OSIRIS (Optical System for Imaging and low-Intermediate-Resolution Integrated Spectroscopy) is an imager and spectrograph for the optical wavelength range, located in the Nasmyth-B focus of GTC. Apart from the standard broad-band imaging and long-slit spectroscopy, it provides additional capability such as the narrow-band tunable filters imaging, charge-shuffling and multi-object spectroscopy.

OSIRIS covers the wavelength range from 0.365 to 1.05 µm with a total field of view of 7.8 x 8.5 arcmin (7.8 x 7.8 arcmin unvignetted), and 7.5 x 6.0 arcmin, for direct imaging and multi-object spectroscopy respectively. The OSIRIS User Manual can be found here.

The following table summarizes the available modes and features for imaging and spectroscopic observations with OSIRIS.

Imaging Spectroscopy
Broad Band Imaging LongSlit Spectroscopy
Medium Band Imaging: SHARDS Filters Multi-Object Spectroscopy
Narrow Band Imaging: Tunable Filters
Fast Photometry
Frame Transfer Photometry

Back to top


icon Index


icon Instrument Features

The following table summarizes some basic parameters of the instrument.

Spectral range 3650-10000 A
F.O.V. 7.8' x 8.5' (imaging; 7.8' x 7.8' unvignetted)
Plate Scale 0.254" (imaging and spectroscopy). Standard OSIRIS observing modes use 2 x 2 binning, hence plate scale can be reduced if needed.
Detector 2 x 2048 x 4096 Marconi CCD42-82 (with a 9.4" gap between them)
Pixel Size 15 µm/pix
Detector Quantum Efficiency (QE) 50% (400 nm), 90% (600 nm), 80% (800 nm), 40% (900 nm)
Image quality EER80 <0.3" (Imaging mode). Distortion <2% in all the detector

Back to top


icon Instrument Detector

The detector of OSIRIS consists of a mosaic of two Marconi CCD42-82 (2048 x 4096 pixels) with a 37 pix (binned) gap between them. The single pixel physical size is 15 µm, which corresponds to a scale of 0.127" in on the sky. However, OSIRIS standard observing modes use 0.254" binned pixels.

The OSIRIS vignetting is significant in the first 250 pixels of CCD1 and the first 250 pixels of the bottom of the image in both CCDs (though this is less prominent). Therefore one needs to be careful when doing photometry of stars near these areas. This vignetting defines a maximum unvignetted field of 7.8 x 7.8 arcmin

Below we show an image with the default pointing positions for all the OSIRIS observing modes: Broad Band imaging (1), Longslit Spectroscopy (2), and Tunable Filter Imaging (3) (the Tunable Filter's center and the MOS reference pointing are also shown).


OSIRIS CCDs mosaic.

The distance between the CCDs (gap) is 9.4" (approximately 37 pixels). This distance is measured in the central area of the CCDs, because it changes slightly from top to bottom (the CCD2 is inclined by about 0.04 degrees from the vertical direction). The CCD2 also is shifted about 0.5" (= 2 pix) down respect to the CCD1.

Back to top


icon Detector Setup

The CCDs control system offers a wide range of readout modes and gain settings, but for the time being the standard observing mode for Imaging and Spectroscopy is shown in the table below. In this mode the detector linearity is guaranteed up to the full 16 bits signal maximum (as well as in the slow readout mode). Linearity curves are shown below. Read noise is better than 5 electrons standard and slow readout modes.

The acquisition mode is generally used for test images. This mode has a significantly high noise pattern so it is not suitable for scientific cases, and also detector linearity regime is not garanteed up to saturation level (in this readout mode linearity is only guaranteed up to 45,000 ADUs). The following table gives an overview of the main characteristics of the OSIRIS CCD readout modes.

Imaging / Spectroscopy
Slow Acquisition
Readout configuration CCD1+CCD2_A CCD1+CCD2_A CCD1+CCD2_A
Readout velocity 200 kHz 100 kHz 500 kHz
Gain (e-/ADU)* 0.95 1.18 1.46
Binning (X x Y) 2 x 2 2 x 2 2 x 2
Readout time 21 sec 42 sec 7.8 sec
Actual readout noise ~4.5 e- ~3.5 e- ~8 e-

(*) These gain values correspond to CCD2, values for CCD1 are 5% smaller than these.

Note that if you wish to use readout mode other than the standard one (200 kHz), the time for taking the night-time calibrations will be charged to the project. Also note that for non-standard modes data quality is not guaranteed.


OSIRIS CCDs linearity curve for the standard readout mode.



OSIRIS CCDs Quantum Efficiency curve.


Back to top


icon OSIRIS Standard Pointing Positions

For the main observing modes of OSIRIS a reference position on the CCD has been adopted as the location where targets will be placed. These locations avoid blemished on the detector and they correspond to the coordinates introduced by the PI in the Phase-2. Alternative locations can be used, but this should be explicitly identified in the Phase-2 form.

  • Broad Band Imaging: The standard pointing is at the CCD2 pixel (256,1024) to maximize the available FOV and in order to avoid possible cosmetic effects, which are more abundant in the CCD1. The coordinates introduced by the PI in the Phase-2 will be positioned at this central pixel.


    Example of a target acquired in the standar pointing for OSIRIS Broad Band Imaging mode.


  • LongSlit Spectroscopy: Objects are centered on the slit at the coordinate X=250 of the CCD2. This position minimize the amount of cosmetic effects of the CCD2 compared to those on the CCD1. On this area the distortion of the spectra is very low and sufficiently far from the central gap in order to allow a good sky subtraction in the extraction of the spectrum. When observing with OSIRIS in Long Slit Spectroscopy mode, an acquisition image and a throughslit image are usually provided to the user. During the observation, after the acquisition image is obtained with the target placed at the pixel X=250 in CCD2 (that is assumed to be the standard pointing position for this observing mode) an iterative process for slit alignment is needed (usually a small offset to improve the target centering), and no intermediate products of this process are included in the data delivered. For this reason the coordinates for the target in the acquisition and troughslit images can be slightly different (never more than a pixel). It's recommended always to use the troughslit image for checking the proper centering of the target, and to use the acquisition image only for ensuring the correct field acquisition and target identification.


    Example of a target acquired in the standar pointing for OSIRIS LongSlit Spectroscopy mode (above) and the corresponding spectrum position after the observation (below).


  • Tunable Filter Imaging: The position of the objects in the Tunable Filter observing mode depends on the requirements of the PI since the value of the wavelength changes with the object's position in the FOV. The PI must indicate, in the Phase-2 form, the coordinates to which the telescope will be pointing and the CCD pixel position for this observations. If there is no specific location at the Phase-2, the pointing will be done at 15 arcsecs of the optical center of the system pixel (50, 976) at the CCD2.


    Example of a target acquired in the standar pointing for OSIRIS Tunable Filter Imaging mode.


  • Multi-Object Spectroscopy: Pointing coordinates will refer to gap center, -pixel (6,1024) in CCD2 coordinates-, but in this case the file coming from the Mask Designer Tool will include that information and no further action within Phase-2 tool is needed.

Note that all the coordinates are referred to standard CCD pixels (that is, using 2 x 2 binning).

Back to top


icon Observing Modes

icon Broad Band Imaging

OSIRIS allows broadband imaging over a FOV of 7.8 x 8.5 arcmin (7.8 x 7.8 arcmin unvignetted) covering the full spectral range from λ=3650A to λ=10000A, with a high transmission coefficient in particular at longer wavelengths. The full spectral range is covered by the Sloan system broadband filters: u' (λ3500 A), g' (λ4750 A), r' (λ6300 A), i' (λ7800 A), z' (λ9250 A).

The broad-band photometric zero-points for OSIRIS are given in the following table, as well as estimates of the sky brightness ADUs/s/pix measured at a Elevation 55 deg in the standard OSIRIS readout mode. Although ETC predictions for sky brightness at the ORM are accurate enough, it is recommended to use the values from the table below for a quick estimation of the sky background counts in long exposed images.

Filter λc(0º) [Δα(0º)] nm Zeropoint (mag) Sky Brigthness
Sky Brigthness
Sky Brigthness
Transmission (0º)
u' 350.0 [60.0] 25.75 ± 0.09 15 10 1 --- / ---
g' 481.5 [153] 28.82 ± 0.07 250 150 25 Image / Table
r' 641.0 [176] 29.29 ± 0.07 350 305 90 Image / Table
i' 770.5 [151] 28.85 ± 0.05 290 265 160 Image / Table
z' 969.5 [261] 28.23 ± 0.07 400 350 325 Image / Table

Standard extinction coefficients for the ORM can be found here

Daily monitorizing of the aforementioned OSIRIS broadband zeropoints yields to a decrease of 0.5-0.7 mag due to presence of thick cirrus (spectroscopic nights), that shows the excellent sensitivity of the instrument even in bad observing conditions. Unusual presence of dust in the atmosphere above the observatory also produces a slightly decrease of this sensibility, but typically no larger than 0.2-0.3 mag respect to the standard zeropoint values. Here can be found the complete table with the OSIRIS zeropoints values, including the night conditions, for the full dataset of OSIRIS observations. These can be used for calibration purposes (see details on this contribution).

The graph below show the zeropoint values for different observing nights with clear and/or photometric conditions.


Evolution of OSIRIS broadband zeropoints with time (only considering clear/photometric nights) since the beginning of the GTC operation. Each year covers the interval between consecutive dashed lines.

For improving the photometric calibration of OSIRIS data, a detailed GTC PSF is available here. This has been produced courtesy of Drs. Trujillo and Fliri (IAC) to be offered to GTC users community.


Back to top


icon Broad Band Detection Efficiency

The graph below shows the overall photon detection efficiency of GTC and OSIRIS in each of the Sloan filters.

BB efficiency

OSIRIS detection efficiency with Sloan filters.


Also, the following plots shows the limiting magnitudes with OSIRIS Sloan filters for getting S/N=3 as a function of the exposure time, assuming dark conditions, seeing = 1.0 arcsec, and airmass =1.2. A detailed view for exposure times lower than 1.0 h is shown in the lower panel.

BB efficiency BB efficiency

Limiting magnitudes achievable in OSIRIS Broad Band Imaging mode as a function of time. The lower panel shows a zoomed view of the upper panel for exposure times lower than 1 h.


Back to top


icon Broad Band Sky Flats

OSIRIS Sky Flats are obtained observing in a selected set of Blank fields containing few (bright) stars, that have been produced specifically for GTC (the selected list of GTC blank fields is given here). This list was compiled using the TESELA tool developed under the umbrella of the Virtual Observatory and where you can also make your own searches. Full details about the search tool for GTC blank fields can be found in Jiménez-Esteban et al. (2012, MNRAS, 427, 679).

The flat fielding homogeneity in each of the OSIRIS Sloan filters is better than 2.5% over the full unvignetted FOV of the instrument, except in Sloan u', where fluctuations up to 5-6% respect to the mean value (caused by the filter coating) are found as measures from many twilight flat fields. Day to day fluctuations in the flatfields are less than 0.05% , and less than 0.1% week to week. Hence, Sky Flat fields obtained with OSIRIS are well usable up to within a week before or after the observations.

Comparisons with SuperFlats derived from GTC scientific observations during bright time (Sky background values around 15,000 - 20,000 ADUs on average) show no variations with respect to the standard Sky Flats up to such a low level as 0.01%, hence they can be considered practically identical for scientific purposes. These percentage variations are measured globally, while of course locally, due to dust particles that can come and go, the variations may be larger. Moreover, differences between the night-sky and the twilight spectrum may result in suble flat fielding differences.

By combining large series of Sky Flats in each filter, obtained during the GTC scientific operations, we constructed a series of MasterFlats frames that can be retrieved here (only Flats for the last two years are shown, but even older Flats can be retrieved under request). Flat fields were all obtained by using the GTC automatic sequence for SkyFlats, that allows to get a series of flats with exposure times always larger than 1 s (to minimize possible photometric effects due to OSIRIS shutter) and a maximum exposure time of about 20 s (where the detection of stars is notable), with an average of 25,000-40,000 ADUs in each individual image. Rejection parameters were chosen accordingly to as much images as possible in each filter. MasterFlats are available separately for each CCD of OSIRIS (as they have a slightly different gain and bias level).


Master Flats
Jan-Mar 2015 Apr-Aug 2015 Sep-Dec 2015 Jan-Mar 2016 Apr-Aug 2016 Sep-Dec 2016
MasterFlat u' MasterFlat u' MasterFlat u' MasterFlat u' MasterFlat u' MasterFlat u'
MasterFlat g' MasterFlat g' MasterFlat g' MasterFlat g' MasterFlat g' MasterFlat g'
MasterFlat r' MasterFlat r' MasterFlat r' MasterFlat r' MasterFlat r' MasterFlat r'
MasterFlat i' MasterFlat i' MasterFlat i' MasterFlat i' MasterFlat i' MasterFlat i'
MasterFlat z' MasterFlat z' MasterFlat z' MasterFlat z' MasterFlat z' MasterFlat z'


Comparisons between Sky Flats and Dome Flats taken with OSIRIS show that these latter are not as good as Sky Flats for the photometry, due to inhomogeneities in the GTC Dome illumination. Differences up to 10-15% are found in CCD2, although they are as small as 2% in CCD1. Therefore Dome Flats are only recommended for obtaining reliable OSIRIS photometry in CCD1 and as last choice in CDD2. In any case, GTC policy of taking FlatFields in a regular basis produces that a series of Sky Flatfields will be always available within a week of any scientific observation, hence Dome Flats are unnecessary.

Back to top


icon Medium Band imaging with SHARDS filters

From June 2012, the number of filters available for general use with the OSIRIS instrument has been drastically extended thanks to a generous gesture by Dr Pablo Pérez González from the Universidad Complutense de Madrid to make available his private optical filters. Dr Pérez González designed and purchased (using funding from the Spanish Government through projects CSD2006-00070 and AYA2009-07723E) a set of medium-band filters for the SHARDS science program that has been executed on the GTC (for further details on this program, see http://guaix.fis.ucm.es/~pgperez/SHARDS/).

This set consists of 25 filters spanning the wavelength range from 500 to 940 nm with bandwidths from 14 to 34 nm. Interested parties who would like to use of any of these filters should contact Dr Pérez González and GTC to request their use, and write the appropriate credits in any paper that may result from the use of these filters. The main characteristics of these filters are summarized in the following table:

Filter λ (nm) FWHM (nm)
U500/17 500 15


As in the rest of filters used in OSIRIS, the SHARDS filters are placed in the collimated beam and close to the pupil of the instrument, at an angle of 10.5º with respect to the optical axis of the instrument. This causes that the central wavelength depend on the position in the field, with a center of symmetry corresponding to the center of rotation of the instrument, located on CCD1 towards the left side of the field.

This effect, small for the standard Broad Band Sloan filters but has a stronger impact in the operation of the SHARDS filters, as the filter widths are notably narrower than in the case of Sloan filters. Hence this is not a defect of the filters, but due to the design of the OSIRIS instrument that becomes more prominent as filter pass band gets narrower. The central wavelength variation effect is more relevant in the case of medium-band filters such as SHARDS' (and also for the order sorter filters) given that the central wavelength shift from edge to edge of the OSIRIS FoV is of the same order of the width of the filter. Note that the typical shift (14-15 nm) would be 10%-20% for broad-band filters. Summarizing, potentially users of the SHARDS filters set have to take two main aspects into account when using these filters:

  • The central wavelength in the rotation center of the instrument, (approximately pixel 462.5,995 in CCD1), is different from the central wavelength at the standard pointing position for OSIRIS Broad Band imaging (pixel 256,1024 in CCD2). This latter can be as much as 5-6 nm bluer than the central wavelength at CCD1, reaching differences up to 12-14 nm at the extremes of the FOV. Taking into account that most SHARDS filters have a bandwidth as narrow as 17 nm (except in two cases), this produces that the wavelength range observed in OSIRIS CCD1 can be notably different than the one observed in OSIRIS CCD2. Hence, for instance when making use of the whole FOV of OSIRIS for mapping a single emission line probably more than one filter has to be used.
    An example of this effect is shown in the following Figure for U840/17 filter. The central wavelength in CCD1 is 843 nm, while the nominal wavelength for an angle of incidence of 10.5º is 840 nm (approximately at the central gap between CCDs). For CCD2, the central wavelength changes drastically, from 840 nm to 830 nm at the edge of OSIRIS FOV, a value that is 13 nm bluer than the central wavelength in CCD1.



Effect of the dependence of the central wavelength variation with the position in OSIRIS FOV.


  • The wavelength variation over the FOV results in that the sky background is inhomogeneous. This is in particular pronounced when strong sky lines fall within the band. This makes that the sky background subtraction is a critical step in the reduction of data taken using SHARDS filters.

    The following Figure demonstrates this effect. In the left panel, we have the wavelength variation for U687/17 filter, in combination with a sky emission spectrum. As it can be seen, the stronger emission lines fall at the reddest wavelengths, -the ones that are observed in CCD1-, while these same lines are nearly undetected at the bluer wavelengths, that are the ones observed in CCD2. This produces that the sky background will be notable different from one CCD to other, being stronger at CCD1 than in CCD2.
    This effect can be clearly seen in the right panel, where a sky image with U687/17 filter is shown. Note the strong gradient observed in the background level, and how this follows a radial geometry from CCD1 to CCD2. This effect should be taken into account when using the sky flat frames that can show some variability as sky emission itself changes (in any case, this can be properly corrected during the data reduction process).



Effect of the wavelength variation over the FOV in the background level when using SHARDS filters.


For a detailed description on the wavelength change along the FOV when using the SHARDS filters, see Table 1 in Perez-Gonzalez et al. 2013 (http://adsabs.harvard.edu/abs/2013ApJ...762...46P), as well as other details on the data reduction, etc. Also, a complete updated table on the SHARDS filters' characteristics can be found in https://sioux.fis.ucm.es/~pgperez/SHARDS/Filters/TAB_filters/TAB_filters.html


The next Figure shows the overall photon detection efficiency of GTC and OSIRIS with the SHARDS filters. The full table with the SHARDS filters' efficiencies can be retrieved here for flux calibration purposes (see details on this contribution).

shards filters

SHARDS filters efficiency curve.


SHARDS filters are available for its public use without any restrictions, but users must be aware the all the science publications based on the use of SHARDS filters should include the following acknowledgement: "This work is (partly) based on data obtained with the SHARDS filter set, purchased by Universidad Complutense de Madrid (UCM). SHARDS was funded by the Spanish Government through grant AYA2012-31277."


Back to top


icon Narrow Band Imaging: Tunable Filters

Narrow-band imaging in OSIRIS is made possible through the use of a tunable filter (TF), which is in essence a low resolution Fabry-Perot, where fhe filter is tuned by setting a specific separation between the optical plates of the Fabry-Perot.

OSIRIS has two different tunable filters: one for the blue range (BTF: λ 450 nm - 671 nm) and another for the red range (RTF: λ 651 nm - 934.5 nm). The use of the tunable filters implies the utilization of Order Sorter filters (OS) in order to select the wavelength band that avoids confusion between different orders of interference of the Fabry-Perot.

When working with the OSIRIS tunable filters the user needs to take into account two parameters: the observing wavelength and the required FWHM. Once it was selected the wavelength to be tuned, the corresponding OS has to be chosen accordingly as for a given wavelength there is only a single OS to be used. The correspondence with wavelength can be found in the Order Sorter tables, both for the Blue Tunable Filter and the Red Tunable Filter.

Back to top


icon Tunable Filters Optical Center and Available Widths

The TF optical center is located at pixel (1051, 976) of CCD1 (binned coordinates), including the 25 pixels of overscan, within the gap of the CCDs, and 2 pixels away from the right edge of the CCD1, or equivalently, the center of the system lies at pixel (-10,976) of CCD2.

Users should be aware that the wavelength tuning is not uniform over the full field of view of OSIRIS. There is a progressively increasing shift to the blue of the central wavelength (λ0) as the distance to the optical centre (r) increase.

The wavelength (in angstroms) observed with the RTF relative to the optical center changes following the law (González et al. (2014, MNRAS, in press)):

λ = λ0 - 5.04 * r (arcmin) 2

(this is slightly different to the pure geometrical expression; see here for the complete details)

While for the BTF, the change in wavelength (in angstroms) can be described by:

λ = λ0 - 3.80 * r (arcmin) 2

These two expressions are different for each Tunable Filter, as a result of the differences in the coating thickness of the internal reflective coating for both TFs (see OSIRIS user manual for details). In both cases, the expressions are very accurate for any wavelength, and the maximum error is of the order of the tuning accuracy (0.1 - 0.2 nm) across the 4 arcmin radius OSIRIS TF FOV.

The image below show a picture of prominent sky lines taken at 7325 A , which clearly shows the center of the system (pixel -10, 976 at CCD2). The 4 arcmin radius that assures the observations will not have any contamination of other interference orders in the filter is also shown. This is the operative FOV of OSIRIS Tunable Filters.



OSIRIS circular FOV available for TF observations.


It should also be noted that the practical use of the tunable filter is more restrictive than was originally anticipated. For the red-optimized tunable filter (RTF) the minimum achievable width is 1.2 nm for most wavelengths, except for the longest wavelengths where even narrower pass bands can be tuned (see Figure below):

TF range (nm) TF available FWHMs (nm)
λ < 800.0 1.2 < Δλ < 2.0
800.0 < λ < 820.0 1.0 < Δλ < 1.5
820.0 < λ < 840.0 0.9 < Δλ < 1.4
840.0 < λ < 880.0 0.8 < Δλ < 1.3
880.0 < λ < 910.0 0.85 < Δλ < 1.2
λ > 910.0 0.9 < Δλ < 1.2


The tunable accuracy is 0.1 nm in the filter central lambda. The range of operation of the red tunable filter is from 651 nm to 934.5 nm.

In addition to the information of the maximum tunable widths with the TFs as a function of wavelength (see table above); the plot below shows the variation of the available widths as a function of wavelength. The minimum width is 1.2 nm for all the wavelength to avoid contamination due to other orders in a circular FOV of 4 arcmin radius, except for the longest wavelengths where even narrower pass bands can be tuned thanks to an upgrade in the RTF Order Sorters definition (see here for details on this work).



RTF available widths vs. wavelength.



For the blue-optimized tunable filter (BTF) the possible achievable pass bands are narrower than the ones provided by the RT.F Due to the particularities of the BTF only a single pass band is available for each wavelength that avoids contamination from other interference orders over the circular 4 arcmin radius field-of-view. In other words, for each wavelength only one passband FWHM is available. The following table shows the available FWHM for the BTF.

TF range (nm) TF available FWHMs (nm)
448 < λ < 464 0.80
464 < λ < 481 0.85
481 < λ < 503 0.80
503 < λ < 522 0.50
522 < λ < 543 0.45
543 < λ < 584 0.50
584 < λ < 610 0.70
610 < λ < 638 0.90
638 < λ < 671 * 1.10

(*) Note that redwards of 651 nm the RTF can be used with higher efficiency and higher bandwidths.

The tunable filter accuracy is 0.1 - 0.2 nm in the filter central wavelength. The range of operation of the blue tunable filter ranges from 448 to 671 nm.

Back to top


icon Variation of the TF line with Rotator Angle

The calibration of both OSIRIS TFs is highly dependent on the angle of the rotator. We can find differences of up to 40 steps (~0.8 nm) between two rotator positions (see fig below). In order to avoid this we define the useful range (-160°<θ<-40º and 50º<θ<160º). This range ensures a stable calibration accuracy ±0.1nm and, if the rotator is moving less than 10°, the calibration can be considered virtually unchanged, with the precision given by the self-calibration (± 0.01 nm = 1 bit).

The global variation in the blue tunable filter is roughly inverse of the behavior of the red tunable filter, as gravity-induced flexure in the reference capacitors is the opposite given their opposite location in the filter wheel.

It is possible to predict, using the coordinates of the object and the position angle (PA) on the sky, the position of the rotator to a specific time in order to ensure that the observations are performed within the useful range. If the PI do not required a particular angle on the sky (IPA =150.54º±PA) then the IPA that best fit the observations would be chosen.


Variation of the TF tuning (defined by an internal parameter Z) with the rotator angle.

Back to top


icon Tunable Filter Observations: Dithering, Ghosts and TF Flat Fields

Observing with a Tunable Filter recquires some previous knowledge of this kind of devices. We strongly recommend to potentially OSIRIS TF users to read Section 3 in the OSIRIS User Manual for a complete description of different observing strategies and problems that are usual in this type of observations.

Tunable Filter observations produce ghosts, as a result of the light reflected by the detector that enters the etalon and is sent back to the detector. For this reason, the ghosts are symmetric with respect to the centre of the Tunable Filter, that correspond with pixel (-10,976) in CCD2. Diametric ghosts can be easily removed by the classical dithering procedure since moving the image in one direction shifts its ghost in the opposite direction with respect to the TF optical centre. Hence, when stacking the images tanking as reference the image of the target, all ghosts will fall in diiferent pixels and can be removed with average sigma clipping or similar algorithms.

Only very bright, usually saturated sources, generate ghosts. For OSIRIS RTF, the integrated flux of the ghost images is less than 1.7% of the integrated flux of the main source. Hence, for RTF observations, unless very bright sources are in the FOV and their ghosts could spoil the image of the target, there are no need to worry about it. However, for OSIRIS BTF, as a result of a larger thickness of the internal reflective coating, the integrated flux of the ghost image can be up to 15% of the integrated flux of the main source for wavelengths bluer than 610 nm. For this reason, for BTF observations dithering procedure is mandatory to remove these ghosts.



Tunable Filters diametric ghosts effect seen on scientific images (left). Ghosts dissapear after applying a classical dithering procedure (right).

During the normal operation of OSIRIS at GTC, Tunable Filter Flat Fields for the TF observations are obtained using Dome Flats, with the TF tuned to the same wavelengths of the science observations (usually from 3 to 5 flatfields at each wavelength). It is nearly impossible to get a series of enough Sky flatfields at all the wavelengths covered in a single OB due to time limitations, and this is even worse when two or more programs with TF imaging are observed in a single night.

Some features that can be observed at some wavelengths in the TF Dome Flats are also present in Sky Flat images, hence they have nothing to do with the dome illumination but with the Order Sorter Filters and their wavelentgh displacement along the OSIRIS FOV. We consider that the dome flat fields are adequate for their purpose and little is gained from using TF sky flat fields. Moreover, the features can be also noted in the science images, for example when an artificial dithering is used to produce a sky image from the raw image. Once this sky image is subtracted to the original image, all the sky features are corrected with high accuracy.



TF Dome and Sky Flat at 715 nm, showing the same features.



TF Dome Flat at 660 nm, and a sky image at 660 nm obtained from an artificially dithered raw science image (to enhance the background level).

Back to top


icon Blue Order Sorter Filters


Filter ID λ (nm) FWHM (nm) TF λ range (nm) Transmission 0º
f451/13 450.7 13.1 448 - 458 Image / Table
f454/13 454.3 13.2 458 - 461 Image / Table
f458/13 457.9 13.3 461 - 464 Image / Table
f461/13 461.5 13.4 464 - 468 Image / Table
f465/13 465.1 13.5 468 - 473 Image / Table
f469/14 469.0 14.0 473 - 476 Image / Table
f473/14 473.1 14.1 476 - 481 Image / Table
f477/14 477.2 14.2 481 - 484 Image / Table
f481/14 481.4 14.4 484 - 489 Image / Table
f486/14 485.6 14.5 489 - 494 Image / Table
f490/15 490.0 15.1 494 - 498 Image / Table
f495/15 494.7 15.2 498 - 503 Image / Table
f499/15 499.5 15.4 503 - 506 Image / Table
f504/16 504.2 15.5 506 - 511 Image / Table
f509/16 509.1 15.7 511 - 516 Image / Table
f514/16 514.0 15.8 516 - 522 Image / Table
f519/16 519.1 16.5 522 - 528 Image / Table
f525/17 524.6 16.7 528 - 533 Image / Table
f530/17 530.1 16.8 533 - 538 Image / Table
f536/17 535.7 17.0 538 - 543 Image / Table
f542/18 541.6 17.8 543 - 550 Image / Table
f548/18 547.8 18.0 550 - 556 Image / Table
f554/18 554.1 18.2 556 - 562 Image / Table
f561/19 560.8 19.0 562 - 569 Image / Table
f568/19 567.9 19.2 569 - 576 Image / Table
f575/19 575.0 19.5 576 - 584 Image / Table
f583/20 582.6 20.4 584 - 593 Image / Table
f591/21 590.5 20.7 593 - 600 Image / Table
f599/22 599.0 21.8 600 - 610 Image / Table
f608/22 607.9 22.1 610 - 618 Image / Table
f617/23 617.4 23.3 618 - 628 Image / Table
f627/24 627.4 23.7 628 - 638 Image / Table
f638/25 638.0 25.0 638 - 649 Image / Table
f649/25 649.2 25.5 649 - 658 Image / Table
f661/27 661.2 27.0 658 - 671 Image / Table

Back to top


icon Red Order Sorter Filters


Filter ID λ (nm) FWHM (nm) TF λ range (nm) Transmission 0º
f657/35 657.20 35.0 649 - 660 Image / Table
f666/36 666.84 35.5 660 - 670 Image / Table
f680/43 680.21 43.2 670 - 685 Image / Table
f694/44 694.38 44.0 685 - 695 Image / Table
f709/45 708.84 44.9 695 - 710 Image / Table
f723/45 723.29 45.2 710 - 725 Image / Table
f738/49 737.98 46.1 725 - 735 Image / Table
f754/50 754.25 49.6 735 - 755 Image / Table
f770/50 770.57 49.7 755 - 770 Image / Table
f785/48 785.58 47.6 770 - 788 Image / Table
f802/51 802.02 51.3 788 - 803 Image / Table
f819/52 819.03 52.4 803 - 818 Image / Table
f838/58 838.57 57.8 818 - 845 Image / Table
f858/58 858.21 57.9 845 - 860 Image / Table
f878/59 878.23 59.3 860 - 885 Image / Table
f893/50 893.21 49.6 885 - 900 Image / Table
f902/44 902.40 40.1 900 - 910 Image / Table
f910/40 910.64 40.5 910 - 912 Image / Table
f919/41 918.95 40.8 912 - 920 Image / Table
f923/34 923.85 34.2 920 - 925 Image / Table
f927/34 927.94 34.4 925 - 930 Image / Table
f932/34 932.05 34.5 930 - 935 Image / Table


Back to top


icon Medium Band Imaging with TF Order Sorters

The OS filters can also be used for direct image observations. Measurements made during the commissioning of OSIRIS provided zeropoint values for some of the most significant OS. These values are given in absolute magnitudes (mAB) at airmass = 1, using these measures of spectrophotometric standard stars

OS mAB(standard) Zeropoint
OS657 15.25 ± 0.05 27.86 ± 0.09
OS666 15.27 ± 0.05 27.72 ± 0.02
OS709 15.35 ± 0.05 27.89 ± 0.05
OS770 15.45 ± 0.10 27.73 ± 0.02
OS858 14.35 ± 0.05 27.58 ± 0.03
OS902 14.48 ± 0.05 27.07 ± 0.09


The graph below shows the overall photon detection efficiency of GTC and OSIRIS with the TF Order Sorters

OS efficiency

OSIRIS detection efficiency with the TF Order Sorters.


The OS high inclination makes impossible to use two contiguous OS to produce a single narrower filter. This is due to internal reflections occurring in different layers of the filters that lead to the formation of ghosts. Their intensity and position in the field vary depending on the combination of filters that is, the position of the rotator, etc. This mode of operation is not offered.

Back to top


icon Tunable Filter Detection Efficiency

The graph below shows the overall photon detection efficiency of GTC and OSIRIS with the Tunable Filters. The full table with the RTF efficiencies can be retrieved here for flux calibration purposes (see details on this contribution).

overall TF efficiency

OSIRIS detection efficiency with Tunable Filters.

Back to top


icon Fast Photometry

For fast photometry, a mask with a 7' x 3" slit is used; the fast photometry mask. This slit is placed in one of the detector edges. The images are obtained while the shutter remains open, and after each exposure the charge is shifted a number of lines, at least the equivalent to the width of the slit (12 pix approximately in standard 2 x 2 binning mode). The minimum exposure time and photometric accuracy is determined by the time needed for the vertical displacement of the charge on the CCD (typically by 50 μs/row). For example, to displace 12 binned pixels requires 0.0012 s (this will be the minimum exposure time allowed in this configuration), hence for exposure times larger than 0.1 s the photometric accuracy will be better than 1%. For larger pixel shifts, the minimum exposure times required will increase accordingly.

Standard configurations for this mode allows obtaining up to 133 consecutive images before readout (see Figure below as an example). However, more conservative numbers are recomended (60-80 consecutive images per frame) in order to avoid possible flux contamination from the previous images along the series. The fast photometry standard mode means using the same broad-band (Sloan) or medium-band (SHARDs) filter throughout the observation, or a tunable filter adjusted to a fixed wavelength, as no delays due to filters exchanging or TF tuning are possible as this is a shutterless mode.

The only delay in the series would be imposed by the readout time once the detector is filled with the individual images, plus the 4 s delay needed for clearing/configuring the detector. In standard OSIRIS readout mode (200kHz) this time will be 25 s. However, this time can be reduced by defining a single readout window limiting the extent of the readout area along the slit. Also the higher readout of 500 kHz can be used to reduce the readout time to 12 s (and even less with windowing). However, this readout speed is a non-standard operation mode in OSIRIS and its performance is not guaranteed.

As the fast photometry mask is restricted to a 3" size in the vertical direction, good seeing conditions are required for its use. It is also possible to use the longslit masks as fast photometry masks, allowing up to 10" of vertical aperture. However, in this case, only half of the detector will be completed before readout since the position of the long slit is centered in the FOV.

Due to the high flexiblilty and the multiple possible combinations in using this observing mode, fast photometry with OSIRIS is only offered in visitor mode. Prior to defining your observing proposal it is strongly recommended to contact GTC staff astronomers to evaluate the optimum mode of operation.

fast photometry mask
Image taken in fast readout mode, where many individual narrow-strip images are combined in a single detector readout.


Back to top


icon Frame Transfer Photometry

Frame transfer capability in OSIRIS uses a half-field mask (see Figure below) with an accessible FOV of 7' x 3.5', approximately. In this operation mode, only half of the detector is exposed while the other half of the detector is being read out. The minimum exposure time allowed is now imposed by the time required to displace the charge over half the number of the detector lines (0.1 s) plus the readout time of this area. In standard OSIRIS readout mode (200 kHz) this time will be 8.25 s, which can be decreased to 4.3 s using 500 kHz readout mode. However, using this higher readout speed is a non-standard operation mode in OSIRIS and its performance is not guaranteed.

Frame transfer standard mode means using the same broad-band (Sloan) or medium-band (SHARDs) filter throughout the observation, or a tunable filter adjusted to a fixed wavelength, as no delays due to filters exchanging or TF tuning are possible since the shutter remains opened. The difference of this mode with respect to the fast photometry is that the sampling interval between exposures is smaller since it is possible to expose while reading out, obtaining a continuous series of temporal data.

A possibility for decreasing the minimum exposure time is to use a single readout window. In this case, the minimum exposure time will be determined by the size of the window combined with the readout time and the time used for skipping the remaining pixels. This is independent on the window placement, hence by knowing the desired window size and the readout speed, the final minimum exposure time can be determined.

There are a lot of possible combinations depending on the desired FOV and sampling requested. Due to the high flexibility and the multiple possible combinations in using this observing mode, Frame Transfer in OSIRIS is only offered in visitor mode. Prior to defining your observing proposal it is strongly recommended to contact GTC staff astronomers to evaluate the optimum mode of operation.

frame transfer mask
Image taken through the frame-transfer mask, showing the half-field blocking.


Back to top


icon Overall efficiency in OSIRIS imaging modes

The following diagram summarizes the overal system efficiency of OSIRIS in imaging mode, as a function of wavelength. Note that in the curves for the tunable filter order sorters, the efficiency of the tunable filter itself is not included in the curve.

overall efficiency

OSIRIS detection efficiency in the different imaging modes.

Back to top


icon LongSlit Spectroscopy

OSIRIS allows longslit spectroscopic observations by using slits of 7.4' in the spatial direction and with different choices for the widths: 0.4", 0.6", 0.8", 1.0", 1.2", 1.5", 1.8", 2.5", 3.0", 5.0" and 10". The spectral direction in OSIRIS coincides with the vertical direction in the detector, hence OSIRIS spectra are not affected by the gap between both CCDs.

Due to the obscuration present in one of the edges of OSIRIS FOV, and the manufacture process in producing the slits, the maximum distance allowed for placing two targets in the same slit configuration is 7.4 arcmin. For a proper sky subtraction, however, no distances larger than 7.0 arcmin are recommended in order to get enoguh pixels for the background estimation on both sides of the targets. Likewise, if offseting is required during the observation, the maximum distance to the targets has to be estimated accordingly (for example, a maximum distance of 6.5-6-7 arcmin between the targets ia a good approximation for this kind of observations).

OSIRIS has a wide variety of grisms and volume-phased holographic gratings (VPHs) covering low to intermediate resolutions, from R=300 up to R=2500. The following table summarises the resolutions and spectral ranges available. Resolutions and dispersions are measured at λc(A) for a slit with of 0.6". Dispersions correspond to binned pixels, that is the standard operation mode, while the physical pixels (unbinned) dispersions are half of those listed in the table.


ID λc(A) λ range (A) D (A/pix) Resolution Peak Efficiency Type Efficiency
R300B 4405 3600 -7200 4.96 360 70% Grism graph
R300R 6635 4800 - 10000 7.74 348 70% Grism graph
R500B 4745 3600 - 7200 3.54 537 68% Grism graph
R500R 7165 4800 - 10000 4.88 587 67% Grism graph
R1000B 5455 3630 - 7500 2.12 1018 65% Grism graph
R1000R 7430 5100 - 10000 2.62 1122 65% Grism graph
R2000B 4755 3950 - 5700 0.86 2165 87% VPH graph
R2500U 3975 3440 - 4610 0.62 2555 70% VPH graph
R2500V 5185 4500 - 6000 0.80 2515 80% VPH graph
R2500R 6560 5575 - 7685 1.04 2475 80% VPH graph
R2500I 8650 7330 - 10000 1.36 2503 80% VPH graph


Back to top


icon Arc Line maps

A complete arc linelist for all the lamps available at the OSIRIS Instrument Calibration Module (ICM) can be retrieved here. As addtional help in the wavelength calibration, some sky spectra at the resolutions covered by OSIRIS are available: R300, R500, R1000, R2000, and R2500.

Spectral Line List and Calibration ARCs
LineList & ARC
(HgAr, Ar, Ne, Xe)
LineList & ARC
(HgAr, Ar, Ne)
LineList & ARC
(HgAr, Ar, Ne)
LineList & ARC
LineList & ARC
(HgAr, Ar, Ne, Xe)
LineList & ARC
(HgAr, Ar, Ne, Xe)
LineList & ARC
(HgAr, Ar, Ne, Xe)

LineList & ARC
LineList & ARC
LineList & ARC
LineList & ARC


For some of the OSIRIS R2000/2500 VPHs long exposed lamp images are required to obtain enough signal which allows a good line identification. For this reason, here can be retrieved a master arc collection with the arc lamps images obtained with OSIRIS R2000/2500 VPHs with the 0.6" slit:

Master ARCs
Xe & Ne & HgAr
Xe & HgAr
Xe & Ne & HgAr
Xe & Ne & HgAr
Xe & Ne & HgAr


All the OSIRIS grisms show some minor ghost effects in the arc images. Those ghosts are due to internal reflections within the grisms, and can be clearly identified as the curvature in these spectra clearly differ than that shown by the arc lines. The intensity of those ghosts is negligible, and do not affect neither to the line identification nor to the science images.



OSIRIS grisms ghost effects shown in the arc images.

Back to top


icon VPHs R2000/2500 ghosting

OSIRIS R2000 and R2500 VPHs suffered from a quite strong ghost image of the spectrograph slit, that was corrected by adapting the baffling of the optics. This has been a delicate and precise task in order to avoid any vignetting that could be detrimental to the throughput. The ghosting is now nearly fully removed; what remains is a very small fraction of the total flux of the object that normally will have a negligible impact on the quality of the spectra.

The ghost is negligible in the R2500I and R2500R VPHs, while in R2000B, R2500U, and R2500V the ghost is only noted in the spectral flat-field images and arc lamp frames, where a very faint slit image can be observed superimposed on the spectral flat / lamp arcs. The approximated position for those ghost images are: R2000B : from pixels Y = 988 to 996; R2500U : from pixels Y = 980 to 988; R2500V: from pixels Y = 992 to 1000.

There are no problems for the line identification as the intensity of the ghost is far below the average of the counts for the spectral lines. However, users must be aware when obtaining the flat-fielding correction in the pixels range described above, and only for R2000B, R2500U and R2500V. In the science images the effect is unrelevant for the complete set of VPHs.

Back to top


icon Second Order Contamination

All the OSIRIS red grisms/VPHs (R300R, R500R, R1000R, R2500R, and R2500I) are used in combination with an spectral order sorter filter (GR), which cuts out the light blue-ward from ∼495 nm. However, there is a slight contamination in the spectrum due to the second order, as the spectral order sorter filter doesn't block completely the contribution for wavelengths lower that the defined cut level (see Figure below). Hence, there is a distinguishable contribution for wavelengths at 480 – 490 nm, whose second order may contribute somewhat at 960 nm – 980 nm, depending of the source spectral distribution.

Below are two example spectra of the flux standard star PG1545+035 taken with a 2.5 arcsec slit and 300 secs exposure time. The low-resolution spectrum with the R300R grism shows the first order of dispersion well centered on the CCD. Also visible is the zeroth order on the left. Although the spectral order sorting filter was used, which cuts out the light blueward from ~495nm, the second order spectrum is visible on the right-hand side of the graph. The second example spectrum shown was taken under identical conditions but using the R1000B grism. Here only the first-order light is detected.


Example of a OSIRIS spectrum taken with a red grism, showing the contribution of the second order contamination in the reddest wavelengths (left). Blue grisms, however, are not affected by this effect (right).


Back to top


icon Flexures

OSIRIS allows a very stable spectral calibration, with no significant drifts with rotator position (<0.3 pix) thanks to its active collimator. Therefore, the calibrations for each observation can be taken either at the beginning or at the end of the night regardless of the orientation of the instrument when the science observation is carried out. Next figure shows an example of the wavelength shift as a function of rotator angle, for different spectral lines, with R1000B grism.


Shift in the spectral direction (Y) for the arcs emission lines with rotator position for OSIRIS R1000B grism. The more extreme variations are less than 0.3 pix (binned).

Back to top


icon Fringing

The measured value of fringing in the OSIRIS CCD is < 1% for wavelengths < 9000 A, and 5% for wavelengths > 9300 A (with a slightly increase to 7% at higher resolutions, R=2500), so it is relevant only at higher wavelengths (and in the range z' in imaging mode). Next figure shows an example of fringing vs. wavelength obtained with OSIRIS R500R.


OSIRIS fringing vs wavelength, obtained with the R500R grism.

Back to top


icon Spectroscopic Photon Detection Efficiency

The overall photon detection efficiency in spectroscopic mode has been measured using a spectrophotometric standard star through a wide slit, as a function of wavelength and for different grisms. The results are displayed in the following graphs showing the end-to-end overall percentage detection efficiency, and the limiting magnitudes (AB) for obtaining S/N=5 in 1 h of integration time with OSIRIS@GTC for the complete set of grisms/VPH available (assuming 1 arcsec seeing, dark night and airmass=1.2).

spectral efficiency

OSIRIS detection efficiency in spectroscopic mode.


spectral efficiency

Limiting magnitudes for obtaining S/N=5 in 1 h of integration time with OSIRIS.

Back to top


icon OSIRIS Overall Efficiency

The following diagram summarizes the overall photon detection efficiency in both imaging and spectroscopic modes of OSIRIS (for the higher resolution grisms/VPHs)

spectral efficiency

Overall detection efficiency in OSIRIS.

Back to top


icon OSIRIS Data Reduction Tools

Some useful data reduction tools can be retrieved here to process OSIRIS data:

  • OOPs (official pipeline), by A. Ederoclite (aederocl [at] cefca.es). It covers the complete set of OSIRIS observing modes. This can be retrieved here, where it's described its use and functionalities.
  • GTCMOS pipeline, by D. Mayya (ydm [at] inaoep.mx). Produced for reducing longslit and MOS data.

Important note: A. Ederoclite and D. Mayya have kindly provided access to the reduction packages to the general OSIRIS user community. However, the kind of support they can provide is only in a best effort basis.

Back to top


icon OSIRIS contact persons at GTC


contact email @ gtc.iac.es
David García Álvarez - OSIRIS Instrument Specialist (main contact) david.garcia
Gabriel Gómez Velarde - OSIRIS MOS Specialist gabriel.gomez

Back to top


icon Useful Documents

Back to top


Last modified: 24 January 2017