基于LAMOST光谱的恒星大气物理参数估计

发布时间：2019-01-08 14:56

【摘要】：基于实测光谱的恒星大气物理参数估计是探索恒星本质的首要任务。随着郭守敬望远镜(LAMOST)进入正式巡天阶段,正以前所未有的速度获取海量的恒星实测光谱数据,这为星系研究带来了新的机遇和挑战。由于LAMOST是多目标光纤光谱天文望远镜,获取的光谱噪音比较大。光谱前期处理中的波长定标和流量定标精度不高,导致光谱存在微小畸变,这些都大大增加了恒星大气物理参数测量的难度。如何对LAMOST实测光谱的恒星大气物理参数进行自动测量是迫切期待需要研究的一个重要课题,关键是如何消除噪声,提高恒星大气物理参数的测量精度和鲁棒性。提出了一个测量LAMOST恒星光谱大气参数的回归模型(SVM(lasso))。基本思路是:首先使用Haar小波对光谱信号进行滤波,抑制光谱中噪声的不利影响,最大限度地保留光谱判别信息。然后采用lasso算法进行特征选择,选取与恒星大气物理参数相关性强的特征。最后将选择的光谱特征输入支持向量机回归模型对恒星大气物理参数进行估计,该模型对光谱畸变和噪音的容忍性比较好,提高了测量的精确度。为了验证上述方案的可行性,在33 963条LAMOST先导巡天恒星光谱库上作了实验研究,三个恒星大气物理参数的精度分别为log Teff:0.006 8dex,log g:0.1551dex,[Fe/H]:0.104 0dex。
[Abstract]:The estimation of stellar atmospheric physical parameters based on measured spectra is the primary task in exploring the nature of stars. As Guo Shoujing Telescope (LAMOST) enters the official phase of sky survey, it is acquiring massive stellar measured spectral data at an unprecedented speed, which brings new opportunities and challenges to galaxy research. Because LAMOST is a multi-object optical fiber spectral telescope, the spectral noise obtained is relatively large. The precision of wavelength calibration and flow calibration in the pre-processing of spectrum is not high, which leads to the small distortion of the spectrum, which greatly increases the difficulty of measuring the atmospheric physical parameters of stars. How to automatically measure the atmospheric physical parameters of stars measured by LAMOST spectrum is an important subject that needs to be studied urgently. The key is how to eliminate noise and improve the measurement accuracy and robustness of stellar atmospheric physical parameters. A regression model (SVM (lasso). For measuring spectral atmospheric parameters of LAMOST stars is proposed. The basic idea is: firstly, the spectral signal is filtered by Haar wavelet to suppress the adverse effect of the noise in the spectrum, and the spectral discriminant information is kept to the maximum extent. Then the lasso algorithm is used to select the features which have strong correlation with the atmospheric physical parameters of stars. Finally, the selected spectral features are input into the support vector machine regression model to estimate the atmospheric physical parameters of stars. The model has better tolerance to spectral distortion and noise, and improves the accuracy of measurement. In order to verify the feasibility of the above scheme, an experimental study has been carried out on 33,963 LAMOST guided star spectropools. The accuracy of the three stellar atmospheric physical parameters is log Teff:0.006 8 dextral log g: 0.1551 dextran, [Fe/H]: 0.104 dextra.
【作者单位】：华南师范大学数学科学学院;
【基金】：国家自然科学基金项目(61273248,61075033) 广东省自然科学基金项目(S2011010003348)资助
【分类号】：P144

【参考文献】