LAMOST恒星光谱物理参数测量软件系统的开发

Galactic astronomy mainly aims to study the structure, formation and evolution of the Galaxy through the physical properties of its member stars. Thus large spectroscopic survey data and automated methods to precisely derive the fundamental stellar atmospheric parameters (Teff, log g, [Fe/H], RV) are in need. Statistical analysis of the large sample of these data could reveal more fine characteristics of the Galaxy. ..LAMOST survey project executed its pilot survey and formal five-year regular survey since Sep. 2012 and Oct. 2011, respectively. Now it already observed more than two million stellar spectra. The LAMOST stellar parameter pipeline (LASP） which exploits ULySS method, was used to automatically determine the stellar parameters. The improvement, validation and calibration of the LASP is very significant and crucial for the subsequent survey implementation and the realization of the scientific objectives. ..While ULySS took ELODIE library as a reference, in this work we alternatively change to use MILES library which has better stellar parameter space coverage. Based on this, a new interpolator is built, which is a function that could return a model spectrum for a given set of atmospheric parameters. This new interpolator performs better than the former one for the parameter estimation, and expands the valid estimation region to more cold and hot stars and the place located around the edge of the parameter space. Through this way we could improve the reliability and precision on the parameter determination for the LASP...The data processing phase of the survey mission normally comprises three vital parts: validation, calibration and in-mission software development. By selecting and collecting proper comparing data sample and parameter set, e.g. LAMOST repeated observed spectra, other survey (medium or low resolution) spectra (like SDSS etc.) and given parameters, high resolution spectra sample with well determined parameters, standard clusters’ member stars with well known physical property etc. By comparing with the LASP determined results, systematical and comprehensive statistical validation tests and analysis could be made. Thus according to different stellar types, SNR, magnitude, observation condition etc., the credibility and accuracy of each physical parameter estimation could be derived, as well as the distribution feature of the individual errors. The potential existing systemic bias for the determined parameters could be identified and corrected by employing effective calibration function. Conversely and iteratively, the results of the substantial comparison and validation, may well guide the improvement of the LASP (data processing way, algorithm, software framework etc.). Besides, it may also help timely adjusting and optimizing the selection strategy for which kind of LAMOST observed stellar spectra will be determined, in order to enhance the effective output rate for LAMOST produced stellar parameters.

对大样本恒星精确的物理特征量的统计分析，有助于更加全面和真实地了解和认识银河系形成、结构和演化的规律和特征。LAMOST巡天项目开展以来，已观测了200多万条恒星光谱。LAMOST恒星光谱物理参数自动测量软件系统的改进和评估对项目的顺利实施和科学目标的实现具有非常重要的意义。本项目一方面通过改用基于更好的恒星模板光谱库训练构建的模型谱生成函数，提升参数估计的可靠性和精度。另一方面，通过选取不同来源的参照数据集（来自其他巡天项目的光谱集、高分辨率光谱集、标准星团等），对该系统给出的参数做细致全面的统计检验分析，按不同光谱型、信噪比、观测条件等，给出各物理参数估计的精度范围和误差分布特征，对可能存在的系统偏差做定标校准等。依据检验分析结果，可针对性地改进系统的处理流程、算法性能，适时调整和优化系统软件对待测巡天恒星光谱的整体选择规则，提高LAMOST恒星物理参数有效产出率。

LAMOST一期巡天任务（2012.9-2.017.6）观测获取了海量恒星光谱数据，LAMOST DR5 里信噪比大于10的恒星光谱有7531398条。通过对大样本恒星物理参数做统计分析，来研究银河系的结构和动力学等方面的课题，具有重要的意义。如何对LAMOST海量恒星光谱做自动处理并测得可靠的物理参数是迫切需要解决的关键问题。本项目研究内容为通过研发有效的算法和软件系统实现这一需求。..我们采用ELODIE、MILES及理论恒星光谱库作为模板库，训练生成了非线性内插函数模型，用于计算生成任意一组给定参数对应的光谱，经过优选模板和多次训练改进模型函数，通过待测谱和模型谱间的卡方拟合来实现参数的估计。LAMOST恒星光谱物理参数自动测量软件系统（LASP）的性能、可靠性和精度得到了提升，对热星和冷星的估计性能也得到了显著提高，LASP能够对温度在2700-40000K范围内的恒星进行参数测量。对LAMOST恒星光谱测得的参数，我们做了大量的统计分析，通过不同参照对比数据集（如 其他星表、巡天数据集、高分辨率光谱、其他软件算法得出的结果等），做了一系列内部和外部定标检验，对参数估计误差及分布特征有了了解和掌握。通过对比还发现了某些参数在局部空间里有些估计偏差，我们通过选用更精准的参数集针对该问题进行了标校改正，进一步提高了参数估计的可靠性。..本项目实现了LASP的研发和运维及对应后端数据库的设计和使用，对Teff、log g、[Fe/H]和RV四个参数的测量精度分别达到：130 K，0.22 dex，0.12 dex和5 km/s，坚实地保障了每次LAMOST官方发布AFGK恒星参数星表产品。2018年初DR5发布的参数星表含5344058条结果（产出率71%），是目前世界上同类产品中最大的，此星表对人们更加深入全面地了解银河系具有重要价值。基于LASP测得LAMOST恒星参数发表的SCI文章已逾60篇。针对LAMOST二期光谱巡天（2017年9月始），已对LASP做调整将其应用于中分辨率恒星光谱测量上，现处于测试运行阶段，有待完善。