Using APOGEE Stellar Abundances

This page has important information on the DR16 APOGEE stellar abundances. We recommend that users interested in using these data review this page. A proper understanding of how the quality of the supplied abundances is flagged using bitmasks is useful to sort out spectra/stars that likely are inaccurately analyzed by the pipeline. If you, after exploration, find that you would like to use the DR16 stellar abundances in a publication, also take the time to read Jönsson et al. (2020), which provided even more information and quality assessment.

One main objective of the APOGEE survey is to extract the chemical abundances of multiple elements for the entire stellar sample. In DR16, we attempt to measure abundances for 26 species: C, C I, N, O, Na, Mg, Al, Si, P, S, K, Ca, Ti, Ti II, V, Cr, Mn, Fe, Co, Ni, Cu, Ge, Rb, Ce, Nd, and Yb. The accuracy of an individual element varies with the element and stellar type; certain parts of the element-star parameter space have even been deemed impossible to explore with the APOGEE data, as is discussed below.

ASPCAP uses the same eight-dimensional grid of synthetic stellar spectra as was used for determining the stellar parameters when determining the stellar abundances. Now, however, the fit is done only for spectral windows that cover spectral lines corresponding to the element of interest. Variation is only made for the appropriate abundance dimension -- [M/H], [α/M], [C/M] or [N/M] -- depending on the element. Each element window has an element-specific weight that serves to minimize the impact of blended lines. As a result, some spectral lines are more important to the abundance determination than others. For elements with few lines, the abundance may be measured from a single line.

Data Access

All of the ASPCAP-derived stellar parameters and abundances are stored in the master allStar file with contents described in the allStar datamodel. Entries in this allStar file are frequently referred to on this page.

For users with only interest in part of the data, the ASPCAP output for all stars in a given field (i.e. location in the sky) is stored in a single aspcapField file. Results for each individual star are stored in aspcapStar files .

See the Data Access for a full description of files provided for the user.

Uncalibrated Abundances

The abundances for all 26 elemental species in all stars as determined by the pipeline is stored in the FELEM array (see allStar datamodel for full descriptions of these arrays). The order of the elements in this array is given in the ELEM_SYMBOL array. Since different elements are derived varying different grid dimensions -- [M/H], [α /M], [C/M] or [N/M] -- some of these abundances are given relative to hydrogen ([X/H]) and some relative to metallicity ([X/M]). The nature of every abundance ratio in the FELEM array is given in the ELEM_VALUE array. The abundances given in the FELEM array have not been quality-assessed but are mainly supplied for completeness and might be used by some users for exploratory purposes.

Calibrated Abundances

The original FERRE-derived abundances supplied in the FELEM array are refined and calibrated in two steps that are described below.
First, the abundances are zero-point shifted such that the median of the elemental abundance in stars near solar metallicity (|[M/H]|<0.05) in the solar neighborhood (within 0.5 kpc in R and Z) is adjusted to be zero ([X/M] = 0). This adjustment places the abundances on a "solar-like" scale; we do not calibrate directly to the Sun because it is not typical of the stars in the APOGEE sample. These abundances are supplied in the X_H, and X_M arrays relative to hydrogen and metallicity, respectively.

Secondly, we supply abundances in the so-called “named” tags (e.g., C_FE, O_FE, etc.), which are measured relative to iron. More importantly, we do not populate these tags if we have a reason or another to think the element or stellar type reliable (see Jönsson et al. (2020) for details). Thus, the named tags are the most conservative sets of abundances and recommended for most users. However, it is important to note that there still certainly are abundances in the named tags that may suffer from systematic errors, in particular for cooler stars.

Abundances Uncertainties

Empirical uncertainties for the abundances are derived from the scatter in abundances from stars in the sample that happen to be observed in two or more different fields and hence independently analyzed more than once by ASPCAP. These uncertainties are parameterized as a function of Teff, [M/H], and S/N, with separate relations for giants and dwarfs. These are included in the X_H_ERR and X_M_ERR arrays, as well as in the named tags, C_FE_ERR, N_FE_ERR, MG_FE_ERR, etc. Our current methodology does not naturally provide upper limits for non-detected lines.

APOGEE Abundance Scale

The abundance scale of DR16, i.e., which solar abundances are used in the normalization of the supplied abundance ratios, is difficult to detangle, but we are likely to be close to the scale of Grevesse et al. (2007) in most elements. The relevant steps in making this a hard question are described below:

  • When constructing the linelist for the analysis, we adjust the atomic data to fit a spectrum of the sun with the Grevesse et al. (2007) abundances and the parameters Teff=5777 K, log g=4.44, [Fe/H]=0.00, vmicro=1.10 km/s. However, we simultaneously adjust the atomic data to also fit a spectrum of Arcturus with the parameters Teff=4286 K, log g=1.66, [Fe/H]=-0.52, vmicro=1.74 km/s, and abundances from the literature, mainly from Ramirez & Allende Prieto (2011). Molecular data is not adjusted. More information about this fitting-procedure can be found in Smith et al. (2019).
  • We use the abundances of Grevesse et al. (2007) when calculating the stellar atmospheres used in the analysis.
  • We use the abundance scale of Grevesse et al. (2007) when calculating the synthetic stellar spectra for the spectral grids.
  • The supplied calibrated abundances have been zero-pointed so that solar-metallicity stars in the solar neighborhood have [X/M]=0. We do not calibrate directly to the Sun because it is not typical of the stars in the APOGEE sample. C and N abundances have not been calibrated for giants since those abundances are expected to be affected by the star’s evolution and not follow Galactic chemical evolution.

This means that the uncalibrated abundances derived for giants from molecular lines -- C, N, O (not C I) -- are not adjusted in any way and, given the molecular data do not have systematic uncertainties, those abundances should be on the Grevesse et al. (2007) scale.

Regarding the uncalibrated abundances derived from atomic lines, the abundance scale varies from element to element. Those elements with strong features in the Sun do not depend much on the fitting of the Arcturus spectrum/abundances when adjusting the atomic data. If these same features happen to have a high weight in the ASPCAP analysis, the abundance scale should be close to that of Grevesse et al. (2007). For elements whose determination relies more on lines whose data was more adjusted using the Arcturus spectrum, the absolute abundance scale is unknown. This sliding scale between the elements is the main reason for providing calibrated stellar abundances.

For the calibrated abundances, C and N in giants (not C I) are still unaltered and should -- if we assume that the molecular data used does not have any systematics -- be on the Grevesse et al. (2007)-scale. For all other abundances, the idea is that they should be on a “true” bracket-scale in the spectroscopic sense, where abundances are presented in a ratio to our own, undetermined, unspecified solar abundance.

ASPCAP bitmasks

Information about potential issues with the ASPCAP parameters is stored as a set of bitmasks. The ASPCAPFLAG bitmask is used to flag potential issues with the star and/or with specific parameters for that star. In addition, there is a separate bitmask for each element that flags possible conditions for that element. In the allStar file, these element bitmasks are stored in a ELEMFLAG bitmask array. Users need to be sure to consult the bitmasks and employ the derived elements accordingly. For stellar parameters, there is an additional PARAMFLAG that should be consulted as described in Using APOGEE stellar parameters.