I present the results oféchelle spectroscopy of a bright H II region in the irregular galaxy IC 4662 and their comparison with results from long-slit spectroscopy of the same region.All observations were obtained...I present the results oféchelle spectroscopy of a bright H II region in the irregular galaxy IC 4662 and their comparison with results from long-slit spectroscopy of the same region.All observations were obtained with the standard spectrographs of the Southern African Large Telescope:(1)low and medium spectral resolution spectrograph Robert Stobie Spectrograph(R≈800)and(2)échelle spectrograph HRS(R=16,000–1,7000).In both types of data the intensities of most of the emission lines were measured and abundances of oxygen and N Ne,S,Ar,Cl and Fe were determined as well as physical parameters of the H II region.The chemical abundances were obtained from both types of data with the Te-method.Abundances calculated from both types of data agree to within the cited uncertainties.The analysis of theéchelle data revealed three distinct kinematic subsystems within the studied H II region:a narrow component(NC,σ≈12 km s^(-1)),a broad component(BC,σ≈40 km s^(-1)),and a very broad component(VBC,σ≈60–110 km s^(-1),detected only in the brightest emission lines).The elementa abundances for the NC and BC subsystems were determined using the Te-method.The velocity dispersion dependence on the ionization potential of elements showed no correlation for the NC,indicating a well-mixed turbulent medium,while the BC exhibited pronounced stratification,characteristic of an expanding shell.Based on a detailed analysis of the kinematics and chemical composition,it was concluded that the BC is associated with the region surrounding a Wolf-Rayet(WR)star of spectral type WN7-8.The stellar wind from this WR star interacts with a shell ejected during an earlier evolutionary stage(either as a red supergiant or a luminous blue variable LBV),which is enriched in nitrogen.These findings highlight the importance of high spectral resolution for detecting small-scale(25 pc)chemical inhomogeneities and for understanding the feedback mechanisms of massive stars in low-metallicity environments.展开更多
天体光谱处理中的一项基本任务是对大量的恒星光谱进行自动分类。到目前为止,恒星光谱的分类工作多是基于一维光谱数据。该研究打破传统的天体光谱数据处理流程,提出了基于二维恒星光谱分类的方法。在LAMOST(the large sky area multi-o...天体光谱处理中的一项基本任务是对大量的恒星光谱进行自动分类。到目前为止,恒星光谱的分类工作多是基于一维光谱数据。该研究打破传统的天体光谱数据处理流程,提出了基于二维恒星光谱分类的方法。在LAMOST(the large sky area multi-object fiber spectroscopic telescope)的数据处理流程中,所有的一维光谱都是由二维光谱抽谱、合并得来。二维光谱是由光谱仪产生的图像,包括蓝端图像和红端图像。基于LAMOST二维光谱数据,提出了特征融合卷积神经网络(FFCNN)分类模型,用于二维恒星光谱的分类。该模型是一个有监督的算法,通过两个CNN模型分别提取蓝端图像和红端图像的特征,然后将二者进行融合得到新的特征,再利用CNN对新特征进行分类。所使用的数据全部来源于LAMOST,我们在LMOST DR7中随机选择了一批源,然后获得了它们的二维光谱。一共有14840根F,G和K型恒星的二维光谱用于FFCNN模型的训练,其中包括7420根蓝端光谱和7420根红端光谱。由于三类恒星光谱的数量并不均衡,在训练的过程中分别为每类恒星光谱设置了不同权重,防止模型出现分类失衡现象。同时,为了加快模型收敛,对二维光谱数据采用Z-score归一化处理。此外,为了充分利用所有样本,提高模型的可靠度,采用五折交叉验证的方法验证模型。3710根二维光谱用作测试集,使用准确率、精确率、召回率和F1-score来对FFCNN模型的性能进行评价。实验结果显示,F,G和K型恒星的精确率分别达到87.6%,79.2%和88.5%,而且它们超过了一维光谱分类的结果。实验结果证明基于FFCNN的二维恒星光谱分类是一种有效的方法,它也为恒星光谱的处理提供了新的思路和方法。展开更多
基金support from the National Research Foundation(NRF)of South Africa。
文摘I present the results oféchelle spectroscopy of a bright H II region in the irregular galaxy IC 4662 and their comparison with results from long-slit spectroscopy of the same region.All observations were obtained with the standard spectrographs of the Southern African Large Telescope:(1)low and medium spectral resolution spectrograph Robert Stobie Spectrograph(R≈800)and(2)échelle spectrograph HRS(R=16,000–1,7000).In both types of data the intensities of most of the emission lines were measured and abundances of oxygen and N Ne,S,Ar,Cl and Fe were determined as well as physical parameters of the H II region.The chemical abundances were obtained from both types of data with the Te-method.Abundances calculated from both types of data agree to within the cited uncertainties.The analysis of theéchelle data revealed three distinct kinematic subsystems within the studied H II region:a narrow component(NC,σ≈12 km s^(-1)),a broad component(BC,σ≈40 km s^(-1)),and a very broad component(VBC,σ≈60–110 km s^(-1),detected only in the brightest emission lines).The elementa abundances for the NC and BC subsystems were determined using the Te-method.The velocity dispersion dependence on the ionization potential of elements showed no correlation for the NC,indicating a well-mixed turbulent medium,while the BC exhibited pronounced stratification,characteristic of an expanding shell.Based on a detailed analysis of the kinematics and chemical composition,it was concluded that the BC is associated with the region surrounding a Wolf-Rayet(WR)star of spectral type WN7-8.The stellar wind from this WR star interacts with a shell ejected during an earlier evolutionary stage(either as a red supergiant or a luminous blue variable LBV),which is enriched in nitrogen.These findings highlight the importance of high spectral resolution for detecting small-scale(25 pc)chemical inhomogeneities and for understanding the feedback mechanisms of massive stars in low-metallicity environments.
文摘天体光谱处理中的一项基本任务是对大量的恒星光谱进行自动分类。到目前为止,恒星光谱的分类工作多是基于一维光谱数据。该研究打破传统的天体光谱数据处理流程,提出了基于二维恒星光谱分类的方法。在LAMOST(the large sky area multi-object fiber spectroscopic telescope)的数据处理流程中,所有的一维光谱都是由二维光谱抽谱、合并得来。二维光谱是由光谱仪产生的图像,包括蓝端图像和红端图像。基于LAMOST二维光谱数据,提出了特征融合卷积神经网络(FFCNN)分类模型,用于二维恒星光谱的分类。该模型是一个有监督的算法,通过两个CNN模型分别提取蓝端图像和红端图像的特征,然后将二者进行融合得到新的特征,再利用CNN对新特征进行分类。所使用的数据全部来源于LAMOST,我们在LMOST DR7中随机选择了一批源,然后获得了它们的二维光谱。一共有14840根F,G和K型恒星的二维光谱用于FFCNN模型的训练,其中包括7420根蓝端光谱和7420根红端光谱。由于三类恒星光谱的数量并不均衡,在训练的过程中分别为每类恒星光谱设置了不同权重,防止模型出现分类失衡现象。同时,为了加快模型收敛,对二维光谱数据采用Z-score归一化处理。此外,为了充分利用所有样本,提高模型的可靠度,采用五折交叉验证的方法验证模型。3710根二维光谱用作测试集,使用准确率、精确率、召回率和F1-score来对FFCNN模型的性能进行评价。实验结果显示,F,G和K型恒星的精确率分别达到87.6%,79.2%和88.5%,而且它们超过了一维光谱分类的结果。实验结果证明基于FFCNN的二维恒星光谱分类是一种有效的方法,它也为恒星光谱的处理提供了新的思路和方法。
