Multi Object Spectroscopy
Multi Object Spectroscopy is used to obtain simultaneous spectra of many objects in an EFOSC2 field. MOS is in principle similar to long slit spectroscopy but differs from it in many critical details. For this reason a separate web page has been devoted to MOS. The main stages in carrying out a MOS programme include :
- Obtaining a pre-image of the target field
- Defining the slits
- Identifying targets such that they do not overlap along the spatial axis
- Defining the slitlets (location and length)
- Punching the slitlets on a MOS plate
- Loading the mask into EFOSC2
- Calibrations (wavelength, flat field, mask image)
- Checking slitlet overlap
- Slitlet mask image
- Other calibrations (bias, flat fields, He-Ar arcs)
- Science observations
- Image through the MOS mask
- Science spectra
- A sample jP2PP MOS science observing block
P84 (October 2009) onwards
Note that with the end of support astronomers on La Silla, visiting astronomers will have to do their own pre-imaging observations. Further details on the La Silla 2010+ operation mode are available here.
Proposers will have to specify how much time they need for pre-imaging in their phase 1 proposals. Up to 20 minutes of this time will be taken from the previous NTT visitor, while the rest will have to be taken during the MOS run itself. Or in other words, if you request less than or equal to 20 minutes of pre-imaging, then you don't lose any time during your run, because it will be taken from the previous visitors.
These 20 minutes will be taken from the EFOSC2 run immediately preceding the MOS run. As observers are required to be on the mountain 2 nights before the start of their run, this means that the observer will take his or her own pre-imaging during a mutually convenient 20 minutes during those 2 nights. In the case when there is a SOFI run immediately prior to the EFOSC2 MOS run, no pre-imaging will be possible due to the additional time that would be lost due to the instrument change. Pre-imaging will then be attempted earlier by the observatory staff on a best-effort basis, but it might happen that it will then have to be done during the first night of the MOS run. The La Silla staff (telescope operators and the system engineer) will endeavour to give fair warning to visiting observers that MOS pre-imaging may interrupt their run, and to MOS observers where such pre-imaging will not be possible (i.e. in case of a SOFI run before the MOS run, or in exceptional cases when the preceding EFOSC2 run cannot be interrupted, such as highly time critical observations).
It should be possible to take pre-imaging for ~3 fields during 20 minutes, meaning that it will be possible to have 3 masks prepared for the 1st night of a MOS run. The remainder of this first night should give plenty of time for pre-imaging the remaining fields, so subsequent nights will be able to use the full 5 masks possible on any night. Note that long-slit (single object) spectroscopy will also be possible on the first night if the observer has not been able to prepare enough masks to fill the night.
It is essential that MOS slits are defined based on a prior image taken with EFOSC2. This minimises the effect of CCD distortion. The telescope team will endeavour to provide these pre-images, as a favour to the observer, subject to the following constraints/conditions :
- The request should be made well in advance of the run: The NTT operates in visitor mode the vast majority of the time, and technical or calibration nights are the only time that the pre-imaging c an be done. The pre-imaging request should reach the team at least 6 weeks in advance.
- All the necessary information including target co-ordinates, finding chart with field centre marked, image orientation (see below), filter and exposure time are provided. Please send an email to the La Silla account email@example.com where a ticket will be made for your request. You should also cc your request to the EFOSC2 instrument scientist.
- 20 minutes worth of pre-imaging will be provided per night of the MOS run; i.e. a three night run can ask for up to 1 hour of pre-imaging. Additional images if necessary should be obtained by the observer during their first scheduled night - it takes about an hour (with some familiarity) from pre-image to loading the masks into the instrument. However, the masks cannot be changed during the night. For this reason, it makes sense to ask for at least 5 fields to be pre-imaged (assuming the total number of fields/masks is more than 5), to allow the maximum number of masks to be prepared and mounted for the first night of observing.
- The pre-image exposure time should not be more than 5 minutes - if an object is not visible in a 5 minute exposure one will not obtain a useful spectrum unless targets happen to be a cluster of strong emission-line objects. Observers who need more than the above should apply for the same as part of their regular proposal.
It is in the observer's interest to place the request well in advance of their run - to avoid weather hitches, competition with other scheduled tasks during calibration nights etc.
MOS observers are recommended to arrive in La Silla 2 days prior to the start of their run due to the need to define and punch plates.
Image Orientation: The orientation (long axis) of the slitlets on the sky, in terms of their position angle (from north through east), is given by PA = 90 + Rotator_Offset_Angle. This means that the Rotator Offset Angle that one must specify for the pre-imaging (and in the acquisition template of the science OBs - these must be the same) is given by PA - 90 degrees. See the Adaptor page for more details on orientating the slits. In general, one uses the default EFOSC2 orientation (North Up, East Right on the CCD) for MOS observations, as the targets are unlikely to have a prefered orientation, however occasionally rotating the field helps to get a better spread of targets without overlapping slits. If in doubt, leave the Offset Angle = 0.
Defining the slits
This first step can be done, or at least begun, before the observer arrives on La Silla; the remainder of these steps can only be done on La Silla, as they require the use of custom software only available at the control room. By inspecting the pre-image the astronomer needs to identify the objects of interest in each frame. He/She then has to design the masks in such a way that the slits do not overlap in the spatial (x-coordinate) direction. This is necessary to keep the resulting spectra separate; the slit defining software descibed in the following sections will not allow you to punch a mask with overlapping slits. There are typically a number of ways to arrange slits on a field, and it will not always be possible to get all objects at once if two interesting objects are at the same x-coordinate: At this point you may realise that you wanted to rotate the field, but remember this will require a new pre-image at the new orientation.
The size of the slits is set by the physical dimension of the punch heads. A choice of 3 punch heads are currently offered:
- 1.02" x 8.6" (punch #5).
- 1.34" x 8.6" (punch #7).
- 1.87" x 8.5" (punch #3).
Defining the Slitlets
While it helps to have a plan in advance, the actual design of the masks is done interactively at La Silla. This is done using a MIDAS interface, which is described here. There is also a more detailed page describing the functions of this interface here.
Note that three bright sources (it is best to use stars which will give a good centroid fit) have to have slits placed on them for acquiring the targets into the slits. These can be science objects, if there are suitably bright objects. The three sources should be spread into a wide triangle, ideally without getting too close to the field edge.
It is worth spending time practising with this software, as the results from MOS depend critically on the accuracy of the slit placement. Once the design is finalised, the next step is to physically punch the desired slits from a blank starplate mask.
Punching the slitlets
This task will be done by La Silla Science Operations staff during the day. The observer needs to tell the operator which mask files are required. The process is desribed here (note that the astronomer does not need to worry about these details - the link is provided for day operators and the overly curious).
Loading the Mask
The MOS masks will be loaded into the instrument by either the telescope operator or the support astronomer. Make sure that the lower side of the plate is clearly marked on it. This takes considerable time and will only be done during the day. MOS masks cannot be changed during the night
Each MOS mask is loaded into a numbered slot on the slit wheel and the numbered slot is provided a name in the instrument database (the name will be of the form MOS#n, n : 1-5). The template definition (in jP2PP) is in terms of MOS#n while the instrument recognises only slot numbers on the slit wheel. The operator will update the database mapping one to the other. Observers with multiple masks in their programme should keep track of the mapping between their target field name and MOS#n and provide clear instructions to the observer as to which mask should be associated with a particular MOS#n. This information should be included in the comments field of the EFOSC setup form. Note that for runs with different masks each night, a new set up form should be filled in each night describing the set up for the following night's observations.
The key to avoiding confusion (and therefore lost time during the night) is accurate book keeping. The astronomer should keep a log of the correspondence between the following parameters for each field/mask:
|Field name||The name of the field for the P2PP "Target name" field.|
|Position||The RA & DEC co-ordinates of the field centre|
|Rotator Offset Angle||Rotator offset as described under pre-imaging above: Must be consistent between pre-imaging and science OBs|
|Pre-image||The file name for the pre-imaging fits file will be either EFOSC_Image# (EFOSC naming convention) or EFOSC.$date (ESO archive convention). It makes sense to rename the copy of the image you will use to define the masks.|
|Mask file||The mask defining software produces two files, with extensions .msk and .mask. They are named by default after the input pre-image frame, hence the advice to rename the input files to something unique and recognisable.|
|MOS plate||The MOS plate number (n = 1-5) that is used to request the plate in P2PP (Starplate = Mos#n)|
|Mask image name||The name of the internal flat image taken through the mask during the afternoon (see below). It will have the format EFOSC_ImaInt#.fits|
Checking Slitlet Overlap
The first thing to be checked as soon as the mask is loaded into the instrument is that there is no overlap between the spectra of adjacent slitlets along the spatial direction. The slitlet defining software checks for and eliminates overlapping slitlets but the user can override this check (user beware!). The other reason why overlaps occur is because the slitlets were defined using a smaller punch head than the one actually used for punching.
There is no template for making this check - but a simple way is to do this (or rather request the telescope operator/support astronomer) is to directly use the OS panel. Select the appropriate grism and MOS plate (filter free) and take a 10-20 second exposure with the quartz lamp on. Any overlap between adjacent slitlets will be immediately obvious as the image is displayed on the RTD. In case of overlap, the observer has to decide whether the mask is still acceptable or a new one has to be made - hence the admonition to reach La Silla with time to spare!
The MOS acquisition template works by matching an image of the sky with an image of the slitlet mask - using 3 reference objects (usually stars) and their corresponding slitlets the procedure calculates the rotation and translation of the telescope field required to align the objects and the slitlets. For this the procedure looks for an image of the slitlet mask (name EFOSC_ImaInt.#.fits , #:1, 2, 3, ...) in the directory wefosc:/vlt/insroot/SYSTEM/DETDATA.
This is actually the directory where images observed on the current day are stored. Thus all one needs to do is take an internal image of the slitlet mask on the same day (during the afternoon or the same night) and the image file will automatically be placed in the proper directory. In case of many internal images the user will have to note down the correspondence between the target field (MOS#n) and the internal image name. This information should be provided to the acquisition procedure in a panel which pops up demanding the same.
Do not change any of the parameters on the jP2PP panel except for the Slit (i.e. choose some other MOS#n as necessary).
The rest of the calibrations including Bias and darks, Flat fields, Wavelength calibration (He-Ar lamps) are identical to those needed for long slit spectroscopy and the observer is referred to that page. The only difference is that one has to choose the appropriate MOS#n for the Starplate in jP2PP instead of a long slit.
Note that usually MOS slitlets are not all aligned along the central column and so different slitlets will cover different spectral ranges. One may have to take a few more arc lamp exposures to compensate for the reduction in photons for slitlets whose spectra has shifted towards the blue.
The MOS acquisition procedure is as follows (after the Preset and Focus described in Acquisition Images):
Rotation to align the slitlet mask with the objects
- The acquisition image is displayed and the user (or rather the telescope operator) is asked to select the three reference stars by clicking on the screen.
- The mask image is displayed next and one has to select the corresponding 3 slits in the same order as the reference stars - this step is only done once during the procedure and the slitlet locations are stored for use during later iterations. Right at the beginning, before the Preset and Focus, the procedure asks for the name of the slitlet mask image.
- The procedure calculates the rotator offset angle required to align the slitlet mask with the 3 reference stars and then offers the following options:
- rotate and move on to the next step in the acquisition procedure (rotation < 1 degree),
- rotate but take one more acquisition image to make sure the rotation worked fine,
- do not rotate but move on to the next step in the acquisition procedure (if the offset < 0.1 degree),
- abort the entire OB.
Translation to move the objects into the slit
- The acquisition image is displayed and the user (or rather the telescope operator) is asked to select the object by clicking on the screen
- The procedure calculates the telescope offset required to move the object to the x-pixel defined in the template and on the y-column where the slit is located. It then offers the following options:
- offset and exit from the acquisition procedure (when the shift is less than 1-3 arcsec depending on the size of the slit),
- offset but take one more acquisition image to make sure the shift worked fine,
- do not offset but exit from the acquisition procedure (if the offset was very small, say <0.1 arcsec, for example),
- abort the entire OB.
The FITS file is named EFOSC_AcqMOS.fits
Image through the MOS mask
Usually, aligning the slitlet mask using the 3 reference star works very well. However sometimes observers tend to live on the edge by choosing stars on the CCD margins (where the distortion is greater) and/or putting targets at the edges of the slits. On such occasions some of the targets may not fall on the corresponding slitlets, especially when narrow slits have been punched (we have never had this problem for the wider slitlets). In order to be sure that all or at least the crucial objects are on the slits, especially for long integration spectra we recommend that observers take an exposure of the field through the slit after the acquisition template and before the spectral templates.
This is a regular Spectrocopy template (Efosc_spec_obs_Spectrum) with the starplate set to MOS#n and the Grism set to Free.
After confirming that the objects are all located where they ought to be one can then relax with nary a worry for the next hour or two..... OR decide, if you so wish, to repeat the acquisition procedure all over again!
The MOS science spectroscopy templates are identical to the ones for long slit spectroscopy except that one uses a MOS mask for a starplate instead of a long slit. Note that there is a new template (Efosc_spec_obs_MOS) that is identical to the long-slit one, but that asks for the name of the .msk mask file. This then uses the slitlet positions from the .msk file to provide these positions in the FITS header of the output spectra. This is primarily for the purposes of pipeline reduction of the spectra, but is recommended for all visitors so that this information is available in the archived data.
A typical MOS observing block includes an acquisition template followed by an image of the field through the slit and finally science spectra. Note that one can combine templates with different grisms. The above example shows an OB comprising a sequence of:
- An acquisition sequence for Mos#4 with a R filter, a 45 second exposure and a rotator offset angle of 40 degrees (i.e. slit position angle = 40 + 90 = 130 degrees)
- Note : the rotator offset angle has to be set to the value used for the pre-image.
- An image through the MOS mask Mos#4 with a 45 second exposure and R filter.
- 1 spectrum of 1200 seconds with Mos#4, Grism Gr#12, normal readout, no filter and 2x2 binning; followed by
- 1 spectrum of 1800 seconds with Mos#4, Grism Gr#7, normal readout, no filter and 2x2 binning.