DR1 Caveats and known issues
Stellar locus calibration and galactic extinction
One important aspect of the calibration with the stellar locus is that implicitly produces a partial correction of the galactic extinction.
The photometric calibration with the stellar locus is done constructing beforehand a reference stellar locus as the median of the stellar locus of several dozens of pointings calibrated using other calibration procedures. In constructing the stellar locus no correction by galactic extinction is performed, and therefore, part of the offset observed between the stellar locus in differente pointings is due to differences in galactic extinction. Since J-PLUS is surveying the same sky areas as J-PAS, and these have been defined as areas with low galactic extinction, the median stellar locus will be affected by a low galactic extinction. As a result, the calibration with respect to this reference stellar locus will bring the photometric calibration of all the pointings to a photometric system which already takes into account the galactic extinction. The galactic extinction is not fully corrected because still remains the small effect on the median stellar locus.
Pseudo-spectra versus real spectra
Multiband photometry of J-PLUS (and J-PAS) is in between traditional photometry and traditional spectroscopy, and the analysis of the spectral energy distributions (SEDs) in many aspects is similar to the analysis of spectra obtained with traditional spectrographs.
However, careful should be taken when analysing J-PLUS pseudo-spectra (or photo-spectra) as if they were actual spectra to avoid misinterpretations:
- Data of photo-spectra are not taken simultaneously. Although for
most of the objects the time variable is not important, for
transient objects this is crucial. For two types of objects, the
time domain is particularly important:
- Moving objects. For moving objects, the position of the objects changes between observations in different bands and, therefore, the dual-mode photometry has no sense.
- Fast variable objects. Objects whose photometry changes in time scales of few minutes will show pseudo-spectra that sometimes will resemble traditional spectra of objects with emission lines and, in extreme cases, they won't resemble the traditional spectra of any object.
Keeping this in mind is of paramount importance in particular when dealing with outliers in relations traditionally used with data from actual spectra.
- Data points of photo-spectra are highly independent. While data
points of a traditional spectrum are taken and calibrated
simultaneously, data points of photo-spectra are almost
independent. The following aspects should be considered when
interpreting a photo-spectrum:
- Differences in sky conditions. Data of the same object in different photometric bands could be taken in different sky conditions. Although the survey strategy has been devised to minimize these differences trying to carried out sequentially the observations in all bands of a given pointing, changes in the sky conditions can happen and, in some cases, observations are carried out in different nights.
- Differences in PSF. Differences among images taken with different filters arise because of the dependency of the PSF with the wavelength as well as due to variation in seeing conditions. Therefore, a consistent pseudo-spectrum should be constructed trying to minimize these differences to avoid the introduction of spurious color gradients. There are several ways to approach this problem and for this reason we provide photometry following different recipes which try to obtain consistent pseudo-spectra.
- Photometric calibration. The photometric calibration of each band can be done in different ways (see the Photometric calibration section) which introduce different degrees of correlation among different bands. While the calibration with the spectrophotometric standard star is the one making the calibration among bands more independent, the calibration with the stellar locus is the one tying at higher level the calibration among all the bands. Depending on the calibration procedure used to calibrate the images, the effects (and in particular the spurious effects) could be different in each data point.
Fake spectral features
Fake spectral features (especially fake emission lines) can appear as consequence of several issues among the known ones are:
- Impact of cosmic rays. Although several procedures have been implemented to minimize the impact of cosmic rays hits on the images (eg., used of LACosmix to detect and mask cosmis rays on single frames, and sigma-clipping rejection in the construction of coadded images), a small fraction of non-detected cosmic rays still remains and they could mimic an emission line in that band affected by the hit. A way to check whether this could happen in a given object is to visually check the poststamps that can be generated when exploring a single object.
- Variability. Changes in the overall flux of an object during the observations will be seen only in a subsample of the bands and if this affect only one band could be confused with an emission line.
Issues with the treatment of large resolved galaxies
The automatic treatment of large resolved galaxies is very complex and, in this data release, they haven't been treated in any special way. As a result, the photometry for these large galaxies is unreliable being affected by at least the following issues:
- Background subtraction. The background subtraction
configuration is unique for all the pointings and it is optimized to
remove large scale background patterns. This background subtraction
is done on each single frame before combining them for generating
the coadded image, and then again when constructing the catalogues
from the coadded images. The effect on very large objects is that
part of the extended light of the object is assigned to the
background and removed from the galaxy, with the corresponding
effect on their total magnitude.
The Swarp background parameters used are BACK_SIZE=1024 and BACKFILTERSIZE=5, while SExtractor's background parameters used to create the final catalogues from the coadds are BACK_SIZE=512 and BACK_FILTERSIZE=3.
- Inappropriate deblending. Large galaxies clearly resolved or with bright knots are usually split in spurious different objects by the deblending procedure of SExtractor given the configuration parameters that we have used which, however, are suitable for most of the objects (DEBLEND_NTHRESH=32, DEBLEND_MINCONT=0.005).