 minutes elapse between the first and second measurement, which is acceptable for subtracting the thermal contribution, while it allows time for 2 cycles of 3-5 broad band filters.

The script notcam.skyflat automatically saves the raw NOTCam flats to the /data/service/calib/ directory.

If you could not get skyflats for all your filters, the second best option is to use differential domeflats. Because of the structure on the inside of the hatch, we recommend setting the telescope to alt 45 or so, where the inside of the dome is a bit "flatter". Differential dome flats can be taken with the NOTCam QC lamp (adjustable from the control room) ON and OFF in each filter. NB! The integration time must be the same for the ON and OFF frames. For the HR camera, it is useful to use mexp 20 10 for all filters, lamps turned on full for K-band and a bit down for J and H. For the WF camera mexp 10 10 is useful for the same adjustment of the lamps.

There is also a NOTCam skyflat archive with master flats for the most commonly used filters taken at different dates.

We recommend to focus every evening using the WF-camera and the Ks band and apply the focus offsets for the other filters as found in the table below. This is done automatically when using the setup scripts notcam.setup-ima and notcam.setup-spec, where the corresponding focus-delta offset is applied, taking as a reference the focus found for WF camera imaging in the Ks band, which is defined as the default NOTCam focus.

The focus offset is wavelength dependent so for the narrow-band filters we use the focus offset for the closest broad band filter.

1. If you have a suitable target for focusing you can use it or go to one of the NOTCam standard stars that are high up. Autoguiding starts automatically if you use:

   auto-positioning-full
   goto-object <my-object>

   where my-object is the name of the object in the TCS catalogue.

2. The following scripts sets up the instrument for focusing the WF and HR camera, respectively. It puts in the focus pyramid and the Ks band filter and sets the internal focus and the aperture mask according to the camera chosen.

   notcam.setup-focpyr <camera>
   where camera is WF or HR.

3. Take an exposure

   exp 15

4. Move the telescope about 15'' in any direction to be able to do sky subtraction.

   teloffset 0 -15

5. Set the sky-subtraction mode to use this image for subtraction and take a new exposure.

   setskysub <image>
   exp 15

6. Determine the offset from the best focus, manually or automatically, or both.

   notcam.focuspyr
   notcam.focuspyrauto

   The value that comes out is the offset from the actual focus position. You must add this number to the current focus value to get the new focus value.

7. Change the telescope focus tcs.focus-position <new-focus-value> and iterate at least once. Don't forget to take the focus pyramid out.

*) This is the default NOTCam telescope foc-pos value, but it may vary by up to around 100 units, and this is the focus value you should determine every night by focusing the telescope in the Ks filter. Usually, the accuracy of the focus correction is around 10-15 steps, but note that with a seeing of FWHM=0.3'', telescope focus steps as small as 5 units make a difference.

Remember to take out the focus pyramid when finished.

If you are doing NOTCam spectroscopy the internal focus setting differs between camera, filter, and slit used, and the easiest way to set both the internal camera focus, as well as the telescope focus offset, is to use the setup script notcam.setup-spec. Also for spectroscopy, the telescope foc-pos value should be set to the default or the one found for the night by doing focusing for the WF camera and the Ks band.

This corresponds to roughly J = 15.9, H = 13.9, and Ks = 13.2 magnitudes per square arc second.

1. Monitoring of reset-read-read mode
2. Monitoring of ramp-sampling mode

The sky background is high in the near-IR, and a sufficient amount of counts to reach BLIP is usually obtained within a few seconds for broad band imaging. For narrow-band imaging, on the other hand, the time to reach BLIP should be carefully considered.

For more information on sensitivity and zeropoints with recent updates check page: NOTCam sensitivity .

The maximum individual integration time is the one where the signal is well below saturation (which starts at 56000 ADU for the new science grade array), and preferrably in the linear range of the detector, which is somewhere below ~22000 ADU (for the new science grade array). For stars the maximum integration time you can use without exceeding these limits is highly seeing dependent. In practice, the limiting factor for faint sources is usually the need to move the telescope (dither). It is recommended to stay at most 1-2 minutes per sky position before making a dithering offset to allow for good sky subtraction.

The background is highly variable, and strongly temperature dependent for the K band. The airglow lines are most prominent in the H band and their brightness may change by up to 50% during a night, while the typical amplitude and period are 10% and 5-15 minutes, respectively. For a typical cold winter night with an ambient temperature of 3 deg Celsius, the background will saturate in individual exposure times of 1000s, 235s, and 160s for J, H, and K, respectively. These saturation times will be shorter for warm summer nights. In addition, you will typically want to have the background within the linear range of the detector, limiting the exposure times to 360s, 80s, and 70s in J, H, and Ks for cold winter nights. In warmer nights, when the water vapor content is higher, or in the case of thin clouds, you may have to limit the exposure times much more, for instance 30s in H and Ks is sometimes found to be the limit.

The detector is non-linear in response, and staying below ~20000-25000 adu only means staying inside the range where the non-linearity is < 1%. It is possible to correct for the non-linearity, pixel by pixel, and the non-linearity correction coefficients can be downloaded form the NOTCam Calibrations page.

The memory (or charge persistency) effect is less than 1% and cleared on the first read for non-saturation levels. For saturated pixels, however, there is a positive memory which persists for many reads, although at a relatively moderate level of less than 0.03% in the 6th readout. It is recommended not to saturate the array, if possible. If impossible to avoid, and if the saturated pixels are close to your target, then it is recommended to clean the array with the command clear which resets and reads the array once without showing or storing the image and which takes 3.5 seconds to perform. Depending on the expected severity of the memory, the clear command may be applied once or several times. There is also a script called notcam.clean3 which makes 3 clear commands and then a dark 0 to show you the result.

It is generally stated that as long as one is in the BLIP regime, it is better to take many co-averages to increase signal-to-noise, rather than increasing the individual integration time. This is not entirely practical with NOTCam. The reason is the large overhead owing to a relatively long readout time (3.6 sec per read, 7.2 sec in total per image) limited by the current detector controller. We expect a readout time of < 1 sec with the new controller.

Note that the ramp-sampling mode gives a large dynamical range, since you save a number of integration times in the fits file (see below). If you saturate (or enter non-linear mode) for very bright targets while you want to go deep for faint targets in the same field, you can select to use one of the first reads in the cube for the bright sources. Note, however, that saturation will produce memory effects in several consequtive images, and you may want to add in the clean array command clear once or more between each exposure to minimize this effect.

Make sure that you repoint the telescope to the start position at the end of every observing script, or to any position you want, such that you keep track of where you are upon possible repetitions of the macro. Between every repetition of a dither pattern you may do a manual offset to some roughly random direction by just typing notcam.teloffset x y in the Sequencer window. In this way the same pattern is not repeated on the same pixels. You can use our available template scripts such as for instance:

1. notcam.9point (a 3x3 grid)
2. notcam.5point (a 5-point dice)
3. notcam.beamswitch (a beamswitch script for extended sources)
4. notcam.abba (ABBA dithering along the slit for spectroscopy)

In general, autoguiding is always recommended. Note that since 2007 there is no additional overhead in dithering sequence owing to autoguiding.

The Script Generator is a useful tool which also gives you an idea about the overheads involved.

Point sources

If the target is not contained fully within one field, make a mosaic. The mosaic can be made as a raster scan which can be repeated again and again until you reach the desired sensitivity. Note that you should dither the telescope a small amount between each repetition of a raster scan to be able to remove the bad pixels and bad features from your sources. For each scan, depending on its size, all or groups of the target frames can be used to create a local sky frame (provided they don't contain extended emission). If there is a small amount of extended emission in a small fraction of the N mosaic frames, these can simply be excluded when making the sky frame.

Note that depending on the dither steps you plan to make, the corresponding guide star area must be selected when doing target acquisition. This is explained in the Autoguiding section in the Step-by-step observing guide.

Extended sources or crowded regions

With NOTCam there are three different commands to start an exposure. Users familiar with BIAS (Brorfelde Image Acquisition System) will recognize two of these from the BIAS versions in use with CCDs (exp, mexp). For IR arrays the data acquisition is different, however. In contrast to CCDs, where the same Si array is used to collect the photons and to read out the photo electrons, IR arrays are hybrid devices with one array for the detection of photons (HgCdTe layered on a a sapphire substrate) and another for reading the photoelectrons (the multiplexer constructed using conventional Si based CMOS technology). The two are manufactured separately, then aligned and connected pixel by pixel with small bumps of indium.

Each individual photodiode (the IR sensitive HgCdTe) and its corresponding electronics on the multiplexer (together comprising one pixel) possess a given electrical capacitance C. The voltage V across this capacitor is set to a given value when the pixel is reset. The reset switch on each single multiplexer circuit is closed at the beginning of an integration, thus charging the capacitance. During an exposure, the reset switch is opened and incoming photons release electrons which discharge the capacitor ( dV = dQ/C). At any time during the integration (before the next reset) the multiplexer can read the detector, i.e. measure the voltage across each capacitance, without changing its value. This is called non-destructive readout.

The multiplexer is divided into four independent quadrants, each with its own clock and bias lines and its own output amplifier. Each quadrant is read out simultaneously. There are many different readout modes for multiplexed arrays. With NOTCam we offer two modes:

1. A basic reset-read-read mode (also called Fowler sampling). What it does is: reset the capacitor, read out immediately after (the so-called reset frame), then integrate for as long as wanted, then read-out again after the end of integration. This method efficiently eliminates any possible long-term drifts as well as the kTC noise, which arises from a fundamental limit to the accuracy of the reset process and causes an uncertainty in the voltage level. One exposure thus consists of two readout frames, one reset frame which must be subtracted from the end of exposure frame in order to obtain the desired signal. This subtraction is done on-line, but the reset frame is kept and stored as a second extension in the [1024,1024,2] MEF file. This readout mode is used with the two commands exp, mexp.
   
   When using the mexp t N command, be aware that there is usually an anomaly with the first of the N images. This one should then be skipped in the final averaging.
   
   
2. More sophisticated is the ramp-sampling mode, where the signal is sampled many times during the exposure in order to reduce the noise. A linear regression analysis applied to the data points provides a mean flux rate and decreases the noise by a factor sqrt(N) where N is the number of such multiple non-destructive readouts. In NOTCam this mode is used with the exposure command frame t N, which produces a final image of exposure time t * N seconds. What the command does is: reset the capacitor, read out immediately after, then integrate for as long as wanted, and during this integration read out at every time step t. The minimum value of t is 3.6 (since the readout time is 3.6s) and the maximum number of N is 14. The frame command produces one output file, a MEF fits file with N+2 extensions. The first image extension is the final result of the linear regression analysis, then follows the N non-destructive read outs (reset-subtracted), and finally the reset frame.

We strongly recommend to save all individual images.

Please, note that the dark current measured with NOTCam is non-linear, variable and higher than the specifications (which state less than 0.1 e/s/pix). See NOTCam Darks - internal summary for more details about the darks, and see NOTCam Calibrations for examples of dark images.

For imaging in the background limited case, we thus do not recommended to correct for the dark level by taking separate dark images that are aquired at a different time. By using differential flats and sky subtraction, as usual in infrared image processing, the dark level will automatically be subtracted out.

For spectroscopy mode, however, we recommend to take a set of darks with the same readout mode and integration time as your observations, although spectroscopy is also done in differential mode (subtracting a dithered image). But since the number of dithers is less, for practical reasons, darks are useful to estimate a hot pixel mask. The number of hot pixels increases with the integration time. For this purpose, the calibration script notcam.notcam-calibs at the end of the night is useful.

NB! Please, note that the linear regression analysis image for darks taken with the dframe t N command is not reliable, since the dark current is not linear (but this is not important for finding hot pixels).

• exp t: T_tot = T_seq + T_read + T_exptime + T_read + T_subt + T_show + T_save + T_teloffset
  
  
• mexp t N: T_tot = N * (T_seq + T_read + T_exptime + T_read + T_subt + T_show + T_save) + T_ave + T_show + T_save + T_teloffset
  
  
• frame t N: T_tot = T_seq + T_exptime + T_read + T_linreg + T_show + T_save + T_teloffset
  
  

New! Since November 2011, the exp mode is speeded up by letting the telescope move as soon as the shutter is closed, i.e. while reading out, subtracting, showing on real-time display and storing the data file (marked in red above). The measured total time spent is lowered as follows: T_now/T_before = 0.76 for exp 3 and 0.82 for exp 10. This does not apply for staring mode observations.

---

• T_seq = 0.6 - 1.5 s (sequencer startup, variable)
• T_read = 3.6 s (controller readout time)
• T_subt = 0.6 s (subtract reset frame, get header info, etc.)
• T_ave = 0.6 s (calculate average image etc.)
• T_linreg = 0.6 s (pixel by pixel linear regression analysis, get header info etc.)
• T_show = 0.3 s (minimum time for real time display on DS9, can be switched off)
• T_save = 1.3 s + 0.1 s x size(Mb) (minimum time to store file)
• T_teloffset = 8s (for 10-15'' step, typical value when autoguiding, depends on how much is az and alt movement)

  ---

These are measured overheads for NOTCam (Feb 2011) :

Note 1) Time to add to the exptime when using exp/dark . This includes the time it takes to read the array twice (2 x 3.6 s) plus the time to store the 4 MB file (2 s) plus any possible delays in the acquisition system. From November 2011, when these exposure commands are followed by a telescope dither, observing is speeded up by letting the telescope move while doing the 2nd readout, saving about 5-6 seconds of total overhead. This does not apply to staring mode observations. (During 2008-2010 the overhead was ~ 13 s, while during April and May 2007 it was as high as 18-20 s.)

Note 2) This includes the time needed to read the array once (3.6 s, since the other reads are interlaced with integration), to store the large file (up to 7s for 32 MB), and any possible delays in the acquisition system including the linear regression analysis (up to ~1.4 s). Note that when the time between non-destructive reads (i.e. the t in "frame t N") is < 5 s, there are internal delays in the PC-board (see NOTCam BIAS for details). This may increase the overhead with 2 s per frame command. The following note was valid from 2007-2010: The total time needed per command occasionally becomes much larger if the program experiences hick-ups (up to 150s), a problem present from Apr-2007 to Oct-2010, for the frame command. Both the average overhead values and the frequency of hick-ups were variable with time. This additional readout overhead was quite impossible to predict.

Note 3) All wheels can be moved simultaneously! The wheels move in one direction only. The two filter wheels, the grism wheel, and the stop wheel all take ~ 30 s to make one full turn.

NB! The time to do telescope dithers has increased slightly since Nov-2007. The reason is the introduction of an extra wait such that the telescope has come to a complete rest before the next integration starts. Tested for 240" teloffsets and short (2 s) integrations using the HR-camera in 0.3" seeing. The earlier appearance of elongated stars in such extreme conditions has now been remedied. Note that the exact time needed to make a given dither depends on how much of the movement is in azimuth. In the above table are given the value ranges measured on a number of tests.

The frame command is normally more time efficient than the mexp command. This is due to the long readout time, and the use of a shutter to determine the exposure time. If you need individual exposure times shorter than 3.6 seconds, you must use the exp or mexp command. Nominal overheads are calculated by the Observing Script Generator.

• notcam.9point frame 6 8 my-object 10 2 3
  This does a 9point dither (3x3 grid) and at each position takes an exposure using the exposure command frame 6 8 (which gives a 48s exptime). The dither step size is 10'' and a skew of 2'' is applied to tilt the grid. The dither is run 3 times. Recommended for deep imaging where it is vital that the sky can be well calibrated.
  
  
• notcam.5point exp 6 1 my-object 15 3 1
  This does a 5point (dice) dither, and at each position takes an exposure using the exposure command exp 6 (note that you need to add N=1 as input). The dither step size is 15'' and a skew of 3'' is applied to tilt the grid. The dither is not repeated. Useful for bright targets where you do not worry so much about the sky.
  
  
• notcam.beamswitch frame 10 10 my-object S-posX 240 10 2
  This is a beamswitch mode going alternatingly ON and OFF target. At each position the exposure mode frame 10 10 (giving an exptime of 100s) is used. The OFF field is in the direction S-posX at the distance 240'' from the ON field. Both the ON and the OFF fields are being dithered by 10'' step sizes in a 3x3 grid fashion, using a skew of 2'' to tilt the grids. Useful for extended targets.
  
  
• notcam.loop-frame 3.6 3
  Staring mode observing, using the exposure mode frame 3.6 3 (giving an exptime of 10.8s). The script is repeatedly taking exposures until interrrupted with Ctrl-C. The same type of loop script exists for the exposure mode exp t.

It is very useful to add one of the two following lines at the end of the scripts:

playphone
astrowakeup

See also: Northern JHK Standard Stars for array detectors by Hunt et al. (1998, AJ 115, 2594) (or see local ps files: paper and erratum ). Contains 86 stars in 40 fields, many of which are faint enough to observe with the frame command. Another advantage is that many of the fields have multiple standards.

We recommend using the same readout mode for the standard as for the target.

Data taken from Jan-2006:

Note that the images are now flipped in the X-axis with respect to images taken before Jan-2006.

Data taken before Jan-2006:

Renaming of FITS header keywords Oct 2012:

Condensation on the entrance window

Condensation on the entrance window.

then there is water condensed on the entrance window of NOTCam. This can happen if the humidity inside the dome is very high over a long time, or if there is some problem with the vacuum in the cryostat. Observations can not be continued. Call staff. NOTCam must be dismounted on its trolley and lowered down to remove the condensation.

The appearance of a dust speck

Large dust speck. This is the ratio image of a J-band masterflat from the 5/3-07 to that of 4/3-07. The largest difference inside the dust speck is on a 3% level. The rest of the image has a value of 1 within 0.1 to 1 %, and a standard deviation around the mean of 0.004.

If the particle is on the entrance window, then NOTCam must be dismounted and the entrance window inspected and cleaned. Before doing this, change to the other camera and make a ratio image of domeflats with the two cameras. This should reveal whether the particle is on the camera lens inside NOTCam. In the example here the ratio image between a HR-camera J-band domeflat and WF-camera J-band domeflat shows that the particle is located on the WF-camera lens. See image below. Note that the feature has moved from lower right to lower left quadrant. Since the particle is inside NOTCam, we can not do much until the instrument is warmed-up and opened.

Ratio J-band domeflat between HR-camera and WF-camera. In addition to the large feature (now moved to the lower left quadrant) there are also 3 smaller and rounder features at about 1% level or so. The greyscale is negative so the dark features means they are present on the WF camera image and not on the HR-camera one.

Pick-up noise

The 9th of June 2009 event

New bad pixels on SWIR3. Green circles mark already existing bad pixels, while the red rings mark the new bad pixel areas appearing overnight. Note that two hair-like features seem to have disappeared on the second image.

Also the flat-field structure changed at the same time. See below the ratio of the flat in J-band from the 8th to that of the 9th (left). The response is about 4% lower in the dark big area in the upper right quadrant. Probably this is due to dust having fallen down on the entrance window. The entrance window was cleaned on the the 10th daytime, and the ratio flat between the 10th and the 9th is shown to the right. The remaining structure is due to (moving) dust on the WF camera lens inside the cryostat. The amplitude in this case is at most 1.5%. The gradient in the lower left corner is 1%.

Ratio flats for WF-camera and J-band, between 8th and 9th (left) and between 9th and 10th (right). To the left is shown the dust on the entrance window which in the darkest part has about 4% lower illumination than the surroundings. To the right is the remaining structure after having cleaned the entrance window, and this is due to dust specks on the WF camera lens that have moved slightly. The peak to peak amplitude here is around 1%.

Strange darks

The problem is currently under investigation.

360 second long darks using "dframe 60 6" taken with the internal focus value at 20 (left) and internal focus value at 5650 (right). The left corresponds to the detector plate being at the furthest distance from the camera wheel. This is the focus value used for HR-camera imaging.

NOTCam package in IRAF

Upgraded to version 2.5 in October 2012.
Note that you need this new version to work on data taken after October 1st 2012! This is due to renaming of FITS header keywords.
Upgraded to version 2.6 in December 2014.
Available for download.

mkflat	Make a differential master flat from bright+faint flats
mklincor	Make linearity correction pixel by pixel on raw images.
reduce	Reduce a small-step dither observation
reduce_bs	Reduce a beamswitch dither observation
skysub	Sky subtraction and flat fielding of a set of images
dophot	Quick-look photometry of individual and combined images

NB! Note that the scripts reduce and reduce_bs does the image registration satisfactorily only on data obtained after 15-Nov-2007, since until then the RA/DEC fits header keywords were not accurate. This was because of a bug in the output coordinates from the TCS, although the actual TCS coordinates were accurate.

Note that for quick-look you don't have to make badpixel and masterflat images, but may download these directly from our archive flat fields and bad pixel masks. You may also wish to download the correction files for non-linearity and optical WF-camera distortion from our archive non-linearity correction models and distortion models. Get started as follows:

• In an xterm on Florence open an xgterm by typing xgterm &
• Inside the xgterm go to directory /home/guest/newiraf/
• Start DS9 by typing ds9 &
• Start IRAF by typing cl (or ecl or ncl)
• Open the notcam IRAF package by typing notcam
• Type help to get a listing of the scripts.
• List script parameters using lpar and edit them using epar

Before starting the reductions you should be aware of the optical distortion of the WF camera (see below) and see the note about NOTCam darks.

Note on the optical distortion of the WF camera

From version 2.4 of notcam.cl you can select to correct the images for this distortion in the scripts reduce and reduce_bs. These will need the input file "notcam.db" which contains distortion models for the three bands J, H and Ks for the WF-camera. These models are based on high-quality (fwhm = 0.5'') data from 2009 of a stellar-rich field (~ 300 2MASS sources). See NOTCam Calibrations - Distortion Correction for more details.

An earlier NOTCam distortion model was provided in 2005 by Magnus Gålfalk based on his observations of B335 made with FWHM=0.5''. The model is calculated from more than 100 stars in the field using the 2MASS catalogue as positional reference. It is available for download at the web page http://www.astro.su.se/~magnusg/NOTCam_dist/. Please, note that images obtained after Jan-2006 are flipped in X and this old model must be correspondingly adjusted.

The HR camera has an excellent optical quality all over. See the NGC4147 image in the H-band obtained with fwhm = 0.44'' and ellipticity typically 0.02 to 0.05 all over the FOV. The pixel scale of the HR camera is 0.078''/pix.

Note on NOTCam darks

Bad pixel masks

It is strongly recommended to correct the flat field images for the zero-valued pixels before making master flats. This is now done inside the mkflat script, and you just have to enter the bad pixel mask to be used. Check the NOTCam bad pixel mask archive for the mask called "bad_zero_sci.fits". See also the page NOTCam calibration images for info about bad pixels.

Check the NOTCam bad pixel mask archive for bad pixel masks from different dates and different arrays.

Image lists

takes all images with prefix NCqd29 and exchanges the string ".fits" with the string "[1]" and pipes them to the file "totlist".

Master flats - imaging

The script searches in the input image list "skyflats" for images with IMAGETYP=FLAT with FILT1ID/NCFLTID1 or FILT2ID/NCFLTID2 corresponding to your input filtid and LENS corresponding to your input camid. In order to have useful flats, make sure that your evening flats and morning flats are separated in two lists, otherwise they will be mixed and the thermal contribution will not be well subtracted. The script sorts the selected images on brightness, subtracts faint from bright, interpolates over bad pixels using your input badpixelimage, optionally corrects the differential images for the "dc-gradient" (if suppress is yes) and combines all the differential images. The master flat is normalized. The output image is diplayed on the DS9.

Since version 2.4 it is possible to autosearch=no and in this case the script will use all images provided in your input list without searching for camera or filter ID. This can be useful for very old data with different keywords.

If for some reason you could not get any useful twilight flat observations during your run, check the NOTCam skyflat archive for master flats from different dates for the most commonly used configurations. Download these to your working directory and use them as input in the following reduction scripts.

Note that the "dc-gradient suppression" is made on all differential flats in the archive from December 2007 (prefix ql13). See below for a comparison of master flats obtained with suppression (left) and without (right). This is a differential twilight flat for WF camera and Ks band.

Non-linearity correction

If you have exceeded the linear range of the detector, and you need good photometric accuracy, you might consider correcting your target images for non-linearity before further data reduction. NB! This should not be done on the /data/notcam/ files from florence! The script mklincor will take the raw image FITS file(s) as input and append another FITS extension to the file(s), where the linearity corrected version of the extension you select to correct is stored. This requires you to have write permission to the raw data. Note that you need two input correction coefficient images that can be downloaded from non-linearity correction models, and more information is found in the document NOTCam Calibrations - Non-linearity correction.

 LINCOR-Y= 'This is the linearity corrected version of image extension: im1'
 LIN-C-BA= 'Linearity correction coefficient BA used: ../nonlin/uj12lin-ba.fits'
 LIN-C-CA= 'Linearity correction coefficient CA used: ../nonlin/uj12lin-ca.fits'

Reduction of a small-step dither - imaging

Any small-step dither can be reduced using the script reduce. The images are optionally trimmed to retain only the overlapping part. If your single expose command per sky position is frame or exp it is sufficient to give as input the first image of the set, like in the below example, where we find the incoming data in the /data/notcam/ directory:

The script is interactive and needs a DS9 to be open. It takes the 'nim' consequtively dithered images, corrects for flatfield using the input flat, optionally interpolates over bad pixels using the input bad pixel mask image, and then makes a sky template using here additive scaling, subtracts the sky (which is for each image re-scaled correctly to conserve the flux), optionally destripes saturated stars (this will not work for extended emission), optionally performs distortion correction on each image using the 'notcam.db' model database, registers the images based on RA/DEC header keywords and the interactively selected stars (place cursor on the DS9 and press 'a') or if no stars selected/found using 'sregister', shifts and optionally trims (here no trimming is selected) all images and finally combines all using here "median combine". If the first image in a dither run is noisy or the array has not had time to stabilize well, it makes sense to exclude the first image in the reductions by setting skip=yes.

NB! If you have used several expose commands per sky position or if you have used mexp then you can not simply give the first image name, but have to prepare the input image list beforehand and remove the average images per position in a list (such as imlist in the example below):

The output result from reduce is stored in the above case as test and displayed on the DS9, while the individually reduced (and shifted) images are stored as test001, test002, ...., test009 . In addition, the sky template image is stored as test.sky, and all image files have extension .fits. (NB! Note that the sky image also includes the dark.)

If there are no alignment stars to be found in the images, or if you do not select any stars, but simply type "q", then in both cases the image registration is done using the "sregister" task which uses the WCS info in the header. This registration is less accurate, but allows the script to work in cases with no alignment stars.

If the resulting image is plagued with horizontal stripes (i.e. from saturated stars and crosstalk), the following workaround usually helps:

Reduction of a beamswitch dither - imaging

The above command reduces the total of 18 images of which 9 are target and 9 sky. The reduction steps are as explained above for reduce depending on the options selected (flat fielding, badpixel fix, sky subtraction, destriping, distortion correction, image registration, shifting, trimming and combining), except that image registration is done twice (for the target as well as the OFF field), and this produces the output target image crab followed by the individually reduced images crab001, crab002, ...., crab009, as well as the sky image crab_OFF and the individually reduced sky images crab_OFF001, crab_OFF002, ...., crab_OFF009. (Again, the sky images include the dark.)

Check out available NOTCam sequencer observing scripts.

Simple sky subtraction - imaging

takes the input image list "grblist" and the input master flat field "flat_h", creates the sky template and subtracts the sky (which in this case has been estimated using multiplicative scaling), and corrects all the 9 images for the flat field. All the 9 images which will have output names grb001, grb002, ...., grb009. Similarly:

takes the input images from the directory /data/notcam/ and the first image name is sufficient if your input images are consequtive in number (if not, you will have to make an image list). Here the sky is obtained from the 16 raw images by additive scaling, and the reduced output images will be called as27001, as27002, ..., as27016 .

Check photometry of point sources

NB! The user has to make sure that the alignment has been performed correctly and that the flux is conserved! It is possible to iterate again the image alignment of the individual files using the IRAF task imalign. NB! Combining images that are not well registered (through taking the average or median per pixel) leads to flux losses. The WF camera has an optical distortion and a distortion correction must be applied in order to have a good image alignment for stars all over the FOV. If the images are not distortion corrected, and if your target(s) are located in the central region, then select stars only in the inner 600x600 pixels for alignment to obtain a reasonable registration for this region.

Check the photometry of a reduction by typing:

Anlaug Amanda Djupvik