Gran Telescopio CANARIAS

Observatorio del Roque de los Muchachos
La Palma

The advanced data products (ADP) at GTC are prepared using automated data reduction pipelines and delivered with the raw data and calibration files to the users. The files are easily recognized by all having the prefix ADP_. Below we list and describe the products for the different instruments and modes that are currently being delivered. The data reduction pipelines are continously being updated and optimized with pipelines for other modes and instruments also in development. We will inform of relevant news and changes on this webpage. Papers describing the pipelines developed at GTC are also being produced. When published they will be linked here and users will be requested to please reference those papers.

DISCLAIMER: Due to the very varied nature of observations at GTC on the one hand and the automization of the data reduction process on the other, GRANTECAN does not guarantee the science readiness of the advanced data products. In particular absolute calibrations (such as flux calibrations) should always be viewed with caution. There will also be special cases, where the observations does not follow the standard setup and/or workflow of an instrument, and thus the data will not be able to be reduced with the automatic system.

NEWS

July 2026: First deliveries of reduced data for EMIR long slit spectroscopy.

Semester 2025B: First deliveries of EMIR imaging reduced data.

Semester 2025A: Implementation of DRAGO, Data Reduction and Analysis for GTC Operations. First official deliveries of OSIRIS+ BBI and LSS reduced data.

Jan/Feb 2025: Reduced data deliveries are tested for observing blocks with OSIRIS+ BBI.

OSIRIS+

The observing modes included in the automatic data reduction for OSIRIS+ are Broad Band Imaging (BBI) and Long Slit Spectroscopy (LSS). Multi Object Spectroscopy (MOS) is foreseen to be included at a later time.

OSIRIS+ Broad Band Imaging

The reduction of OSIRIS+ BBI is done with SAUSERO, a python pipeline developed specifically to reduce OSIRIS+@GTC broad band imaging. The provided frames are the sky-included reductions of each invidual raw frame and a stacked image if more than one raw frame is taken in each filter. The set of calibrations frames used are 11 bias, 7 skyflats and 1 photometric standard field observation. To automate the process and be able to run many reductions in parallel the astrometrization is done using astrometry.net in local mode with Gaia-DR3 instead of the astrometry.net web service.

The configuration.json file used for the redution is provided with the reduced data to allow for a replication of the reduction on the users own machine. With the same configuration file users can also alter the reduction if they are interested in intermediate products or sky-subtracted frames.

The Table below shows the naming scheme for OSIRIS+ imaging reduced data.

Known Issues

• For various reasons it can happen that no standard star is available for the reduction. In this case the ADP frames will have no flux correction. That means the units will be in adu instead of adu/s and the keywords ZP, EZP, EXT and EEXT in the header wil not be generated. This should only happen in cases where it is previously known that the flux calibration is not needed. If however a flux estimate is wanted, the broad-band photometric zeropoint reference values can be found in the table in the Broad Band Imaging section on the OSIRIS+ page and standard extinction coefficients for the ORM can be found here.

• At the moment SAUSERO is only able to reduce observations done with one of the Sloan broad band filters. That means observations done with medium band SHARDS filters or without a filter will not be reduced. It is foreseen in upcoming versions of SAUSERO to include the capability of reducing frames taken without a filter.

OSIRIS+ Long Slit Spectroscopy

The reduction of OSIRIS+ LSS is done with pypeit, a python package for reduction of astronomical spectroscopic data. The set of calibration frames used are 11 bias, 5 spectral flats, at least two arcs for wavelength calibration depending on the grism and 1 spectro-photometric standard star observation. Due to recurring issues with the auto-classification done by pypeit, in particular with science and standard frames, the pypeit configuration file is not created using the pypeit_obslog command, but instead generated by DRAGO. The configuration file delivered with the reduced data shows the specific setup used.

The data are divided in subfolders by instrument setup (slit and grism) and each setup is reduced individually. Due to the mask support covering the middle of the detector, pypeit considers the OSIRIS+ long slits as two individual slits and for each of these only the brightest object is extracted. The brightest object in lower part of the slit (including pixels 1051,750 as the default for spectroscopy) is in the first extension and the brightest object in the upper part is in the second extension. The flux calibration is done with a sensibility function calculated from a standard star spectrum taken with the same grism in the same night. If no standard star is observed in the same night with the same grism as the science data, the spectra will not be flux calibrated. If more than one spectrum is observed in a given setup within the same observing block, a co-added stacked spectrum will also be included in the delivery. The above-mentioned behavior of pypeit seeing OSIRIS+ LSS data as two slits also applies to the stacking of the spectra. This means that if the observations are done with offsets/nodding that move the target between the two parts of the detector, the stacked spectra will also include the off-position.

The Table below shows the naming scheme for OSIRIS+ LSS reduced data.

The format of the 1D extracted spectra is described here with a table of column descriptions in the Current Data Model. Pypeit also provides an XSpecGUI for easy visualization. When invoked with default parameters it will show the columns OPT_WAVE and OPT_COUNTS of the 1st extension. The files with 1D spectra has the science spectrum in the first extension. The last extension will always include information about the extraction. The format of the co-added spectra are likewise desribed here again with the column description in the Current Data Model section. The files with the 2D extraction has several extensions explained briefly in the table below, for more information see the corresponding pypeit pages.

Known Issues

• For spectra taken with the lower resolution grisms where the spectra doesn't cover the whole detector, the sensitivity function is unreliable and likewise the flux calibration is unreliable. When creating the sensitivity function pypeit masks strong lines and then fits a polynomial to the whole spectrum to cover the strong line wavelengths. This can cause a bad fit at the bluest and reddest wavelengths of the spectrum.

• The flux calibration at wavelengths above 8000 AA with grism R300R cannot be trusted, because the sensitivity function is poorly defined at those wavelengths leading to a flux calibration that is an estimate at best.

• For high snr spectra, the tracing of the spectrum at the edges of the detector can be poor. If that is the case it will be visible in the extracted 2d spectra.

EMIR

The only observing mode included in the automatic data reduction for EMIR is Imaging. The pipeline is still in a beta version and therefore the reduced data are not delivered for all observing blocks.

EMIR Imaging

In the nearinfrared (NIR), it is not possible to combine or subtract sky frames that are widely separated in time, as the background varies significantly on timescales of only a few minutes and would no longer be comparable. In addition, in imaging mode the individual exposure times are typically short in order to avoid detector saturation.

Imaging frames with EMIR are reduced with the dedicated SERRADERO pipeline written in python. The first step is to identify images in dithering blocks with compatible sky backgrounds and combine them into subgroups. For each subgroup a so-called superflat is constructed. This superflat is generated from the science images themselves, removing stars and other objects during the combination process so that only the smooth background pattern is preserved. This method of flat-field generation is far more efficient in the NIR than the use of traditional twilight sky flats. The use of sky flats is only recommended for extremely crowded fields, such as the Galactic Centre or similar regions.

After applying the normalised flat, a similar procedure is followed to construct a sky image, which is then subtracted from the corresponding science images. All these steps are always performed on subgroups of FITS files for which the sky background remains coherent.

For the astrometric calibration, SERRADERO uses the software packages developed by the University of Cambridge and the CASU Tools group, which identify sources in public catalogues (2MASS, Gaia, etc.) within the field and correlate them with the objects detected in the FITS images, computing the required rotation and translation transformations to refine the astrometry in each header. Finally, all subgroups of images are combined into a single stacked image for each filter.

Due to the often very large amount of individual frames, the reduction of each individual frame or pattern of frames is not delivered, only a final stacked image of each filter observed.

The Table below shows the naming scheme for EMIR imaging reduced data.

EMIR Long Slit Spectroscopy

Long Slit spectroscopic observations with EMIR are reduced with the dedicated TAGASASTE (Tool for Analysis of GTC Acquired Spectra And Standards Taken with EMIR) pipeline written in python. Spectroscopic observations in the infrared are carried out using the nodding approach to be able to correct for the more variable sky background. For this reason, only one 2D reduced frame and one 1D extracted spectrum is produced for each setting. The 2D reduced frame is in ADU/s.

TAGASASTE automates the organization of the observations, the generation of calibration products, and the production of the final scientific data products. This includes instrumental corrections, the construction of sky and superflat frames directly from the science observations whenever the observing strategy allows it, sky background subtraction, exposure combination, and the extraction of the resulting two-dimensional and one-dimensional spectra. The wavelength calibration is performed using the OH emission lines naturally present in the night sky. These lines provide a wavelength reference acquired simultaneously with the scientific observations, allowing a reliable and consistent wavelength calibration to be achieved without requiring dedicated arc-lamp calibration exposures in most observing configurations.

The following plot shows an example of a reduction with the trace on the 2D image in the top panel, the extracted 1D spectrum in the middle panel and the OH sky lines used for the wavelength solution in the bottom panel.

The Table below shows the naming scheme for EMIR long slit spectrscopic reduced data.

Known Issues

• In the middle of the detector there is at times a ghost present, in particular for bright targets observed with the Ksp and H grisms. An example is seen in the top panel of the above plot.

Last modified: 23 July 2026