星系物理参数测量方法研究及LAMOST星系增值星表的构建

Galaxy spectra contain plentiful information about the properties of galaxies. Determining the parameters of galaxies is crucial for investigating galaxy formation and evolution. In this project, we will present a full spectral fitting method for deriving the stellar population properties and parameters related to the gas, including age, metallicity, dust, stellar mass and nebular emission features, chemical abundance of gas. And this method will be used to measure the properties of LAMOST galaxies. We focus on the calibration, comparison and evaluation of measurement results, and then construct the LAMOST galaxy value-added catalogue. This method is a chi-squared minimization fitting code that fits spectroscopic data to combinations of single-burst stellar population models. We will derive a set of models with a finer metallicity grid. In the beginning of fitting, the continua of spectra and models are “filtered” in order to eliminate the uncertainty of the shape of continua of spectra caused by the uncertainty of the flux calibration or extinction. Moreover, the fitting process uses an iterative loop to progressively increase the models and adjusts the weights of the linear combination. The final “best-fitting” combination may be a set of solutions rather than a solution, which can explore the spectral degeneracies between the stellar population properties. The results of the project will not only provide the parameter measurements of galaxies with full spectral fitting based on line features, but also give some valuable code packages for the improvement of LAMOST pipeline, and further improve the scientific output of LAMOST galaxy spectra.

星系光谱中包含大量与其物理性质相关的信息，星系参数测量对我们研究星系的形成和演化至关重要。本项目拟建立一种全谱匹配方法以确定星系的星族成分和气体相关参数，包括年龄、恒星金属丰度、消光、质量和发射线强度、气体金属丰度等，并将此方法用于LAMOST星系的参数测量，着重研究测量结果的定标、比较和评价，构建LAMOST星系的增值星表。该方法拟构建一套金属丰度网格细化的星族模板库，将光谱与模板进行卡方最小化拟合，在拟合前先将待测光谱与模板的连续谱“拉平”，以消除流量定标或消光不确定而带来的连续谱形状不准确的问题。另外，拟合过程采用迭代法逐步增加模板谱，调整线性组合的权重，寻找一组解而不是一个解，以探索星族成分的简并问题。其研究结果有助于深入研究基于谱线特征的全谱匹配参数测量方法，同时也为优化LAMOST星系光谱pipeline提供重要算法，从而进一步提高LAMOST星系光谱的科学产出。

星系的光谱包含其内部恒星的年龄和金属丰度等信息，从观测光谱数据中测量这些信息对于深入了解星系的形成和演化至关重要。LAMOST巡天发布了大量的星系光谱，本项目提出了两种方法测量LAMOST星系光谱的年龄和金属丰度，以消除流量定标或消光不确定而带来的连续谱形状不准确的问题。一种是基于小尺度特征的模板匹配方法对LAMOST星系估计平均年龄和金属丰度，即使用全谱范围内的谱线特征来估计星族参数，该方法使得LAMOST星系参数测量不受连续谱的影响。另一种是直接使用LAMOST星系光谱的全谱信息进行有监督的深度学习，该方法利用深度学习处理多维、海量的非线性数据的优势，构建卷积神经网络模型，实现LAMOST星系年龄和金属丰度的自动测量，不需要对光谱连续谱进行处理。实验结果表明，两种方法得到的星族参数与传统方法得到的参数值基本一致，误差均在0.2dex以内，并且随着光谱信噪比的增大，预测误差越来越小。本项目发布了LAMOST星系星族参数的增值星表，这是LAMOST星系年龄和金属丰度星表的首次发布，有利于提高LAMOST星系数据的利用率。并且提供LAMOST星系物理参数自动测量算法软件包，为优化星系光谱pipeline提供算法参考。