Individual Methods Used in SSPP
The SSPP implements a number of different methods for determining stellar parameters. The following table shows the range of g-r color, signal-to-noise ratio (S/N), and wavelength range used for each method in the SSPP. The details on each method and the meaning of the name of each technique can be found in Lee et al. (2008a), except for T9 (IRFM), which
is presented in Rockosi et al. in prep.
| Teff | log g | [Fe/H] | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Name | Method | g-r | Name | Method | g-r | Name | Method | g-r | S/N | Wavelength range (Å) | Method |
| T1 | ki13 | 0.0 – 0.6 | G1 | ki13 | 0.0 -0.6 | M1 | ki13 | 0.0 – 0.6 | ≥20 | 4400 – 5500 | Spectral matching |
| T2 | k24 | 0.0 – 0.6 | G2 | k24 | 0.0 – 0.6 | M2 | k24 | 0.0 – 0.6 | ≥15 | 4400 – 5500 | Spectral matching |
| T3 | WBG | -0.3 – 0.3 | G3 | WBG | -0.3 – 0.3 | M3 | WBG | -0.3 – 0.3 | ≥10 | 3900 – 6000 | Spectral Matching |
| T4 | ANNSR | -0.3 – 0.7 | G4 | ANNSR | -0.3 – 0.7 | M4 | ANNSR | -0.3 – 0.7 | ≥15 | 3850 – 9000 | Spectral Matching |
| T5 | ANNRR | -0.3 – 1.2 | G5 | ANNRR | -0.3 – 1.2 | M5 | ANNRR | -0.3 – 1.2 | ≥10 | 3850 – 9000 | Spectral Matching |
| T6 | NGS1 | -0.3 – 1.3 | G6 | NGS1 | -0.3 – 1.3 | M6 | NGS1 | -0.3 – 1.3 | ≥10 | 4500 – 5500 | Spectral Matching |
| G7 | NGS2 | 0.0 – 1.3 | M7 | NGS2 | 0.0 – 1.3 | ≥20 | 4500 – 5500 | Spectral Matching | |||
| G8 | CaI1 | 0.1 – 1.2 | M8 | CaIIK1 | -0.1 – 1.2 | ≥15 | 3850 – 4250 | Spectral Matching | |||
| M9 | CaIIK2 | 0.1 – 0.7 | ≥15 | ~ 3933 | Line index | ||||||
| M10 | CaIIK3 | 0.1 – 0.7 | ≥15 | ~ 3933 | Line index | ||||||
| T7 | HA24 | 0.1 – 0.8 | ≥10 | ~ 6563 | Line index | ||||||
| T8 | HD24 | 0.1 – 0.6 | ≥10 | ~ 4102 | Line index | ||||||
| T9 | IRFM | -0.3 – 1.3 | ≥10 | N/A | Color-temperature | ||||||
The ki13, k24, NGCS1, and NGS2 are grid-matching based methods. WBGis the method from Wilhelm, Beers, & Gray (1999). ANNSR and ANNRR are the neural network approaches. CaIIK1 and CaI1 are determined from the 3850 – 4250 Å region. CaIIK2 and CaIIK3 are [Fe/H] estimates based on the Ca II K line. S/N is the average signal to noise ratio per pixel over the spectral region 4000 – 8000 Å. HA24 and HD24 are the temperature estimates from the Hα and Hδ line indices in 24 Å widths, respectively. See Section 4 in Lee et al. (2008a) for detailed descriptions of individual methods. Normally, estimators that satisfy the criteria are used in the final average.
The SSPP now adopts a much improved color (g-i)-temperature relation, the InfraRed Flux Method (IRFM). This relation is derived from about 14,000 stars, having both SDSS (u, g, r, i, z) and near infrared (J, H, and K) photometry, as well as surface gravity and metallicity measurements from the SSPP. Because the IRFM relation depends on the metallicity and surface gravity of the stars in question, an iterative procedure is used for the IRFM temperature estimate. The metallicity and gravity determined by NGS1 are used in this equation to obtain the first guess for effective temperature. With this first pass of temperature held fixed during the χ2 minimization in the NGS1 synthetic spectra grid, a new set of log g and [Fe/H] is obtained. These log g and [Fe/H] are again plugged into the derived color-temperature relation to determine the final estimate of Teff. This IRFM temperature estimate is considered in averaging to obtain the adopted Teff for the range of -0.3 ≤ g-r ≤ 1.3 and S/N ≥ 10.
