SDSS-III APOGEE Visit Combination

Introduction

Most APOGEE fields are observed multiple times to obtain our desired high signal to noise ratio for faint stars and to detect stellar binaries by their radial velocity oscillations. These multiple "visit" spectra need to be combined into one spectrum for each star. This process is called "visit combination" and involves several steps:

Determination of doppler shift (i.e. radial velocity) for each visit spectrum.
Resampling of each visit spectrum onto the same rest (i.e. no doppler shift) wavelength scale.
Continuum normalization of each visit spectrum using a median filter.
Weighted combination of all the resampled visit spectra.
Re-application of mean continuum shape.

The output of this process is a single apStar file for each star. This file contains the combined spectrum and all the resampled visit spectra on the same wavelength scale.

Radial Velocity Determination

The radial velocities (RV) in visit combination are derived in two separate steps:

Relative radial velocities using the combined spectrum as the spectral template. This is done iteratively.
Absolute radial velocity determination of the combined spectrum against a grid of synthetic spectra spanning a large range of stellar parameters.

The visit relative radial velocities and the absolute velocity of the combined spectrum are then combined to produce absolute velocities for all visit spectra.

This two-step scheme was employed because RVs derived from the combined spectrum (of the star itself) should be more precise than RVs derived from a small set of synthetic spectra. It allows us to create a high-quality combined spectrum without even knowing what type of object we are dealing with. However, the final combined spectrum must be on the rest wavelength scale so that it can be properly compared to the large grid of sythetic spectra in the abundance pipeline (ASPCAP). Therefore, the second step in the RV determination is to derive the absolute radial velocity of the combined spectrum against a small grid synthetic spectra (the "RV mini-grid").

The procedure for determining radial velocities is further described here.

Resampling

In order to combine the visit spectra their individual doppler shifts must be removed and then they must be resampled onto the same wavelength scale. Using the radial velocity determined in step (1) the wavelengths are corrected to the rest wavelengths (λ_rest=λ_obs/(1+RV/c), where c is the speed of light). Using the de-doppler-shifted wavelengths the spectrum is resampled onto the final logarithmically-spaced wavelength scale using sinc interpolation.

Continuum Normalization

Since the visit spectra are taken under different conditions the absolute fluxes can vary from visit to visit. Therefore, each visit spectrum needs to be roughly continuum normalized before they are combined. A 500-pixel median filter (excluding bad pixels) is used to calculate the continuum and normalize the spectrum. This continuum is saved for a final re-normalizing step at the end.

Weighted Combination

The final step is to combine the rest-frame shifted, resampled and normalized visit spectra. The combination is done in two different ways: (1) global weighting, where each visit spectrum is weighted by its (S/N)², and (2) pixel-by-pixel weighting, where each pixel is weighted by its (S/N)². Finally, the combined spectrum is multiplied by the average (over the multiple visit spectra) of the continua found in step (3).

output star spectra: apStar files

As described in the data access description, the combined spectra are provided in apStar files in FITS format. The primary HDU of each file contains an image which gives two versions of the combined spectrum for the object, plus the individual visit spectra that went into the combination. For these files, all of the individual spectra have been resampled to a common logarithmically-spaced wavelength scale, with the radial velocity of each individual spectrum removed. Note that the APOGEE wavelength scale is based on vacuum wavelengths. The logarithmic wavelength grid spacing is the same for all objects (log₁₀ λ_i+1 - log₁₀ λ_i = 6E-6) with a common starting wavelength of 15100.802 Angstroms. These spectra are roughly flux-calibrated. Additional HDUs contain the estimated uncertanties in each pixel, masks, and other information.

HDU2 stores the uncertainties per pixel. It is set to a large number for pixels that should be ignored entirely due to, e.g., bad columns (another way of thinking about it is that they have infinite error). In the spectra shown above the errors per pixel are shown as the grey band surrounding the spectrum; for masked pixels the grey band covers the full vertical extent of the figure.

The pixel mask information is stored in HDU3. These images yield a bitmask for each pixel, in particular the APOGEEPIXMASK bitmask. Since the final spectrum is a combination of 3 or more individual exposures, it may be that some bits were flagged in some exposures but not in others. HDU3 include both an "and mask" and an "or mask" for the combined spectra.