Very Large Telescope Paranal Science Operations VISIR data reduction cookbook

Transcription

1 EUROPEAN SOUTHERN OBSERVATORY Organisation Européene pour des Recherches Astronomiques dans l Hémisphère Austral Europäische Organisation für astronomische Forschung in der südlichen Hemisphäre ESO - European Southern Observatory Karl-Schwarzschild Str. 2, D Garching bei München Very Large Telescope Paranal Science Operations VISIR data reduction cookbook Doc. No. VLT-MAN-ESO-xxxxx-yyyy Issue 80.0, Date 01/09/2007 D. Nürnberger Prepared Date Signature A. Kaufer Approved Date Signature O. Hainaut Released Date Signature

2 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy ii This page was intentionally left blank

3 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy iii Change Record Issue/Rev. Date Section/Parag. affected Reason/Initiation/Documents/Remarks

4 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy iv This page was intentionally left blank

5 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy v Contents 1 Introduction Purpose Reference documents Abbreviations and acronyms Stylistic conventions Observing in the MIR from the ground The Earth s atmosphere MIR background Chopping and nodding VISIR Instrument Description Instrument Overview The spatial resolution Sensitivity Detectors VISIR Data Acquisition 13 5 VISIR Data Description General Data Layout General frames Imaging frames Spectroscopy frames VISIR Pipeline Description 17 7 VISIR Data Reduction Imaging Data Burst Mode Imaging Data Spectroscopy Data A Reference Frames 19

6 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 1 1 Introduction 1.1 Purpose This document is intended for astronomers who want to reduce VISIR data. It describes the various data formats delivered by VISIR, observational scenarios and reduction procedures. It concentrates on the methodology rather than individual routines that are available in either IDL, IRAF or MIDAS. However, some of these routines are mentioned explicitely if the author have found them particularly useful. This document also describes the algorithms implemented in the eclipse data reduction package. The eclipse package is an integral part of the VISIR pipeline which produces calibration products, quality control information and reduced data. The VISIR pipeline produces reduced data; however, by no means it can replace more general reduction packages such as IDL, IRAF or MIDAS. The pipeline does not replace interactive analysis and can not make educated choices. Thus, the data products of the VISIR pipeline should be helpful for a quick assessment of the quality of the data (a quick-look tool if you like) and considered as a first step in the reduction of the data. Throughout this document we will list the eclipse routines which are used to reduce the VISIR data and we will give a short description of how they can be used. We will also list the shortcomings these routines have, so that users can decide if they need to reduce their data more carefully. For completeness, we have also included a description of the eclipse routines whose primary aim is to provide quality control information. These routines are probably of little interest to astronomers. This document does not describe the VISIR instrument, its mode of operations, the offered templates, the different ways to acquire a target, or the various issues related to Phase II Proposal Preparation. The reader is assumed to have read the VISIR User Manual beforehand and to have a basic knowledge in the reduction of infrared imaging data and spectroscopic data. This document is a living document, and as such follows the evolution of the recipes, the implementation of new algorithms, any enhancements, new supported observing modes, etc. Any information given in this document has to be understood to be valid at the date of writing. 1.2 Reference documents 1.3 Abbreviations and acronyms The following abbreviations and acronyms are used in this document:

7 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 2 SciOp ESO Dec eclipse ESO-MIDAS FITS IRAF PAF RA UT VLT Science Operations European Southern Observatory Declination ESO C Library Image Processing Software Environment ESO s Munich Image Data Analysis System Flexible Image Transport System Image Reduction and Analysis Facility PArameter File Right Ascension Unit Telecope Very Large Telescope 1.4 Stylistic conventions The following styles are used: bold in the text, for commands, etc., as they have to be typed. italic for parts that have to be substituted with real content. box for buttons to click on. teletype for examples and filenames with path in the text. Bold and italic are also used to highlight words.

8 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 3 2 Observing in the MIR from the ground 2.1 The Earth s atmosphere Our atmosphere absorbs the majority of the MIR radiation from astronomical sources. The main absorbing molecules are H 2 O, CH 4, CO 2, CO, O 2, O 3. However, the atmosphere is quite transparent in two atmospheric windows: the N and Q bands. They are centered around 10 µm and 20 µm, respectively. The transmission in the N band is fairly good at a dry site and becomes particular transparent in the wavelength range µm. However, the transmission of the Q band is rapidly decreasing with wavelength and can be viewed as the superposition of many sub-bands having a typical spectral coverage of λ = 1 µm at an average transmission of 60%. Observations in this band require low water vapor content in the atmosphere. The atmospheric transmission in the N and Q bands is displayed on Fig. 1. Figure 1: MIR atmospheric transmission at Paranal computed with HITRAN for an altitude of 2600 m and 1.5 mm of precipitable water vapor at zenith. The US standard model atmosphere is used. 2.2 MIR background The atmosphere does not only absorb MIR photons coming from astrophysical targets, but also emits a strong background with the spectral shape of a blackbody at about 253 K (Kirchhoff s law). The telescope gives an additional MIR background. The VLT telescopes emits at 283 K with a preliminary emissivity estimate of < 15% in N. The VISIR instrument is cooled to avoid internal background contamination. The detectors are at 5 6 K and the interior of the cryostat at 33 K. The background radiation at 10 µm is typically m N = 5 mag/arcsec 2 (3700 Jy/arcsec 2 ) and at 20 µm m Q = 7.3 mag/arcsec 2 (8300 Jy/arcsec 2 ). Consequently, the number of photons reaching the detector is huge, often more than 10 8 photons/s.

9 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 4 Therefore, the exposure time of an individual integration the Detector Integration Time (DIT) is short, of the order of a few tens of milli-seconds in imaging mode. 2.3 Chopping and nodding The basic idea to suppress the MIR background is to perform differential observations, using the chopping/nodding technique. In the chopping technique two observations are performed. One set of exposures on-source include the background and the astronomical source. A second set of off-source exposures measures the pure background. The on-source and off-source observations have to be alternated at a rate faster than the rate of the background fluctuations. In practice, it is achieved by moving the secondary mirror of the telescope. For VISIR at Paranal, a chopping frequency of 0.25 Hz has been found to be adequate for N-band imaging observations, while 0.5 Hz are adopted for Q-band imaging. Spectroscopic observations are performed with lower chopper frequencies, at 0.1 Hz or less. The chopping technique cancels most of the background. However, the optical path is not exactly the same in both chopper positions. Therefore a residual background remains. It is varying at a time-scale which is long compared to that of the sky. This residual is suppressed by nodding, where the telescope itself is moved off-source and the same chopping observations as in the on-source position is repeated. An illustration of the chopping and nodding technique is shown on Fig. 2. Depending on the choice of chopping and nodding amplitudes and directions, up to 4 images of the source can be seen on the frame and used for scientific analysis. Of course, the free field-of-view on the chop/nod images can be severely reduced depending on the particular chopping and nodding parameters chosen. 3 VISIR Instrument Description This section provides a brief description of the VISIR instrument. A more complete documentation can be found in the VISIR User Manual, downloadable from VISIR has been developed under ESO contract by CEA/DAPNIA/SAP and NFRA/ASTRON. The instrument has been made available to the community and started operations in Paranal in April Instrument Overview The VISIR instrument is located at the Cassegrain focus of UT3 of the VLT at Paranal. It provides diffraction-limited imaging in the two mid infrared (MIR) atmospheric windows: the N band between 8 to 13µm and the Q band between 16.5 and 24.5µm, respectively. It also features a spectrometer offering long-slit spectroscopy at low resolution (down to 150) in the N band, medium resolution in the N and Q band and high resolution (up to 30000) for a limited set of wavelengths, as well as cross-dispersed high resolution spectroscopy over most of the N and Q band. Because of the very high background from the ambient atmosphere and telescope, the sensitiv-

10 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 5 Figure 2: Illustration of the chopping and nodding technique on observations of the blue compact galaxy He2-10. The intrinsically faint galaxy only appears after chopping and nodding (courtesy VISIR commissioning team, June 2004).

11 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 6 ity of ground-based MIR instruments can t compete with that of space-borne ones. However, ground based instruments mounted on large telescopes offer superior spatial resolution. For example VISIR at the VLT provides diffraction limited images at 0.3 (FWHM) in the N band. This is an order of magnitude better than what can be reached by the Spitzer Space Telescope (SST). The VISIR imager and spectrometer are each equipped with a DRS (former Boing) 256x256 BIB detector. The quantum efficiency of the detectors reaches close to 70% in the N-band and has a sharp absorption feature at 8.8µm. See also??. SEE CHAPTER 7.3 ON PAGE 28, IN PARTICULAR FIGURE For a more detailed description of the VISIR instrument, please see the VISIR User Manual. 3.2 The spatial resolution The spatial resolution of an instrument is ultimately limited either by the diffraction of the telescope or the atmospheric seeing. The diffraction limit as measured by the diameter of the first Airy ring increases with wavelength as 1.22 λ/d, where λ is the observing wavelength and D the diameter of the telescope mirror (see solid line in Fig. 3). The wavelength dependence of the seeing can be derived by studying the spatial coherence radius of the atmosphere in the telescope beam and is to first order approximated by the Roddier formula, where the seeing is λ 0.2 (see dot-dashed lines in Fig. 3). However, initial results from VISIR data indicate that this formula overestimates the measured MIR seeing at Paranal by 20 50%, as the size of a UT mirror is comparable to the turbulence outer scale. As a result, VISIR data are already diffraction limited for optical seeing below 0.6. The results of measures obtained in 2005 are shown in Fig. 4. Figure 3: VLT diffraction limit (full line) versus seeing. The Spitzer Space Telescope diffraction limits (dashed) are shown for comparison. The Roddier dependence is shown for two optical seeings (dashed-dot).

12 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy FWHM (VIS) Figure 4: Measurements of the VISIR image quality versus optical seeing obtained during The dashed lines indicates the prediction of Roddier s formula.

13 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy Sensitivity Measurements of VISIR sensitivities are based on observations of mid-ir calibration standard stars (Cohen et al. 1999, AJ 117, 1864). In imaging mode, the stars are recorded in the small field (0.075 ) and intermediate field (0.127 ) by perpendicular chopping and nodding patterns with amplitudes of 10. Calibrators are frequently observed during the night (see section??). Flux and noise levels are extracted by multi-aperture photometry using the curve-of-growth method: the aperture used for all 4 beams in a given frame is the one for which the flux to noise ratio is the largest. By combining all 4 beams, the sensitivity in a given setup (filter, field-of-view) is defined as the limiting flux of a point-source detected with a S/N of 10 in one hour of on-source integration. The growing calibration database allows a statistical analysis of the sensitivity with respect to instrumental and atmospherical conditions. The values for each filter given in Table?? refer to the median of more than 600 different observations during September and December A graphical compilation is presented in Fig. 5 for the N-band and Q-band imaging filters. Some of the best measurements approach theoretical expectations, i.e., they are close to background limited performance (BLIP). Sensitivity estimates for the VISIR spectroscopy observing modes are obtained in a similar way. However, in this case, chopping and nodding are executed in parallel. Consequently, only 3 beams are obtained, with the central one containing twice as much flux as the two other ones. Tables?? to?? list typical sensitivities measured in low, medium and high resolution modes away from strong sky emission lines for the wavelength ranges offered in P76. Figures?? to?? in the Appendix (see section??) shows the dependence of sensitivity on wavelength. The median sensitivities are the reference for classification of VISIR service mode observations, and the basis to assess the feasibility of an observing programme. In particular, classification of service mode OBs will be based on sensitivity measurements made at zenith. Calibrations will be provided following the guidelines given in section??. For up to date information, please consult The use the VISIR exposure time calculator (ETC, located at is recommended to estimate the on-source integration time. 3.4 Detectors The VISIR imager and spectrometer are each equipped with a DRS, former Boeing, BIB detector. The quantum efficiency of the detectors is greater than 50% and reaches 65% or more at 12 µm (Fig. 6). The detector noise has to be compared with the photon noise of the background. As shown in Fig. 7, the measured noise in an observation consists of read-out noise and fixed pattern noise, which are both independent of the detector integration time (DIT). At the operating temperature of the detector ( 6 K), the dark current, which is the signal obtained when the detector receives no photons, is negligible compared to the background generated by the photons emitted by the telescope and the atmosphere. The dark current is removed by the observation technique (chopping or nodding). It is at least 6 times lower than the photon noise for the spectrometer and negligible for the imager. The detectors have a switchable pixel ( well ) capacity. The large capacity is used for broadband imaging and the small capacity for narrow band imaging and spectroscopy. Detector

14 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy median small field median intermed. field sensitivity [mjy 10σ/h] 10 1 PAH1 ARIII SIV_1 SIV SIV_2 PAH2 PAH2_ wavelength [µm] SIC NEII_1 NEII NEII_2 median small field sensitivity [mjy 10σ/h] median intermed. field Q1 Q2 Q wavelength [µm] Figure 5: Sensitivities for the VISIR imager for the N-band (top) and Q-band (bottom). Small and intermediate field observations are displaced for clarity. Background noise limits are indicated for the individual filter bandpasses.

15 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 10 Figure 6: Detector quantum efficiency at 12 K provided by DRS (solid line). The same curve (dashed) but scaled by 0.72 reflects a lower limit of the quantum efficiency. The scaling was derived from laboratory measurements. Note the sharp absorption feature at 8.8 µm that will appear in raw spectroscopic data. Figure 7: Noise as a function of the incoming flux in the large (left) and small (right) capacity mode. Superimposed is the theoretical photon noise. BLIP performances are approached for higher fluxes and larger DIT, respectively. saturation due to the enormous MIR background is avoided by a storage capacity of e in small and e in large capacity modes, respectively. For background limited noise performance (BLIP), the optimal operational range of the detector is half of the dynamic range for the large capacity, and between 1/2 and 1/5 for the small capacity. The detector is linear over 2/3 of its dynamic range (Fig. 8) and its working point is set in the middle of the dynamic range. During commissioning it was found that, for about half of the array, the gain does not differ by more than 2% peak-to-peak. By comparison with other limitations, flat-field corrections, which are difficult to implement in the MIR, are not considered important. The detector integration time (DIT) is a few milli-seconds in broad-band imaging and may increase to 2 s in high resolution spectroscopy. The DIT is determined by the instrument software according to the filter and pfov. It is not a parameter to be chosen by the observer. The DRS detectors contain a fair fraction of bad pixels (<2%; see Fig.9). The imager detector suffers from striping and appearances of ghosts. The relatively wide rectangular area in the lower right corner (south-west corner for PA = 0 deg) of the imager detector or some other

16 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 11 Figure 8: Linearity curve of the detector in the large (left) and small (right) capacity modes. The break in the response at 2/3 at e of the large and at e of the small capacity are indicated by full lines. The top lines indicate the well capacities. rectangular areas are masked out to avoid such disturbances (Fig. 10). For bright objects the DRS detector shows memory effects. Stabilization is ensured by introducing dead times where necessary. It is advised to observe only sources fainter than 500 Jy in N and 2500 Jy in Q. Figure 9: Bad pixel maps of the imager (left) and spectrometer (right) detectors. The large grey rectangular areas correspond to pixels masked electronically in order to decrease detector striping. These artifacts are less important in spectroscopy due to the lower light levels but clearly visible on objects brighter than 2% of the background. However, a TEL.CHOP.THROW between 9 to 13 shoud be avoided, in particular for objects bright enough to be seen in individual DITs, as one of the beams will hit some particularly hot pixels in the lower-left of the spectrometer detector (see Fig.11).

17 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 12 Figure 10: The DRS detector shows stripes and repeating ghosts for very bright sources (left). The ghosts are distributed every 16 columns. For other sources striping is not apparent (right). Figure 11: Sequence of chop/nod, reduced spectra obtained in the Medium Resolution mode with a central wavelength = 8.8 µm. The TEL.CHOP.THROW = 8, 11, 13 and from left to right. Note the presence of significant striping when the left beam hits some hot pixels at the lower left of the detector. For the location of the object along the slit (pixel X = 123 at row Y = 128), this occured for TEL.CHOP.THROW between 10 and 13 00, approximatively. The horizontal lines at the middle of the images are caused by the lack of detector response at 8.8 µm.

18 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 13 4 VISIR Data Acquisition Both VISIR detectors are controlled by the ESO standard IRACE acquisition system. In imaging mode, the read-out rate of the detector is high. Up to 200 frames per second are read for a minimum detector integration time of DIT = 5 ms. Such a frame rate is too high to store all exposures. One VISIR image is of size ; each pixel is coded with 4 bytes (long integer). Thus one read-out has a size of 262 kb. During each chopping cycle the elementary exposures are added in real time and only the result is stored on disk. At a chopping frequency of f chop = 0.25 Hz every T chop = 4 s one VISIR image is stored as a plane in a data cube of a FITS file. The number of chopping cycles, within one nodding position, is defined by the time spent integrating in that nodding position, T nod. This nodding period is typically T nod = 90 s for science observations. The chopper frequency, DIT and also T nod are predefined by the system. The number of saved A-B frames in one FITS file is: N cycl chop = T nod /T chop (1) The number of nodding cycles is computed from the total integration time as given by the observer. The total number of stacked images for each secondary position, respectively chopper half cycle, is NDIT. This parameter is computed according to: NDIT = (2 DIT f chop ) 1 NDITSKIP (2) and is given by the system. It depends on DIT, chopping frequency and NDITSKIP: some read-outs at the beginning of each chopper half cycle are rejected during stabilization of the secondary. Typical stabilization times of the secondary are 25 ms. The number of rejected exposures is given by NDITSKIP. Similar, during stabilization after each telescope movement, respectively nodding position, a number NCYSKIP of chopping cycles is ignored. The timing organization of data is shown in Fig. 12. The total on source integration time is: The total rejected time is: t source = 4 N cycl nod N cycl chop NDIT DIT (3) t skip = 4 N cycl chop DIT (NDITSKIP N cycl nod + NDIT NCYSKIP) (4) and the total observing time is: Typical duty cycles (t source /t tot ) are about 70%. t tot = t source + t skip (5) 5 VISIR Data Description VISIR data uses the FITS format and can be separated into raw frames and product frames. Raw frames are the unprocessed output of the VISIR instrument observations, while product frames are the result of the VISIR pipeline processing. In addition, the VISIR pipeline uses a set of calibration (FITS-) files (standard star catalogs, detector characteristics, etc.).

19 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 14 NCYSKIP N_cycl_chop NCYSKIP N_cycl_chop An Bn Bn An T_nod NDITSKIP NDIT NDITSKIP DIT Ac Bc Ac Bc T_chop Figure 12: Data timing in VISIR. Ac and Bc refer to the two chopper positions, An and Bn refer to the two nodding (telescope) positions. Note the AnBnBnAn cycle sequence for the nodding to save observing time. Any raw or product frame can be classified on the basis of a set of keywords read from its header. Data classification is typically carried out by the DO or by Gasgano [6] which both apply the same set of classification rules. The association of a raw frame with calibration data (e.g., of a science frame with a standard star catalogue) can be obtained by matching the values of a different set of header keywords. Each kind of raw frame is typically associated to a single VISIR pipeline recipe, i.e., the recipe assigned to the reduction of that specific frame type. In the pipeline environment this recipe would be launched automatically. In the following all raw and product VISIR data frames, that can be reduced by the VISIR pipeline (version 1.3.7), are listed, together with the keywords used for their classification and correct association. The indicated DO category is a label assigned to any data type after it has been classified, which is then used to identify the frames listed in the Set of Frames (see section??). SEE SECTION on page 17. Raw frames can be classified as imaging frames or spectroscopy frames. Their intended use is implicitly defined by the assigned recipe. 5.1 General Data Layout A raw VISIR file is an extension-less FITS-file. The data unit is a cube with NAXIS3= 2n + 1 planes 1, where n is the number of chopping cycles, which is specified in the FITS-card with the key HIERARCH ESO DET CHOP NCYCLES. For each chopping cycle two so called Half-Cycle exposures are made, the A-image from the on-source position of the chopper, and the B-image from the off-source position of the chopper. Each Half-Cycle image is normalized by IRACE to an exposure time of one DIT; in other words, each Half-Cycle image is the average over the 1 Before another data layout was used. The description of this now obsolete format is limited to the statement that it is also supported by the VISIR pipeline

20 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 15 NDIT individual exposures. For each chopping cycle two planes are stored in the cube. The first two planes correspond to the first chopping cycle and contain: The Half-Cycle A-image, A 1. The pixel values in each Half-Cycle image are offset by 32768, i.e., has to be added to each pixel in order to obtain the physical pixel value. The difference between the two Half-Cycle images, A 1 B 1. Similarly, the (2 i 1)th and (2 i)th planes correspond to the ith chopping cycle and contain The Half-Cycle A-image, A i, stored with an offset identical to A 1. The average of the current and all previous Half-Cycle difference images, (A 1 B 1 + A 2 B A i B i )/i. The last plane of the cube contains the average of all Half-Cycle difference images, i.e., it is identical to the (2 n)th plane. 5.2 General frames These are data that can be obtained using any of the two instrument modes (imaging, spectroscopy), as is the case for flat field exposures. The keyword ESO INS MODE is set accordingly to IMG for imaging frames and to SPC for spectroscopy frames, to indicate the intended use for the data. Flat field: Processed by: visir img ff Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB FLAT IMAGE,DIRECT IM CAL FLAT CALIB FLAT SPECTRUM,DIRECT SPEC CAL FLAT TECHNICAL FLAT IMAGE,DIRECT IM TECH FLAT TECHNICAL FLAT SPECTRUM,DIRECT SPEC TECH FLAT See [4] for a definition of the values of DPR.CATG, DPR.TYPE and DPR.TECH. 5.3 Imaging frames Science Observation: Processed by: visir img combine Association keywords: INSTRUME = VISIR Classification:

21 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 16 DPR.CATG DPR.TYPE DPR.TECH DO Category SCIENCE OBJECT IMAGE,CHOPNOD,JITTER IM OBS CHO NOD JIT SCIENCE OBJECT IMAGE,CHOPPING,JITTER IM OBS CHO JIT SCIENCE OBJECT IMAGE,NODDING,JITTER IM OBS NOD JIT SCIENCE OBJECT IMAGE,DIRECT,JITTER IM OBS DIR JIT Standard Star: Processed by: visir img phot Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB STD IMAGE,CHOPNOD IM CAL PHOT See [4] for a definition of the values of DPR.CATG, DPR.TYPE and DPR.TECH. 5.4 Spectroscopy frames These frames are generated with the VISIR spectrometer. Long Slit Wavelength Calibration: Processed by: visir spc wcal Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB WAVE SPECTRUM,DIRECT SPEC CAL LMR WCAL Long Slit Science Observation: Processed by: visir spc obs Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category SCIENCE OBJECT SPECTRUM,CHOPNOD SPEC OBS LMR Long Slit Standard Star: Processed by: visir spc phot Association keywords: INSTRUME = VISIR Classification:

22 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 17 DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB STD SPECTRUM,CHOPNOD SPEC CAL PHOT Echelle Wavelength Calibration: Processed by: visir spc wcal ech Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB WAVE ECHELLE SPEC CAL HRG WCAL Echelle Science Observation: Processed by: visir spc obs ech Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category SCIENCE OBJECT ECHELLE SPEC OBS HRG Echelle Standard Star: Processed by: visir spc phot ech Association keywords: INSTRUME = VISIR Classification: DPR.CATG DPR.TYPE DPR.TECH DO Category CALIB STD ECHELLE SPEC CAL PHOT HRG See [4] for a definition of the values of DPR.CATG, DPR.TYPE and DPR.TECH. 6 VISIR Pipeline Description A more detailed description of the VISIR pipeline and the recipes used by the VISIR pipeline is given in the VISIR Pipeline User Manual, which can be downloaded from the VISIR Instrument Web Pages. 7 VISIR Data Reduction 7.1 Imaging Data In principle, the reduction of mid IR data is quite similar to data taken at near IR wavelength. However, the basic reduction of mid IR data always needs to include chopping and nodding corrections. Given the large background fluctuations at mid IR wavelengths, there is usually no flat fielding done during the reduction of mid IR data.

23 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy Burst Mode Imaging Data The reduction of imaging data taken in burst mode is described in great detail by Doucet et al. (2006, proceeding of the VIRA conference). 7.3 Spectroscopy Data (Thanks a lot to Eric Pantin for the provided input.) First, basic data reduction (i.e., chopping and nodding correction) is performed as for imaging data. In parallel, a reference frame of the infrared background (neither chopping nor nodding correction) is produced from the same dataset. Second, the frames are corrected for image distorsion (the spectrometer optics produce some optical distorsion) to rectify the spectra and to make sure that they are aligned with the detector grid. This step must be applied on both the chopping-nodding corrected science frames and the background frames. Third, in case the cross-dispersed high-resolution mode has been used, the relevant part of the detector (i.e., the one containing the relevant optical order) is extracted as a new (smaller) frame. Fourth, the spectrum is extracted. This extraction is based on the optimum extraction by Horne (1986, PASP 98, 609) but extended in order to consider also the negative beams present in most of the spectroscopic data (i.e., if the chopping throw is smaller than half of the available field-of-view. A weight map is produced using a source profile. This source profile is obtained from the collapse of the spectrum in spectral dispersion direction. The profile is then normalized and expanded (i.e., copied in spectral dispersion direction) to produce a weight map. After multiplying the science frame by the weight map, the final extracted spectrum is obtained by summing in spatial direction. The extracted spectrum has exactly the same number of data points as the detector size in dispersion direction (i.e., 256 pixels with the current DRS detector). Fifth, the spectrum is wavelength calibrated. The background frame is simply averaged over the spatial axis and the resulting profile is compared to a synthetic (modeled) spectrum of the atmosphere in the observed wavelength range (depending on central wavelength and grism setting). A cross correlation between the observed spectrum and the synthetic spectrum then allows to derive any wavelength shift. The wavelength table is then corrected accordingly. Sixth, the spectrum is flux calibrated. Please note that this step is not performed by the ESO provided VISIR pipeline. For this purpose, usually a spectrophotometric standard star (e.g., selected from the Cohen list of infrared standard stars) is observed using the same instrument setting as for the science target. In the optimal case (but not always possible), this observation is done immediately before and/or after the science target. The same steps 1-5 (see above) are applied to the spectrophotometric standard star data as for the science data. Then, if necessary, the wavelength ranges of the science target and the standard star are merged, i.e., the common wavelength range is determined and the spectra are rebinned onto the same wavelength range. Finally, the science spectrum is divided by the standard star spectrum and the result is multiplied by the theoretical standard star spectrum (given, e.g., in Jy or in W/m/µm) to obtain the final calibrated spectrum.

24 VISIR data reduction cookbook VLT-MAN-ESO-xxxxx-yyyy 19 A Reference Frames VISIR reference frames are accessible via the VISIR Instrument Web Pages (to be verified!). ooo