Estimation Of [α /Fe] Abundance Ratios From Low-resolution LAMOST Stellar Spectra

We present a method for determining [α/Fe] ratios from low-resolution(R =1800) LAMOST stellar spectra. The method is an extension of the LSP3(LAMOST Stellar Parameter Pipeline at Peking University) which is based on a template-matching technique, and it estimates the [α/Fe] values by minimizing χ2 between the observed spectrum and the template spectrum taking the [α/Fe] as a free parameter. The grid of synthetic spectra covers 4000 K ≤ Teff ≤ 8000 K in steps of 100 K, 0.0 ≤ log g ≤ 5.0 in steps of 0.25 dex,-4.0 ≤ [Fe/H] ≤- 1.0 in steps of 0.2 dex, and-1.0 ≤ [Fe/H] ≤ +0.5 in steps of 0.1 dex. The range of [α/Fe] values is from-0.4 dex to +1.0 dex with step of 0.2 dex. To determine the overall behavior for the four individual α-elements Mg, Si, Ca, and Ti, we choose spectral bands in the range of 440 – 460 nm and 500 – 530 nm because it contains the prominent lines of Mg I, Ti I and Ti II, as well as Ca I and Si II, and their features are very sensitive to the abundance of [α/Fe].In order to validate the method in use of determining the [α/Fe], it is applied to two stellar samples from the spectra libraries ELODIE and MILES. By comparing the measurements of LSP3 with the orresponding results derived from high-resolution spectra analysis, we find that our estimations of [α/Fe] are slightly lower than those derived from high-resolution spectra with an offset of-0.13 dex and a standard deviation of 0.08 dex. The star-to-star errors of [α/Fe] between LSP3 and HR appears to be little or no covariance in the determined [α/Fe] along with various ranges for the individual stellar parameters. In addition, we also used a sample of stars belong to two globular clusters and two open clusters selected from SDSS survey as an independent validation of our method, the results show that our measurements of [α/Fe] are consistent with the values derived from high-resolution spectra.In order to estimates the uncertainties of our measurements of [α/Fe], we use a sample of 77 stars which have spectra both from LAMOST survey and high-resolution observations, which can be served as a calibration sample to determine and correct the system deviation. We use about 90,000 stars which has multi-epoch duplicate LAMOST spectra to estimates the random errors of our [α/Fe] measurements. From those testing results, we conclude that our method is validating for estimating the [α/Fe] values from LAMOST spectra with S/N higher than 30. For stars with [Fe/H] >-1.0, [α/Fe] can be determined with a precision of 0.1 dex, while for star with [Fe/H] <-1.0, the precision will drop to about 0.14 dex.