© ESO, 2015

1. Introduction

The Calar Alto Legacy Integral Field Area (CALIFA) survey (Sánchez et al. 2012a, hereafter S12) is an ongoing large project of the Centro Astronómico Hispano-Alemán at the Calar Alto observatory (Almería, Spain) to obtain spatially resolved spectra for 600 galaxies in the local Universe by means of integral field spectroscopy (IFS). The CALIFA observations started in June 2010 with the Potsdam Multi Aperture Spectrograph (PMAS, Roth et al. 2005), mounted on the 3.5 m telescope, utilizing the large hexagonal field-of-view (FoV) offered by the PPak fiber bundle (Verheijen et al. 2004; Kelz et al. 2006). Each galaxy is observed using two different setups: one at intermediate spectral resolution (V1200, R ~ 1650) and the other at low resolution (V500, R ~ 850). A diameter-selected sample of 939 galaxies was drawn from the 7th data release of the Sloan Digital Sky Survey (SDSS DR7, Abazajian et al. 2009), which is described in Walcher et al. (2014, hereafter W14). From this mother sample, the 600 target galaxies are randomly selected.

Combining the techniques of imaging and spectroscopy through optical IFS provides a more comprehensive view of individual galaxy properties than any traditional survey. CALIFA-like observations were collected during the feasibility studies (Mármol-Queraltó et al. 2011; Viironen et al. 2012) and the PPak IFS Nearby Galaxy Survey (PINGS, Rosales-Ortega et al. 2010), a predecessor of this survey. First results based on those data sets already explored their information content (e.g., Sánchez et al. 2011, 2012b; Rosales-Ortega et al. 2011, 2012; Alonso-Herrero et al. 2012). The CALIFA survey can therefore be expected to make a substantial contribution to our understanding of galaxy evolution in various aspects, including (i) the relative importance and consequences of merging and secular processes; (ii) the evolution of galaxies across the color–magnitude diagram; (iii) the effects of the environment on galaxies; (iv) the AGN-host galaxy connection; (v) the internal dynamical processes in galaxies; and (vi) the global and spatially resolved star formation history and chemical enrichment of various galaxy types.

Compared with previous IFS surveys, e.g., Atlas3D (Cappellari et al. 2011) or the Disk Mass Survey (DMS) (Bershady et al. 2010), CALIFA covers a much wider range of morphological types over a large range of masses, sampling the entire color–magnitude diagram for M_r> −19 mag. While the recently started SAMI (Croom et al. 2012; Bryant et al. 2015) and MaNGA (Law & MaNGA Team 2014) surveys have a broad scope similar to CALIFA and aim to build much larger samples, CALIFA still has an advantage in terms of spatial coverage and sampling. For 50% of the galaxies, CALIFA provides data out to 3.5 r_e, and for 80% out to 2.5 r_e. At the same time, the spatial resolution of ~1 kpc is typically better than in either SAMI or MaNGA, revealing several of the most relevant structures in galaxies (spiral arms, bars, bulges, giant H ii regions, etc.). The spectral resolution of CALIFA is lower than these two surveys in the red wavelength range, but is comparable for the blue wavelength range.

So far, a number of science goals have been addressed using the data from the CALIFA survey: (i) new techniques have been developed to understand the spatially resolved star formation histories (SFH) of galaxies (Cid Fernandes et al. 2013, 2014). We found solid evidence that mass-assembly in the typical galaxies happens from the inside-out (Pérez et al. 2013). The SFH and metal enrichment of bulges and early-type galaxies are fundamentally related to the total stellar mass, while for disk galaxies it is more closely related to the local stellar mass density (González Delgado et al. 2014b,a); (ii) we developed new tools to detect and extract the spectra of H ii regions (Sánchez et al. 2012b), building the largest catalog currently available (~6000 H ii regions and aggregations). This catalog has been used to define a new oxygen abundance calibrator, anchored to electron temperature measurements (Marino et al. 2013). From these, we explored the dependence of the mass-metallicity relation with star formation rate (Sánchez et al. 2013), and the local mass-metallicity relation (Rosales-Ortega et al. 2012). We found that all galaxies in our sample present a common gas-phase oxygen abundance radial gradient with a similar slope, when normalized to the effective radius (Sánchez et al. 2014). This agrees with an inside-out scenario for galaxy growth. This characteristic slope is independent of the properties of the galaxies, and, in particular, of the presence or absence of a bar, contrary to previous results. More recently, this result has been confirmed by the analysis of the stellar abundance gradient in the same sample (Sánchez-Blázquez et al. 2014); (iii) we explored the origin of the low intensity, LINER-like, ionized gas in galaxies. These regions are clearly not related to star formation activity, or to AGN activity. They are probably most closely related to post-AGB ionization in many cases (Kehrig et al. 2012; Singh et al. 2013; Papaderos et al. 2013); (iv) we explored the aperture and resolution effects on the data. The CALIFA survey provides a unique tool to understand the aperture and resolution effects in larger single-fiber (e.g., SDSS) and IFS surveys (e.g., MaNGA, SAMI). We explored the effects of the dilution of the signal in different gas and stellar population properties (Mast et al. 2014), and proposed a new empirical aperture correction for the SDSS data (Iglesias-Páramo et al. 2013); (v) we analyzed the local properties of the ionized gas and stellar population of galaxies where supernovae (SNe) have exploded. Core collapse SNe are found closer to younger stellar populations, while SNe Ia show no correlation to stellar age (Galbany et al. 2014); (vi) CALIFA is the first IFS survey that allows gas and stellar kinematic studies for all morphologies with sufficient spectroscopic resolution to study (a) the kinematics of the ionized gas (García-Lorenzo et al. 2015); (b) the effects of bars in the kinematics of galaxies (Barrera-Ballesteros et al. 2014); (c) the effects of the interaction stage on the kinematic signatures (Barrera-Ballesteros et al., in prep.); (d) the bar pattern speeds in late-type galaxies (Aguerri et al. 2015); (e) the measurements of the angular momentum of galaxies to previously unexplored ranges of morphology and ellipticity (Falcón-Barroso et al., in prep.); and (vii) we explored the effects of a first stage merger on the gas and stellar kinematics, star formation activity and stellar populations of the Mice merging galaxies (Wild et al. 2014).

In this article, we introduce the second data release (DR2) of CALIFA, which grants public access to high-quality data for a set of 200 galaxies (400 datacubes). All the cubes in the data release have been reduced with the latest pipeline, which includes improved spectrophotometric calibration, spatial registration, and spatial resolution. This DR supersedes and increases he amount of data delivered in DR1 (Husemann et al. 2013, hereafter H13) by a factor of two.

The DR1 opened CALIFA to the community, and allowed for the exploration of several different scientific avenues not addressed by the collaboration (e.g., Holwerda & Keel 2013; De Geyter et al. 2014; Martínez-García et al. 2014; Davies et al. 2014). The properties of the galaxies in the DR2 sample are summarized in Sect. 2. We describe the observing strategy and setup (Sect. 3), processing (Sect. 4), structure (Sect. 5), and data (Sect. 6), which comprise essential information for any scientific analysis of the distributed CALIFA data. Several interfaces to access the CALIFA DR2 data are explained in Sect. 7.

2. The CALIFA DR2 sample

The CALIFA “mother sample” (MS) consists of 939 galaxies drawn from SDSS DR7. The main criteria for the target selection are: angular isophotal diameter (45″ <isoA_r< 79.2″) of the galaxies¹; redshift range 0.005 <z< 0.03; cut in Galactic latitude to exclude the Galactic plane (| b | > 20°); flux limit of petroMag_r< 20; and declination limit to δ> 7°. Redshift limits were imposed so that the sample would not be dominated by dwarf galaxies and to keep relevant spectral features observable within a fixed instrumental spectral setup. Redshift information was taken from SIMBAD for all galaxies where SDSS DR7 spectra were unavailable. The cut in declination was chosen to reduce problems due to differential atmospheric refraction (DAR) and PMAS flexure issues, but was not applied to the SDSS Southern area because of the sparsity of objects in this region. A comprehensive characterization of the CALIFA MS and a detailed evaluation of the selection effects implied by the chosen criteria is provided in W14. From the CALIFA MS, 600 galaxies are randomly selected for observation, based purely on visibility.

The 200 DR2 galaxies, which include the first 100 galaxies of DR1, were observed in both spectral setups between the start of observations in June 2010 and December 2013. We list these galaxies in Table 1 together with their primary characteristics. The distribution of galaxies in the sky follows the underlying SDSS footprint (Fig. 1). The number of galaxies in DR2 is not homogeneous as a function of right ascension, α(J2000), and has three clear peaks at around α~ 15°, 255°, and 345°. All three peaks are located in the same observing semester, in the period from April to October. As noted in H13, there was a downtime of the 3.5 m telescope from August 2010 until April 2011 due to operational reasons at the observatory, which delayed the survey roughly by eight months. In addition to this, because of scheduling matters, a large part of the granted time was allocated in summer seasons. Regardless of the observing time issue, the distribution of physical properties for DR2 is nearly random, as expected, and covers galaxies with a wide range of properties as discussed below.

Figure 2 shows the distribution of galaxies in the color–magnitude diagram. The DR2 sample covers nearly the full range of the CALIFA MS. On average, the DR2 targets comprise ~37% of each color–magnitude bin of the total 600 objects in the full CALIFA sample. The deficit of low-luminosity galaxies with intermediate colors, noted in DR1, has improved. Fluctuations can be explained by the effect of low-number statistics, especially within those color–magnitude bins in which the MS contains fewer galaxies. This point is highlighted in Fig. 2 and emphasizes the need to eventually observe the full CALIFA sample to obtain a sufficient number of galaxies in each bin for a meaningful multi dimensional statistical analysis.

Figure 3 compares the redshift distribution of the CALIFA galaxies in the DR2 and DR1, as a percentage of the CALIFA MS. Except for a few particular bins, the redshift distribution is homogeneous with respect to the MS.

One important test to be made is whether the number density of galaxies estimated from the DR2 sample is not biased with respect to the MS. Figure 4 shows the r-band luminosity function (LF) of the DR2 sample as compared to the MS and the reference SDSS sample of Blanton et al. (2005). We refer to W14 for all technical details on how the LFs are obtained and for the explanation of the turnover of the LF at M_r ≈ −18.6. The DR2 sample already reproduces the CALIFA MS LF very closely for most of the magnitude bins.

An important characteristic of the CALIFA MS is that it contains galaxies of all morphological types. Galaxy morphologies were inferred by combining the independent visual classifications of several collaboration members as described in W14. Figure 5 shows a histogram of bar strength as well as the percentage of DR2 galaxies with respect to the CALIFA MS distribution for different morphological types grouped into elliptical, lenticular, and spiral galaxies (and subtypes). A more detailed classification of ellipticals (from 0 to 7) is available, but we do not distinguish between them here because of the low number of galaxies per elliptical subtype within DR2. From 200 galaxies in DR2, 18 have been classified as ongoing mergers² (of any type). As clearly seen in Fig. 5, the percentage of DR2 galaxies with respect to the CALIFA MS is almost constant for all types, implying that the DR2 coverage seems to be consistent with a random selection. Axis ratios (b/a) were measured from the SDSS r-band image using growth curve analysis, by calculating light moments after proper sky subtraction and masking of foreground stars (see W14 for details). The axis ratios can be used as a proxy of the inclination of spiral galaxies. Figure 6 shows that the DR2 sample covers the same range of axis ratios as the MS. A Kolmogorov-Smirnov test gives a <5% confidence that the DR2 sample is drawn from a different distribution than the underlying MS.

In Fig. 7, we present the distribution of stellar masses for the DR2 galaxies. Galaxy stellar masses are from González Delgado et al. (2014a), and they have been estimated following the process described in Pérez et al. (2013), Cid Fernandes et al. (2013, 2014), and González Delgado et al. (2014b). These masses account for spatial variations in both M/L ratio and stellar extinction. In short, we use the starlight code (Cid Fernandes et al. 2005) to fit each spectrum extracted from the datacube with a combination of SSP models from the Granada (González Delgado et al. 2005) and MILES (Vazdekis et al. 2010) libraries, that cover the full metallicity range of the MILES models (log Z/Z_⊙ from −2.3 to +0.22), and ages from 0.001 to 14 Gyr. We assume a Salpeter IMF. The DR2 galaxies cover intermediate to high-mass galaxies, including at least ten galaxies per 0.25 dex bin between 10¹⁰ and 10^11.75M_⊙ and a median value close to 10¹¹M_⊙. The asymmetric distribution is expected from the distribution in absolute magnitudes (see Fig. 2) and is inherited from the CALIFA MS because of its selection criteria (see W14 for details).

A more general “panoramic view” of the DR2 sample characteristics is presented in Fig. 8. Several of the main properties observable in 2D are highlighted for 169 randomly-selected galaxies, shown individually in hexagons that together form the shape of a CALIFA-like FoV. The galaxies have been ordered by r-band absolute magnitude, from top right (brightest absolute magnitude) to bottom left (faintest absolute magnitude). The highlighted properties derive from several different analysis pipelines developed within the collaboration. Stellar properties, like ages and mass surface density, were measured with the starlight code (see references in the preceding paragraph describing the distribution of stellar masses in the sample) while gas properties and emission lines were measured using FIT3D (Sánchez et al. 2007). This plot is only intended to demonstrate the diversity of the DR2 sample.

3. Observing strategy and setup overview

For the sake of completeness, we provide here a brief summary of the instrument layout and observing strategy. All the details can be found in S12. The PPak fiber bundle of the PMAS instrument has a FoV of 74″ × 64″. There are 382 fibers in total, distributed in three different groups. The PPak Integral Field Unit (IFU) holds 331 “science” fibers in a hexagonal grid with a maximum diameter of 74′′  while each fiber projects to 2.̋7 in diameter on the sky. The fiber-to-fiber distance is 3.̋2, which yields a total filling factor of 0.6. An additional set of 36 fibers devoted to measuring the surrounding sky background are distributed in six bundles of 6 fibers each, located in a ring 72′′ from the center. Finally, there are 15 extra fibers connected to the calibration unit.

Every galaxy in the CALIFA sample is observed in the optical range using two different overlapping setups. The V500 low-resolution mode (R ~ 850) covers the range 3745–7500 Å, but it is affected by internal vignetting within the spectrograph giving an unvignetted range of 4240–7140 Å. The blue mid-resolution setup (V1200; R ~ 1650) covers the range 3400–4840 Å  with an unvignetted range of 3650–4620 Å. The resolutions quoted are those at the overlapping wavelength range (λ ~ 4500 Å). To reach a filling factor of 100% across the FoV, a 3-pointing dithering scheme is used for each object. The exposure time per pointing is fixed. We carry out V1200 observations during dark nights with an exposure time of 1800 s per pointing (split in 2 or 3 individual exposures). We take V500 observations during gray nights with 900 s per pointing.

In the following section, we describe the improvements to the CALIFA data reduction pipeline used to produce the DR2 data.

4. Data processing and error propagation

4.1. Overview of the reduction scheme

As described in H13, since V1.3c the CALIFA pipeline has a Python-based architecture (Py3D package). Here we present a summary of the main steps of the reduction process. The particular improvements on the new CALIFA pipeline, V1.5, are described in Sect. 4.2.

Each raw frame is stored in four different FITS files, corresponding to each of the four amplifiers of the detector. As a first step, these four files are combined into a single frame and bias subtracted. We developed a new tool (Husemann et al. 2012) for the detection and clipping of cosmic rays, which uses a Laplacian edge detection scheme combined with a point spread function convolution approach. Relative flexure offsets are estimated with respect to the continuum and arc-lamp calibration frames and the wavelength solution is corrected for each individual science frame. The stray-light map is reconstructed using the gaps between the fiber traces and subtracted from the calibration and science exposures. For spectral extraction, the widths of the fiber profiles are measured using the position of the fibers obtained from the continuum lamp. An optimal extraction algorithm (Horne 1986) is used to extract the spectra based on the previous measurements of the position and widths. The extracted flux, for each pixel in the dispersion direction, is stored in a row-stacked-spectrum file (RSS). The wavelength solution is obtained from the HeHgCd calibration lamp exposure for each CALIFA data set. The spectra are resampled to a linear grid in wavelength and the spectral resolution is homogenized to a common value along the dispersion axis using a Gaussian convolution. Flexure offsets in the dispersion direction are included at this step. The Poisson plus read-out noise (and bad pixel masks) are propagated in the reduction process. For the wavelength solution, errors are analytically propagated during the Gaussian convolution and a Monte Carlo approach is used to estimate the noise vector after the spline resampling of the spectra. Fiber transmission throughput is corrected by comparing the RSS science frames with the corresponding sky exposures taken during twilight.

In the V1.3c pipeline, standard star observations were used to derive the sensitivity curve to perform the flux calibration of the science exposures. Aperture losses were empirically corrected by comparing the CALIFA spectra with SDSS images. In the V1.5 pipeline, we follow a completely new approach described in Sect. 4.2. After flux calibration, frames are corrected for telluric lines. The sky spectrum is obtained by taking the mean of the 30 faintest sky fibers out of the 36 sky fibers of PPak, and then subtracted from the science frames.

The science spectra corresponding to the three dithered exposures are combined into a single frame of 993 spectra. In V1.3c, these three pointings were rescaled to a common intensity, by comparing the integrated spectra within an aperture of 30′′/diameter. The new procedure followed in V1.5 is explained in Sect. 4.2. After correction for Galactic extinction, the RSS is ready for the spatial rearranging of the fibers and creation of the datacube. We use a flux-conserving inverse-distance weighting scheme to reconstruct a spatial image with a sampling of 1′′. The pipeline V1.5 uses a new set of parameters (see Sect. 4.2) that improves the spatial resolution of the datacube. First, we reconstruct the datacube and estimate the differential atmospheric refraction (DAR) effect. In a second step, we reconstruct the cube again where the position of the fiber against the regular grid is changed according to the DAR offset measured in the first reconstruction. This two-stage iteration avoids another resampling step, important for accurate error propagation. Finally, a new procedure is applied for the absolute flux recalibration and astronomical registration, explained in the following section.

4.2. Improvements to the CALIFA data reduction scheme

The main improvements to the current pipeline, V1.5, are: i) a new sensitivity curve for the V500 setup obtained from a dedicated calibration program for several CALIFA elliptical galaxies (Husemann et al., in prep.); ii) a new registering method, comparing individual CALIFA pointings with SDSS images; and iii) an improved image reconstruction method (cube interpolation). Step ii) also improves the absolute photometric matching of the three dithered pointings.

The new version starts with the RSS files of the three individual pointings after sky subtraction produced by pipeline V1.3c. The V500 RSS files are spectrophotometrically recalibrated with the new sensitivity curve. A new estimate for the sensitivity curve was necessary because it was based on standard star observations in V1.3c, which result in significant uncertainties. Although this is a common procedure for optical spectroscopy, it is an issue for CALIFA because of the finite size of the fibers of PPak. Aperture losses can be very significant, and come in two flavors for PPak: 1) a global aperture loss that depends on the relative position of star with respect to its nearest fibers; and 2) a wavelength-dependent aperture loss caused by atmospheric dispersion, which leads to a smooth change in the apparent position of the star and the FWHM of the seeing as a function of wavelength. While the global aperture loss can be corrected for by renormalizing the spectra based upon photometry, the wavelength-dependent losses affect the shape of the sensitivity curve and are harder to correct.

The aperture losses of the stars are mainly introduced by their point-like nature. Wavelength dependent aperture losses are significantly reduced when an extended source with a smooth surface brightness distribution is observed. Bright nearby early-type galaxies, such as those observed by CALIFA, come close to a flat surface brightness distribution except for their very center. In addition, they are predominately composed of old stellar populations, so that the shape of their optical spectra only varies smoothly with radius.

For this reason, we decide to adopt smooth elliptical galaxies as secondary spectrophotometric calibrators from which the sensitivity curves for the CALIFA PMAS-PPak observations are derived. This, in turn, requires that we have this set of secondary calibrators calibrated against primary calibrators, i.e., standard spectrophotometric stars.

A dedicated calibration program was initiated to reobserve about two dozen elliptical CALIFA galaxies chosen as secondary calibrators and the standard stars with the PMAS Lens-Array (LArr) and the V300 grism. The LArr covers a continuous 16″ × 16″ FoV centered on the elliptical galaxies with a sampling of 1″. A robust and accurate spectrophotometric calibration is achieved for these observations because almost the entire flux of the star is captured within the large continuous field, except the flux in the broad wings of the PSF (Sandin et al. 2008). We mainly targeted the primary HST spectroscopic standard stars, which are DA white dwarfs, for which high spectral resolution reference spectra and very accurate model spectra exist. Special care was taken in monitoring the atmospheric extinction curve during the different observing runs by targeting the same standard star twice a night with an airmass difference of about ~1.

During the data reduction we homogenized LArr data taken with the V300 grism to a common spectral resolution of 9 Å (FWHM) across the FoV and across the wavelength range. The sensitivity curve for the LArr data is the ratio of the observed counts in the standard star spectrum (corrected for atmospheric extinction at the given airmass) to the reference spectrum, smoothed to the adopted spectral resolution. Then we calibrate the early-type galaxy observations with the derived sensitivity curve and atmospheric extinction curve for each night comparing the photometry of the calibrated LArr data with aperture matched SDSS photometry in the g and r bands, we achieve an absolute spectrophotometric accuracy of <0.03 mag. Further details about the calibration program and its application to obtain a sensitivity curve for coarse fiber IFU spectrographs will be presented in a separate publication (Husemann et al., in prep.). Below we briefly outline how we processed the CALIFA data to determine the sensitivity curve based on the LArr observations of early-type galaxies.

We start with the reduced, but spectrophotometrically uncalibrated, RSS spectra for individual CALIFA pointings of the early-type galaxies. Again, we need to compare the count spectrum with the reference spectra in physical units to derive the sensitivity curve. Specifically, we smooth the RSS spectra to a spectral resolution of 9 Å (FWHM, significantly coarser than the public CALIFA data) to match the LArr V300 grism characteristics. In addition, we resample the spectra to the heliocentric frame. All CALIFA fibers that overlap with the LArr FoV and are more than 3″ away from the galaxy center are coadded. A corresponding reference spectrum is extracted from the LArr data based on the relative position and apertures of the CALIFA fibers considered. The ratio of the flux-calibrated LArr and uncalibrated CALIFA spectrum then corresponds to the instrumental sensitivity curve. In this case, we adopt the mean atmospheric extinction curve for Calar Alto presented by Sánchez et al. (2007) because the specific extinction curve for a given CALIFA observing night is not measured. Then we smooth the sensitivity curve by a high-order polynomial to obtain a noise-free representation, while masking out spectral regions suffering from telluric absorption. The final master sensitivity curve is computed as the average of all sensitivity curves independently derived from different reference early-type galaxies. We anticipate that most of the remaining systematic spectrophotometric uncertainty for CALIFA will be driven by the uncertainties in the wavelength-dependent atmospheric extinction at the time of each CALIFA observation.

The current pipeline also implements a new scheme for the registration of the images. First, sky-subtracted and calibrated images are created from SDSS DR7 (in the r-band for the V500 setup and the g-band for the V1200) based on the so-called “corrected frames” (fpC)³. Then, the magnitude in the corresponding SDSS filter is computed for each RSS spectrum. The predicted SDSS flux for each CALIFA fiber is estimated using diameter apertures, adopting the PPak layout projected on the SDSS image. This layout is displaced in steps in RA and Dec across a search box in the SDSS image. Then a χ² map is computed to obtain the best offsets for each pointing, taking errors in the flux measurements into account (only fibers with S/N> 3.0 are considered) and allowing for a photometric scaling factor between the SDSS and the CALIFA observations as an additional parameter. The minimum value of the χ² map is used to obtain the best-fitting RA and Dec for the center of the PPak IFU with respect to the center of the CALIFA galaxy seen by SDSS. Figure 9 shows an example of the described procedure. The photometric scale factor at the best matching position is used to rescale the absolute photometry of each particular RSS pointing to bring them onto the same flux scale.

The photometric anchoring to the SDSS images of the V1.5 data is more accurate than that of the previous version. However, there are a few datacubes where the new registering method does not return optimal results, particularly in low surface brightness edge-on galaxies or in the presence of bright foreground field stars⁴. This effect is more likely to occur in the V1200 setup, given its lower S/N on average compared to V500. In these cases, we apply the photometric SDSS matching of pipeline V1.3c described in H13 (to both setups, for the sake of consistency). We included a new “REGISTER” keyword in the header of the datacubes (see Sect. 5.4) and added a dagger symbol to the quality tables (Tables 6 and 7) to easily identify these galaxies.

The third step in the reduction sequence is the interpolation method used to convert from RSS to cube format, aimed at improving the spatial resolution. We use the position of each RSS pointing obtained in the previous step for the image reconstruction. In a series of tests, we found that an inverse-distance weighted image reconstruction scheme performs more favorably than, e.g., the drizzle method (Fruchter & Hook 2002). To increase the spatial resolution and reduce the correlation between nearby pixels, we have reduced the extent of the Gaussian kernel for the interpolation. We adopt 0.75′′ for the dispersion of the Gaussian (instead of 1′′ in V1.3c) and limit the kernel to a radius of 3.5′′ (instead of 5′′). This results in a much sharper image and a lower value for the correlated noise. In the previous pipeline V1.3c, a minimum number of 3 contributing fibers was imposed in the reconstruction of the image to achieve a homogeneous data quality across the field. With the new maximum radius in pipeline V1.5, this type of prescription results in the absence of data in the outer 2″ of the FoV, due to the wider fiber separation in the outer ring of the fiber bundle. Thus, we decided to lower this limit to 1 as the minimum number of fibers needed to fill a spaxel. We have added a new header data unit (see Sect. 5.3) that records the number of fibers used to compute the total flux per spaxel. This allows the user to control which spaxels to include if a particular science case requires a minimum number of fibers for the reconstruction of the flux. The final sampling of the produced datacube is 1″ × 1″ per pixel.

4.3. Characterization of spatially correlated noise

Because of the interpolation procedure used to obtain a regular grid, the output pixels in the final datacube are not independent of one another. The Gaussian interpolation method distributes the flux from a given fiber between several pixels, which are combined with neighboring pixels within a certain radius, as described in Sect. 4.2. This causes the noise in the adjacent pixels to be correlated (in the spatial dimension). The correlation implies that a measurement of the noise in a stacked spectrum of N pixels will be underestimated (noise is underestimated on scales larger than pixel units). Characterizing this effect is essential for estimating the statistical errors when spectra in datacubes are coadded.

First of all, it is important to check that the error spectra derived from the pipeline for individual spaxels are reliable. Spectral fitting analysis can provide an approximate assessment of the accuracy of the error spectra. In Fig. 10 we update Fig. 9 of H13 for DR2 data. The plot shows the histogram of reduced residuals, i.e., the difference between the observed (O_λ) and synthetic (M_λ) spectra obtained with starlight in units of the corresponding error ϵ_λ (details on the fitting procedures can be found in 6.5). The distribution is very well described by a Gaussian centered at 0.03 with σ = 0.87, only slightly less than expected if residuals are purely due to uncorrelated noise.

The correlated noise can be taken into account by providing the spatial covariance (Sharp et al. 2015). However, a more practical approach consists of using the datacubes to calculate the expected rms noise⁵, with the noise correlation ratio β(N), as a function of the number of pixels. To obtain a sample of coadded spaxels with different areas, we used the Voronoi adaptive binning method (implemented for optical IFS data by Cappellari & Copin 2003) with a target S/N of 20. We removed individual spaxels with S/N< 5 from the analysis, and coadded bins with areas larger than 60 spaxels. The β correlation ratio (or correction factor) is the ratio of the “real” or measured error to the analytically propagated error of the binned spectra as a function of bin size. The results obtained for all DR2 datacubes, shown in Fig. 11, can be well described by the logarithmic function $\mathit{\beta }\mathrm{\left(}\mathit{N}\mathrm{\right)}\mathrm{=}\mathrm{1}\mathrm{+}\mathit{\alpha }\log \mathit{N,}$(1)with N the number of spaxels per bin.

The values for the slope α are equal within the errors (0.01) in both setups, with a value of 1.06 for V1200 and 1.07 for V500. The slope is lower than the DR1 value (mean ~1.4), indicating that the noise in DR2 datacubes is less correlated than in DR1. This is expected since we changed the parameters in the interpolation (reducing the number of adjacent fibers contributing to a particular spaxel) and the registering method. In Appendix A we give some instructions on how to estimate the final coadded error spectrum and the limit of the application of Eq. (1).

5. CALIFA data format and characteristics

The CALIFA data are stored and distributed as datacubes (three-dimensional data) in the standard binary FITS format and consist of several FITS Header Data Units (HDU). These datacubes contain, (1) the measured flux densities, corrected for Galactic extinction as described in S12, in units of 10^-16 erg s^-1 cm^-2 Å^-1 (primary datacube); (2) associated errors; (3) error weighting factors; (4) bad pixels flags; and (5) fiber coverage (Table 2). The last HDU is new added content absent in DR1, as explained in Sect. 4.2, but the others share the same properties as the previous data release. The first two axes of the cubes correspond to the spatial dimension along right ascension and declination with a 1″ × 1″ sampling. The third dimension represents the wavelength and is linearly sampled. Table 3 summarizes the dimensions of each datacube (N_α, N_δ, and N_λ), as well as the spectral sampling (d_λ) and constant resolution (δ_λ) along the entire wavelength range.

5.1. Error and weight datacubes

The 1σ noise level of each pixel as formally propagated by the pipeline can be found in the first FITS extension. Section 4.3 discusses the accuracy of the formal noise and the correlation, important when CALIFA data need to be spatially binned, and an empirical function is provided to account for the correlation effect. The second FITS extension (ERRWEIGHT) stores the error scaling factor for each pixel in the limiting case that all valid spaxels of the cube would be coadded (see also Appendix A). In the case of bad pixels, we assign an error value that is roughly ten orders of magnitude higher than the typical value.

5.2. Bad pixel datacubes

Bad pixel datacubes are stored in the third FITS extension (BADPIX). This information, in combination with the error vector, is essential to properly account for the potential problems in each spaxel. Pixels with flag =1 report the absence of sufficient information in the raw data due to cosmic rays, bad CCD colums, or the effect of vignetting⁶. These bad pixels have been interpolated over and we strongly suggest not to use them for any science analysis.

Finally, the uncovered corners of the hexagonal PPak FoV are filled with zeros and flagged as bad pixels for consistency. The residuals of bright night-sky emission lines are not flagged as bad pixels.

5.3. Fiber coverage datacubes

Pipeline V1.5 adds a new FITS extension (FIBCOVER) to the datacubes, not available in previous DR1 datacubes. As explained in Sect. 4.2 we have reduced the maximum distance of fibers that can contribute to the flux of a given spaxel. The outer hexagonal-ring of fibers do not have the same coverage as any other fiber inside the hexagon. In pipeline V1.3c we imposed a minimum of 3 fibers for computing the flux of given spaxel. In V1.5, with the new radius limit this would yield an empty outer hexagonal-ring of ~2′′ in the FoV. Thus, we relax to 1 the minimum number of fibers. In order to control which spaxels have enough flux “resolution”, we include a new HDU reporting the number of fibers used to account for the computed flux.

5.4. FITS header information

The FITS header contains the standard keywords that encode the information required to transform the pixel-space coordinates into sky and wavelength-space coordinates, following the World Coordinate System (WCS, Greisen & Calabretta 2002). Each CALIFA datacube contains the full FITS header information of all raw frames from which it was created. Information regarding observing and instrumental conditions such us sky brightness, flexure offsets, Galactic extinction or approximate limiting magnitude is also kept in the FITS header of each datacube. See Sect. 4.3 of H13 for nomenclature and their Table 4 for a summary of the main header keywords and meaning.

The most important new keyword added in DR2 datacubes is “REGISTER” and takes a boolean value. It indicates if a particular datacube has been successfully registered using the new method explained in Sect. 4.2 (True) or it has used the old V1.3c scheme (False). Datacubes with a False value are marked with a dagger in Tables 6 and 7.

6. Data quality

This second CALIFA data release (DR2) provides science-grade data for a sample of 200 galaxies, including the 100 galaxies released in the first data release (DR1), identified by an asterisk in Tables 6 and 7. As for DR1, we have run a careful quality control (QC) on the data products and selected only those galaxies that passed a series of QC checks in both setups (V500 and V1200), as we detail in this section. The QC checks are based on a set of measured parameters and/or visual inspection, resulting in a set of flags that allows one to quickly assess the quality of the data and their suitability for scientific use. Quantities and flags are organized into three distinct categories, related to: observing conditions (denoted by the obs prefix); instrumental performance and effectiveness of the data reduction (red); accuracy and quality of the final data products (cal). The flags in each category are computed based on thresholds of measured quantities, possibly combined with flags given by human classifiers based on visual inspection, as detailed below and summarized in Tables 4 and 5. Thresholds are determined from the distribution of the parameters to exclude outliers and also to analyze the effects of anomalous parameters on the final quality of the datacubes. The tables of the relevant QC parameters, along with the QC flags, are available on the DR2 website.

Each flag can have one of the following values:

• −1 = undefined;

• 0 = good quality – ok;

• 1 = minor issues that do not significantly affect the quality – warning;

• 2 = significant issues affecting the quality – bad.

By selection, DR2 excludes galaxies with bad flags, with just a few minor exceptions affecting previously released DR1 galaxies: in these cases, the revised QC criteria adopted here would have prevented us from including these galaxies in the DR, but given the incremental nature of our data releases we keep them in the current sample.

In naming the QC parameters, we adopt the following convention: the first part is the category prefix (obs, red or cal), followed by a measured parameter, and sometimes a final suffix indicating the statistics applied to combine the parameter as measured in different observations/pointings/fibers (i.e., mean, min, max, rms).

In the following subsections, we describe the QCs in each of the above-mentioned categories.

6.1. Quality of the observing conditions (obs)

Three quantities are considered crucial in determining the quality of the observing conditions of the CALIFA data: the airmass, the brightness of the sky, and the atmospheric extinction. While seeing is in general an important parameter of the observing conditions, the imaging quality and spatial resolution of the CALIFA cubes is mostly limited by the sampling of the fibers on the plane of the sky and the resampling process (see Sect. 6.4.1 for more detail), rather than by the seeing. Moreover, the seeing measurement is only available for a small fraction of the objects (see Sect. 6.4.2), and therefore cannot be used as a reliable QC parameter.

For the airmass, we consider the average and the maximum airmass of the observations over all contributing pointings (obs_airmass_mean and obs_airmass_max) and its rms (obs_airmass_rms). For each of these quantities, we defined two thresholds (the same for V500 and V1200, see Tables 6 and 7), above which the warning or the bad flags, respectively, are raised. The combined flag_obs_am is the worst of the three cases.

The surface brightness of the sky in V-band during the observations is another critical parameter, which mainly limits the depth of the observations and the accuracy of the sky subtraction. The quantity skymag is measured in each pointing from the sky spectrum obtained from the 36 sky fibers⁷. The mean and the rms over all pointings are considered to define the corresponding flags. Note that stricter requirements are applied to V1200 data (blue setup, high resolution) with respect to the V500 data.

The transparency of the sky during each pointing (ext) is obtained from the monitored V band extinction at the time of the observation. We consider the following properties as symptoms of low/bad quality observations: large extinctions on average, a large maximum extinction or a large rms variation across the pointings (indicating inhomogeneous observing conditions).

6.2. Quality of the instrumental/data reduction performance (red)

The quality of the instrumental and data reduction performance is assessed via a series of four quantities measured on the reduced data before combining them into the final datacube: straylight, spectral dispersion, cross dispersion cdisp, and the residuals from the subtraction of bright skylines (namely, the 5577 Å O₂ line in the V500 setup and the 4358 Å Hgi in the V1200 setup). In addition, we consider the limiting surface brightness corresponding to a 3σ detection per spaxel and spectral resolution element measured on the final cube.

The so-called straylight is an additional source of illumination internal to the instrument, possibly as a distributed scattered light component. Straylight, if not subtracted properly, introduces systematic errors and thus limits the final sensitivity and accuracy of the data reduction⁸. High mean levels of straylight in a frame (meanstraylight), as well as high maximum values (maxstraylight) and large rms (rmsstraylight), are indications of poor performance. Levels above the thresholds provided in Tables 4 and 5 in at least one of the exposures (_max suffix) raise a warning or a bad flag_red_straylight flag.

The spectral dispersion and cross dispersion are measured on individual fiber spectra as the FWHM of skylines and the FWHM of the spectral trace⁹, respectively. Thresholds are set on the mean values to ensure that the typical parameters do not depart too much from the nominal target specifications, and on the maximum and rms to check for anomalies in the data. Any failure to comply with the thresholds reported in Tables 4 and 5 raises a flag_red_disp¹⁰ or flag_red_cdisp.

To assess the performance of the sky subtraction, we consider the minimum and the maximum over all pointings of the average (over all fibers) flux residual of a bright skyline within an individual pointing (red_res4358_min and red_res4358_max, and red_res5577_min and red_res5577_max for the V1200 and the V500 setup, respectively). We also consider the maximum over all pointings of the rms residuals (over all fibers in an individual pointing), red_rmsres4358_max and red_rmsres5577_max. Average residuals that are too negative or too positive are indications of systematic bias in the sky subtraction. An rms that is too large can be regarded as a symptom of localized failures or noisy data. In these cases, the flag_red_skylines is set.

Finally, the 3σ continuum flux density detection limit per interpolated 1 arcsec²-spaxel and spectral resolution element¹¹ for the faintest regions is used to identify cubes whose depth does not fulfill the survey requirements, which is reflected in the flag_red_limsb flag. More about the depth of the final datacubes is discussed in Sect. 6.6.

6.3. Quality of the calibrated data products (cal)

The quality of the calibrated data products is determined by checks on the global spectrophotometry, on the stability of the wavelength calibration across the spectral range, and on the quality of the resulting 2D flux distribution (synthetic image) and its ability to match the SDSS broadband imaging.

The quality of global spectrophotometric calibration is assessed by comparing the photometric fluxes derived from spectra integrated within 30′′-radius apertures with the corresponding fluxes derived from SDSS imaging, as explained below in Sect. 6.5. For the V500 setup, in particular, it is possible to derive the flux ratio between SDSS and CALIFA in g and r-bands (cal_qflux_g and cal_qflux_r, averaged over all pointings for a given galaxy): values of these ratios departing from 1 by more than the tolerances listed in Table 4 are flagged. Large rms variations of these values over the three V500 pointings (cal_qflux_rms, which combines g and r bands) are also considered symptoms of poor quality. In addition to these quantitative parameters, we visually check that the spectral energy distribution (SED) measured via SDSS photometry matches the CALIFA integrated spectrum. For this check we also consider the u and the i band data points: although the CALIFA spectra do not cover the full extent of these pass-bands, they prove helpful in judging the matching of spectral shapes. Five members of the collaboration performed these checks independently and assigned flags ok-warning-bad: the second-to-worst classification is retained. This flag is then combined with the flags based on the quantitative flux ratios to create the final flag_cal_specphoto flag.

In order to check the stability of the wavelength calibration over the full spectral range we performed the same measurements presented in Sect. 5.3 of H13: for each galaxy and setup, the spectra within 5′′ of the center of the galaxy are integrated and the systemic velocity is estimated first for the full spectrum and then for 3 (4) independent spectral ranges in V1200 (V500). The rms of these values with respect to the systemic velocity from the full spectrum (cal_rmsvelmean) is an estimate of the stability of the wavelength calibration across the wavelength range and is used to set the corresponding quality flag flag_cal_wl. In >97.5% of the cases, we obtain cal_rmsvelmean well below 2 km s^-1 for the V1200 and 3 km s^-1 for the V500 grating.

Finally, the quality flag on the 2D flux distribution and plane-of-sky registration, flag_cal_ima, is defined by combining the information on the goodness of matching between SDSS images and synthetic images from the CALIFA datacube with a series of visual checks. The former is provided by the chi-squared of the registration procedure (see Sect. 4.2). The visual checks include: a check on possible artefacts in the synthetic broadband image from the final CALIFA cubes (e.g., mismatched features, elongated PSF); a comparison of the CALIFA fiber footprints of each pointing with the registered SDSS image, looking for apparent mismatched and miscomputed spatial offsets; a check of the chi-squared surface plot displaying the dependence of the registration procedure (see below) on the x and y spatial offsets, whereby irregular chi-square surfaces and lack of clear minima imply the possibility of an inaccurate registration. Out of five independent classifiers we chose the median value of the attributed flags and combine it with the flag corresponding to the chi-squared measurements. We note that a small number of objects already released as part of DR1 do not reach the imaging quality standards using the registration procedure adopted in the pipeline V1.5 (see Sect. 4.2), which uses cross-correlation with SDSS images: in these cases we revert to the old registration scheme adopted for DR1 (pipeline V1.3c) and mark the objects with a dagger in Tables 6 and 7.

6.4. Astrometric accuracy and spatial resolution

6.4.1. Astrometric registration accuracy

Pipeline V1.5 implements a new method (see Sect. 4.2) to register the absolute astrometry of the datacube coordinate system to the International Coordinate Reference System (ICRS). The previous pipeline, V1.3c, used tabulated coordinates of the galaxy V band photometric center that were assigned to the barycenter measured in the reconstructed image from the datacubes (just one point, instead of the global match applied in V1.5).

To check the accuracy of the new astrometric registration for V500 and V1200 datacubes, we performed independent tests using SDSS r and g-band images (DR10) for each galaxy. Synthetic r and g-band PPaK images were computed using the V1.5 reduced data. The coordinates of the peak centroid PPAK images are used as an approximate galactic center, and the corresponding peak was measured in the SDSS images. The offsets between the SDSS and CALIFA are less than 3′′ (rms ~1′′) for the majority of the DR2 sample. Large offsets are mostly due either to edge-on galaxies, centers of the galaxies not well defined because of dust lanes, irregular morphologies or bright field star(s) near the center of the galaxy. Objects with offsets larger than 3′′ measured in V500 setup are: IC1 652, NGC 0444, UGC 00809, UGC 00841, NGC 0477, IC 1683, NGC 0499, NGC 0496, NGC 0528, UGC 01938, NGC 1056, NGC 3991, MCG-01-01-012, and NGC 7800. For the V1200 setup: IC1 528, IC1 652, NGC 0444, UGC 00809, UGC 00841, NGC 0477, NGC 0499, NGC 0496, NGC 0528, UGC 02222, NGC 3991, UGC 11792, MCG-01-01-012, and NGC 7800.

6.4.2. Seeing and spatial resolution

To cover the complete FoV of the central bundle and to increase the final resolution of the CALIFA datacubes (PPak fibers have a diameter of 2.7′′), we adopt a dithering scheme with three pointings, as described in S12. For imaging, in addition to the telescope aperture, instrumental and atmospheric seeing determine the final spatial resolution. This has to be added to the particular IFU characteristics.

The average atmospheric seeing conditions throughout CALIFA observations were derived from the measurements acquired by the DIMM (DIMM, Aceituno 2004), which operates fully automatically at the Calar Alto observatory during the night. The DIMM has different operational constraints from the 3.5 m telescope (humidity lower than 80% and wind speed less than 12 m s^-1). Thus seeing information is not available for every CALIFA observation and these values are missing from Tables 6 and 7, but the overall seeing distribution is not expected to be very different¹². Figure 12 shows the DIMM seeing distribution for the DR2 sample, which has a median value of FWHM (the distribution is very similar to the DR1 sample), and therefore atmospheric seeing is not a limiting factor in the spatial resolution of the CALIFA cubes. At any rate, the final spatial resolution of the CALIFA data is mainly set by fiber size and the dithering and interpolation scheme.

We use the following approach to measure the PSF in the datacubes. Since January 2012 standard stars were observed using the same dithering pattern adopted for the science observations for the V500 setup. We observed a total of 107 nights in this period. Only 70% of the nights had weather conditions good enough to acquire a calibration star and 2/3 were observed adopting the dithering scheme, yielding a total of 45 datacubes. We reduced these data using the same procedure described before for the science objects. The PSF can be measured very precisely because the calibration stars have a very high S/N. We synthesize a SDSS g-band image simulated from the datacubes for each of these stars. For each of these images, we fit a 2D Moffat profile using the software imfit (Erwin 2015)¹³. Figure 13 shows the normalized distributions of FWHM and β_M parameters of the Moffat profile, weighted by the likelihood of the fit. We obtain a mean value and 1 σ scatter of the FWHM = 2.39 ± 0.26 arcsec, with β_M = 1.73 ± 0.11. The ellipticity (1 − b / a, with a and b being the semimajor and semiminor axes, respectively) is also measured, with mean value and 1 σ scatter of 0.08 ± 0.06. Given the uncertainties, this value means the PSF can be considered effectively axisymmetric. The uncertainties in these measurements correspond to 1 σ of the distributions.

As a consequence of the improvements in the interpolation scheme, the PSF FWHM has substantially decreased with respect to DR1 (pipeline V1.3c). The improvement in spatial resolution is illustrated on the Hα maps presented in Fig. 15. This figure shows Hα maps obtained using FIT3D on the CALIFA datacubes for NGC 5406 (ID 684) for DR1, DR2, and one image taken with the William Herschel Telescope (WHT) using a narrowband filter. The last image has also been degraded to the DR2 nominal resolution for the sake of comparison. This improvement impacts directly, for example, on the detection rate of H ii regions. Using hiiexplorer (Sánchez et al. 2012b) on the V1.3c datacubes of the 200 galaxies, a total of 5878 are recovered, while this number rises to 7646 H ii regions for the DR2 galaxies using pipeline V1.5, which represents an increase of ~30%.

6.5. Spectrophotometric accuracy

As mentioned in Sect. 4.2, the new registration scheme of the pipeline uses r-band for the V500 setup and g-band for the V1200 of the SDSS DR7. Each V500 datacube is rescaled in the absolute flux level to match the SDSS DR7 broadband photometry using the photometric scale factor at the best matching position for each pointing. On the other hand, the V1200 data is matched to the V500 data (S12). This procedure, together with the new recalibrated sensitivity curve (see Sect. 4.2 and Husemann et al., in prep.), improves the spectrophotometric calibration over DR1. This is clearly shown in Fig. 14. As part of the CALIFA pipeline V1.5, a 30′′ diameter photometric aperture in r and g is measured both in the SDSS DR7 images and the equivalent synthetic CALIFA broadband images. The mean SDSS/CALIFA g and r band ratios in DR2 and their scatter are 1.00 ± 0.05 and 0.99 ± 0.06, respectively. In the right panel of Fig. 14 the distribution in Δ(g − r) color difference between the SDSS and CALIFA data shows that the spectrophotometric accuracy across the wavelength range is better that 3%, with a median value of 0.01 ± 0.03.

Spectral fitting methods can be used to perform useful tests of the data and their calibration, and this has been done before in CALIFA. H13 used starlight fits to evaluate the accuracy of the error estimates in DR1 datacubes, while Cid Fernandes et al. (2014) used these fits to map systematic features in the spectral residuals that may indicate calibration issues.

We repeated the same experiments for the DR2 datacubes. Results are shown in Fig. 16. The top panel shows in blue the mean spectrum of 170670 Voronoi bins of the 200 galaxies in DR2¹⁴. The average is taken after normalizing each spectrum by its median flux in the 5635 ± 45 Å window. The mean synthetic spectrum (overplotted orange line) as well as the mean residual (at the bottom of the upper panel, purple line) are also plotted. The middle panel zooms in on the residual spectrum, which now excludes emission lines and bad pixels, which are masked in the fitting process. Finally, the bottom panel shows what fraction of all spectra was used in the statistics at each λ.

The layout of Fig. 16 is identical to Fig. 13 of Cid Fernandes et al. (2014), to which it should be compared¹⁵. Focusing on the middle panel, one sees that from ~5000 Å to the red the residuals are very similar, including the humps around 5800 Å associated with the imperfect removal of telluric features. Toward the blue however, the new reduction pipeline leads to smaller residuals. For instance, the broad feature around Hβ seen with V1.3c spectra is essentially eliminated by the new reduction. A systematic excess blueness persists for λ< 3900 Å, but overall the improvement is clear.

Residuals for the 200 DR2 nuclear spectra are shown in Fig. 17, where galaxies are sorted by redshift and stacked. This visualization facilitates the identification of telluric features, which appear as slanted lines in the image. Comparison with an identical plot in H13 (their Fig. 16) shows the improvements achieved with the new pipeline.

6.6. Limiting sensitivity and signal-to-noise

To assess the depth of the data, we estimated the 3σ continuum flux density detection limit per interpolated 1 arcsec²-spaxel and spectral resolution element for the faintest regions. Figure 18 shows the limiting continuum sensitivity of the spectrophotometrically recalibrated CALIFA spectra. The depth is plotted against the average S/N per 1 arcsec² and spectral resolution element within an elliptical annulus of ± 1″ around the galaxies’ r-band half-light semimajor axis (HLR), with PA and radius values taken from W14. A narrow wavelength window at 4480–4520 Å for the V1200 and at 5590–5680 Å for the V500 was used to estimate both values¹⁶. These small windows are nearly free of stellar absorption features or emission lines. The 3σ continuum flux density detection limit per spaxel and spectral resolution element¹⁷ for the V1200 data (I_3σ = 3.2 × 10^-18 erg s^-1 cm^-2 Å^-1 arcsec^-2 in the median at 4500 Å) is a factor of ~2–3 brighter than for the V500 data (I_3σ = 1.2 × 10^-18 erg s^-1 cm^-2 Å^-1 arcsec^-2 in the median at 5635 Å) mainly because of the difference in spectral resolution. These continuum sensitivities can be transformed into equivalent limiting broadband surface brightnesses of 23.0 mag arcsec^-2 in the g-band for the V1200 data and 23.4 mag arcsec^-2 in the r-band for the V500. The variance of the sky brightness of each night might be one of the main factors causing the dispersion in the limiting continuum sensitivity. Dust attenuation, transparency of the night, and other atmospheric conditions might also affect the depth achievable at fixed exposure times.

The limiting sensitivity is a measure of the noise and thus it correlates mildly with the S/N. The mean S/N in the continuum per 1 arcsec² and spectral resolution element at the half-light semimajor axis (HLR) of all objects is ~9.5 for the V1200 setup, while it is ~22.2 for the V500 data. Thus, we achieve a for a significant number of the objects even for the V1200 setup.

7. Access to the CALIFA DR2 data

7.1. The CALIFA DR2 search and retrieval tool

The public data are distributed through the CALIFA DR2 web page¹⁸. A simple web form interface, already in use for the first data release, allows the user to select data of a particular target galaxy, or a subsample of objects within some constraints on observing conditions or galaxy properties. Among the selection parameters, we include the instrument setup, galaxy coordinates, redshift, g-band magnitudes, observing date, Hubble type, bar strength, inclination estimated from axis ratio, V band atmospheric attenuation, airmass, and relative accuracy of the SDSS/CALIFA photometric calibration.

If any CALIFA data sets are available given the search parameters, they are listed in the follwing web page and can be selected to be downloaded. The download process requests a target directory on the local machine to store the data, after the downloading option is selected. The CALIFA data are delivered as fully reduced datacubes in FITS format separately for each of the two CALIFA spectral settings, i.e., the V500 and V1200 setup. Each DR2 data set is uniquely identified by their file name, GALNAME.V1200.rscube.fits.gz and GALNAME.V500.rscube.fits.gz for the V1200 and V500 setup respectively, where GALNAME is the name of the CALIFA galaxy listed in Table 1.

All the QC tables discussed throughout this article are also distributed in CSV and FITS-table formats in the same webpage. In addition, we distribute the more relevant tables discussed in W14 regarding the characterization of the MS, using similar formats. These tables could be useful in further science explorations of the cubes.

7.2. Virtual observatory services

CALIFA data is also available through several Virtual Observatory (VO) facilities:

• 1.

  The FITS files of the full cubes are accessible through GAVO’sObsCore (Louys et al. 2011)service, which is part of the TAP (Dowleret al. 2011)service¹⁹. ObsCore provides ahomogeneous description of observational data products of allkinds and thus allows for a global data set discovery. The systemalready supports the upcoming IVOA DataLink standard forperforming cutouts and similar server-side operations.

• 2.

  At the same TAP endpoint, the califadr2.cubes and califadr2.objects tables enable queries versus CALIFA-specific metadata, and in particular, the quality control parameters given in Tables 6 and 7.

• 3.

  Individual, cutout spectra can be located and retrieved from the CALIFA SSA service²⁰; advanced SSAP clients like Splat (Draper 2014) also support server-side spectral cutouts on this service via a DataLink prototype.

• 4.

  The spaxels can also be queried in database tables via GAVO’s TAP service mentioned above (the tables are called califadr2.fluxv500 and califadr2.fluxv1200).

An overview of VO-accessible resources generated from CALIFA, possibly updated from what is reported here, is available at http://dc.g-vo.org/browse/califa/q2. This page also gives some usage scenarios for CALIFA VO resources.

8. Summary

In this article we have presented the main characteristics of the second public data release of the Calar Alto Legacy Integral Field Area (CALIFA) survey. This data release comprises 200 galaxies (400 datacubes) containing more than 1.5 million spectra²¹, covering a wide range of masses, morphological types, colors, etc. This subset of randomly selected objects comprises a statistically representative sample of the galaxies in the Local Universe. The CALIFA DR2 provides science-grade and quality-checked, integral-field spectroscopy publicly distributed to the community²².

We described in detail the main quality parameters analyzed in the validation process, which are provided to the users with complete tables to select the objects for their science cases. We reduced the data using a new version of the pipeline (V1.5), which considerably improves the quality of the data in terms of: (i) the spatial resolution; (ii) the covariance between the adjacent spectra; and (iii) the spectrophotometric calibration.

Compared with other ongoing major surveys, CALIFA offers a similar spatial resolution. The PSF of the datacubes has been improved considerably, with a mean value of ~2.5′′ (Sect.6.4.2), similar to SAMI (Sharp et al. 2015). In the case of MaNGA, the combination of an average seeing at the Sloan Telescope (~1.5′′) and the fiber size (2′′), would produce a PSF with a very similar FWHM. The redshift range of SAMI and MaNGA surveys is considerably larger than CALIFA, reaching up to z ~ 0.1. This means that only galaxies at the lowest redshift range in SAMI and MaNGA will offer a similar physical resolution. On the other hand, the spatial coverage of CALIFA is larger than any of those surveys, both in physical and in projected terms (five times larger than SAMI and two times larger than MaNGA). In summary, CALIFA is the survey that samples the galaxies with the largest number of spatial elements for the largest FoV. The penalty for this wider coverage is the lower number of galaxies observed (6 times lower than SAMI and 15 times lower than MaNGA), and a lower spectral resolution of CALIFA over the full wavelength range.

The dataset analysed so far have produced significant advances in our knowledge of the stellar and gas composition in galaxies, their kinematical structure, and the overall star formation history and chemical enrichment (as reviewed in the introduction). We have uncovered new local relations within galaxies, tightly connected to the global relations described using classical spectroscopic surveys. With this new DR, we open to the astronomical community the possibility to futher analyze the spatially resolved properties of galaxies, presenting a panoramic view of the galaxy properties.

---

Acknowledgments

CALIFA is the first legacy survey being performed at Calar Alto. The CALIFA collaboration would like to thank the IAA-CSIC and MPIA-MPG as major partners of the observatory, and CAHA itself, for the unique access to telescope time and support in manpower and infrastructures. The CALIFA collaboration thanks also the CAHA staff for the dedication to this project. R.G.B., R.G.D., and E.P. are supported by the Spanish Ministerio de Ciencia e Innovación under grant AYA2010-15081. S.Z. is supported by the EU Marie Curie Integration Grant “SteMaGE” Nr. PCIG12-GA-2012-326466 (Call Identifier: FP7-PEOPLE-2012 CIG). J.F.B. acknowledges support from grants AYA2010-21322-C03-02 and AIB-2010-DE-00227 from the Spanish Ministry of Economy and Competitiveness (MINECO), as well as from the FP7 Marie Curie Actions of the European Commission, via the Initial Training Network DAGAL under REA grant agreement number 289313. Support for L.G. is provided by the Ministry of Economy, Development, and Tourism’s Millennium Science Initiative through grant IC12009, awarded to The Millennium Institute of Astrophysics, M.A.S. L.G. also acknowledges support by CONICYT through FONDECYT grant 3140566. A.G. acknowledges support from the FP7/2007-2013 under grant agreement n. 267251 (AstroFIt). J.M.G. acknowledges support from the Fundação para a Ciência e a Tecnologia (FCT) through the Fellowship SFRH/BPD/66958/2009 from FCT (Portugal) and research grant PTDC/FIS-AST/3214/2012. RAM was funded by the Spanish programme of International Campus of Excellence Moncloa (CEI). J.M.A. acknowledges support from the European Research Council Starting Grant (SEDmorph; P.I. V. Wild). I.M., J.M. and A.d.O. acknowledge the support by the projects AYA2010-15196 from the Spanish Ministerio de Ciencia e Innovación and TIC 114 and PO08-TIC-3531 from Junta de Andalucía. AMI acknowledges support from Agence Nationale de la Recherche through the STILISM project (ANR-12-BS05-0016-02). M.M. acknowledges financial support from AYA2010-21887-C04-02 from the Ministerio de Economía y Competitividad. P.P. is supported by an FCT Investigador 2013 Contract, funded by FCT/MCTES (Portugal) and POPH/FSE (EC). P.P. acknowledges support by FCT under project FCOMP-01-0124-FEDER-029170 (Reference FCT PTDC/FIS-AST/3214/2012), funded by FCT-MEC (PIDDAC) and FEDER (COMPETE). T.R.L. thanks the support of the Spanish Ministerio de Educación, Cultura y Deporte by means of the FPU fellowship. PSB acknowledges support from the Ramón y Cajal program, grant ATA2010-21322-C03-02 from the Spanish Ministry of Economy and Competitiveness (MINECO). C.J.W. acknowledges support through the Marie Curie Career Integration Grant 303912. V.W. acknowledges support from the European Research Council Starting Grant (SEDMorph P.I. V. Wild) and European Career Re-integration Grant (Phiz-Ev P.I. V. Wild). Y.A. acknowledges financial support from the Ramón y Cajal programme (RyC-2011-09461) and project AYA2013-47742-C4-3-P, both managed by the Ministerio de Economía y Competitividad, as well as the “Study of Emission-Line Galaxies with Integral-Field Spectroscopy” (SELGIFS) programme, funded by the EU (FP7-PEOPLE-2013-IRSES-612701) within the Marie-Sklodowska-Curie Actions scheme. We thank the referee David Wilman for very useful comments that improved the presentation of the paper.

References

Appendix A: Computing the error spectrum for co-added spectra

Some science cases require a minimum S/N in the spectra, especially in the outer parts of the galaxies. This is achieved by spatially coadding spaxels in the datacubes, often by means of an adaptive binning method, such as the Voronoi-binning scheme, implemented for optical IFS data by Cappellari & Copin (2003). However, the final error spectrum of the coadded spectra cannot be simply quadratically summed since the spectra are not independent of each other. As described in Sect. 4.2, we adopt an inverse-distance weighted image reconstruction which, like many other image resampling schemes, introduces a correlation between spaxels in the final datacube. In Sect. 4.3, we provide an equation that relates the analytically propagated error recorded in the datacubes with the final “real” error of the coadded spectrum²³.

Let B be a bin of size N spectra, i.e., we want to coadd N spectra and compute the corresponding error spectrum for that bin. Since we are adding the flux to obtain an integrated spectrum, first we need to add the errors of each individual spectra in quadrature, $\begin{array}{ccc}{\mathit{ϵ}}_{\mathit{B}}^{\mathrm{2}}\mathrm{=}\sum _{\mathit{k}\mathrm{=}\mathrm{1}}^{\mathit{N}}{\mathit{ϵ}}_{\mathit{k}}^{\mathrm{2}}\hspace{0.17em}\mathit{.}& & \end{array}$This would be the error spectrum of the bin B if the spaxels where completely independent. To account for the correlated noise, we simply need to multiply by the corresponding “correlation factor” (Eq. (1)) for a given number of spectra in a particular bin, $\begin{array}{ccc}{\mathit{ϵ}}_{\mathrm{real}\mathit{,B}}^{\mathrm{2}}\mathrm{=}\mathit{\beta }\mathrm{(}\mathit{N}{\mathrm{)}}^{\mathrm{2}}\mathrm{\times }{\mathit{ϵ}}_{\mathit{B}}^{\mathrm{2}}\mathit{,}& & \end{array}$when the bin B contains a large number of spaxels (N 80), the use of Eq. (1) is not recommended. In this case, the ERRWEIGHT HDU extension of the CALIFA FITS file datacube should be used (see Table 2) as a correction factor for each spaxel, $\begin{array}{ccc}{\mathit{ϵ}}_{\mathit{B}}^{\mathrm{2}}\mathrm{=}\sum _{\mathit{k}\mathrm{=}\mathrm{1}}^{\mathit{N}}{\mathit{w}}_{\mathit{k}}^{\mathrm{2}}\mathrm{\times }{\mathit{ϵ}}_{\mathit{k}}^{\mathrm{2}}\hspace{0.17em}\mathit{,}& & \end{array}$where w_k is the error weight of each individual spaxel. The error weighting factor is estimated for each pixel such that the formal error of the coadded spectrum of the entire cube is identical at the obtained by coadding the individual 993 fibers of the RSS.

All Tables

All Figures