Preamble

Vol. 66 No. 5

September 2025

ACTA ASTRONOMICA SINICA Vol. 66 No. 5 Sept., 2025 doi: 10.15940/j.cnki.0001-5245.2025.05.009

Morphological Analysis of Dense 13CO&C18O Structure in the Milky Way Imaging Scroll Painting (10°≤l≤20°)

XIAO Yan-shan¹,² ZHANG Hai-xia¹,² HUANG Yao¹,²† JIANG Zhi-bo³ CHEN Zhi-wei³ ZHENG Sheng¹,2
ZHANG Peng¹,² LUO Xiao-yu¹,² JIANG Yu³ PAN Xue-jiao¹,²

(1 Center for Astronomy and Space Sciences, China Three Gorges University, Yichang 443002)
(2 College of Mathematics and Physics, China Three Gorges University, Yichang 443002)
(3 Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing 210023)

Abstract

Dense molecular clumps are the sites of star formation, and their morphological characteristics are closely linked to the physical properties of the molecular gas. This study is based on data from the Purple Mountain Observatory 13.7 m millimeter-wave radio telescope, covering galactic longitudes (l) of 10°≤l≤20° and galactic latitudes (b) of -5°.25≤b≤5°.25. We conducted morphological classification and analysis of dense structures traced by 13CO(J=1−0) & C18O(J=1−0) emission lines. Using the FacetClumps detection algorithm, we detected and manually verified molecular cores traced by C18O spectral line data, obtaining a total of 544 C18O molecular cores. Approximately 5.97% of the 13CO clumps in this region contain C18O cores. We categorized the 13CO clumps based on the presence or absence of C18O cores and found that those containing C18O cores have significantly higher peak intensities, fluxes, and angular areas compared to those without C18O cores. However, no significant differences in eccentricity or form factor were found between the two categories. Furthermore, a comparative analysis of form factor shows that the velocity-integrated intensity contours of C18O cores are notably closer to circular shapes than those of 13CO clumps.

Key words stars: formation, ISM (interstellar medium): molecules, ISM: lines and bands, methods: data analysis

1 Introduction

The discovery and observation of interstellar molecules have profoundly impacted modern astronomy, establishing that stars form in molecular clouds. The complex hierarchical structure of molecular clouds can be subdivided into substructures including clouds, clumps, and cores. Based on extensive observational results, it is now believed that stars form in dense molecular clouds. Molecular clouds undergo quasi-static contraction to form filamentary structures, and under certain physical conditions, when these filaments become sufficiently dense, they fragment internally to produce molecular clumps of various sizes, shapes, and density structures. When the density and mass of a molecular clump reach critical values and can no longer resist gravity, collapse begins, leading to the formation of protostellar objects. Dense molecular clumps are the sites of star formation, and their morphological and physical characteristics determine the initial conditions for star formation, making them important research subjects.

In molecular clouds, the primary component is molecular hydrogen (H₂). Besides H₂, the most important and abundant molecule is carbon monoxide (CO). As the most abundant and widely used molecular probe, CO is employed to investigate the physical state, distribution, and kinematic properties of molecular clouds. Different CO isotopologues with varying optical depths serve as effective probes of different layers and environments in the interstellar medium. In particular, the completely optically thin C18O molecule provides opportunities to study smaller or denser regions that are typically closest to or about to undergo star formation.

To comprehensively understand the distribution of molecular gas in the Galactic plane, numerous CO molecular line surveys of Galactic structure have been conducted both domestically and internationally, such as the CO(J=1–0) survey by the Harvard-Smithsonian Center for Astrophysics, the Galactic Ring Survey (GRS) by the Five College Radio Astronomy Observatory, the 13CO/C18O Heterodyne Inner Milky Way Plane Survey (CHIMPS) by the Astrophysics Research Institute of Liverpool John Moores University, the Structure, Excitation and Dynamics of the Inner Galactic Interstellar Medium (SEDIGISM) survey by the Max Planck Institute for Radio Astronomy, and the Milky Way Imaging Scroll Painting (MWISP) survey by the Purple Mountain Observatory of the Chinese Academy of Sciences. These survey projects have yielded massive observational datasets, enabling astronomers to conduct numerous meaningful studies on the relationship between the properties of Galactic molecular gas and star formation.

Yuan et al. correlated cloud morphology, properties, and Galactic environment, classifying clouds into filamentary and non-filamentary structures. They found that approximately 10% of filamentary molecular clouds contribute about 90% of the total CO flux. Neralwar et al. first completed morphological classification of molecular cloud data from the SEDIGISM survey using the J-plots algorithm and visual inspection, finding that most clouds exhibit elongated structures. Subsequently, Clarke et al. updated the classification labels in the SEDIGISM catalog using an improved RJ-plots algorithm, demonstrating a strong correlation between the central concentration of molecular cloud structures and their star formation efficiency and dense gas fraction, but no significant correlation with filamentary morphology. These studies selected larger-scale CO molecular clouds as classification objects to analyze morphological characteristics and patterns. However, morphological classification studies of denser structures traced by other isotopologues (13CO & C18O) remain relatively limited.

In recent years, the rapid development of machine learning technology has enhanced the performance of detection algorithms for small-scale dense structures in molecular clouds. For example, the GaussClumps algorithm obtains clump regions through iterative three-dimensional Gaussian fitting at maximum peaks; the ClumpFind detection algorithm uses contour levels to assign pixels containing extremum points to that extremum as a dense structure; the FellWalker algorithm finds local maxima by following the direction of maximum gradient from low-intensity points, assigning all points on paths converging at the same peak location to the same structure; the Dendrograms algorithm displays the hierarchical structure of molecular spectral lines, identifying molecular clumps by constructing hierarchical structures. Additionally, new detection algorithms specifically designed for molecular clumps and cores have emerged. The Local Density Clustering algorithm uses local density clustering methods to determine clump centers, members, and boundaries; the ConBased algorithm divides signals into small regions and uses merging rules based on connectivity, peak distance, intensity differences, and volume to merge signals from the smallest regions; the FacetClumps algorithm applies the multivariate function extremum theorem to determine clustering centers, then combines connectivity and minimum distance to cluster local regions around centers for clump identification. These improved detection methods have produced extensive catalogs of dense molecular structures, providing data sources for studying the morphological characteristics of small-scale molecular clumps and cores.

The large-scale, unbiased, and high-sensitivity CO spectral line data from the MWISP survey provide an opportunity for systematic studies of the spatial distribution and morphological characteristics of molecular clumps. In this paper, we use the FacetClumps algorithm to detect and verify C18O molecular cores in the region with galactic longitudes of 10°≤l≤20° and galactic latitudes of -5°.25≤b≤5°.25. Combining this with the 13CO molecular clump catalog for this region, we complete the matching of 13CO clumps and C18O cores. Based on the matching results, we conduct classification studies and statistical analysis of the morphology of 13CO molecular clumps and explore morphological differences in gas structures traced by different density probes.

This paper is organized as follows: Section 2 describes the acquisition and matching of the 13CO molecular clump and C18O molecular core samples. Section 3 presents statistical analysis of parameters for two categories of 13CO clumps, using the presence or absence of matched C18O cores as the classification criterion. In Section 4, we analyze and discuss the morphological characteristics of 13CO clumps and C18O cores. Section 5 presents the conclusions of this paper.

2 Data Samples

2.1 Data Sources

The Milky Way Imaging Scroll Painting (MWISP) uses the Purple Mountain Observatory 13.7 m millimeter-wave telescope to simultaneously observe three transition lines: CO, 13CO, and C18O(J=1–0). It is an unbiased survey project with high sensitivity, wide velocity coverage, and high resolution. According to the observational strategy, the survey employs a sideband-separating superconducting spectroscopic array receiver system. This receiver uses double-sideband superconductor-insulator-superconductor mixers, incorporates a fast Fourier transform spectrometer, widely adopts digital technology, and combines On-The-Fly (OTF) observing mode, providing good operational stability and efficiency. CO is observed in the upper sideband with a main beam width of , while 13CO and C18O have a main beam width of , both with a bandwidth of 1000 MHz and 16384 channels. The velocity resolution is approximately 0.159 km·s⁻¹ for CO and 0.166 km·s⁻¹ for 13CO and C18O. The total sky coverage of the survey is , comprising 10,941 tiles, with each tile measuring . After baseline calibration of the spectral lines, three-dimensional data cubes of CO, 13CO, and C18O spectral lines are produced with a grid spacing of . The typical root mean square (rms) noise levels for CO, 13CO, and C18O spectral lines are 0.47 K, 0.22 K, and 0.21 K, respectively.

2.2 Data Samples

2.2.1 13CO Molecular Clumps

Luo et al. utilized 13CO(J=1–0) emission line data from MWISP and, following the Facet-SS-3D-Clump pipeline, completed the extraction and confidence assignment of molecular clumps in the region (cumulative coverage area of approximately 100 deg²). The FacetClumps algorithm obtains candidate molecular clumps, which are cross-matched with clumps detected by the Dendrograms algorithm to obtain high-confidence clumps. These high-confidence clumps are then used as prior knowledge to train a semi-supervised deep clustering model, SS-3D-Clump, and the trained model is subsequently applied to verify candidate clumps, providing confidence levels for them. The final catalog containing 18,757 13CO clumps is published at https://www.scidb.cn/en/s/qEfe2m. This region encompasses very active massive star-forming regions such as M16, M17, W31, W33, and W39, containing numerous samples of molecular clumps and cores at different evolutionary stages, making it a key region for understanding star evolution processes. The 13CO clumps used in this paper are selected from this catalog. [TABLE:1] shows the parameter information of molecular clumps used in the statistical analysis. The clump confidence (Confidence parameter in Table 1) and clump flux (Flux parameter) are used as source selection criteria.

The distribution of clump confidence is shown in [FIGURE:1], ranging from 0.8 to 1.0. To ensure the reliability of 13CO molecular clump samples, high-confidence samples must be selected. This paper chooses 13CO clump samples with confidence levels greater than or equal to 0.9, i.e., Confidence ≥ 0.9. The flux distribution is shown in [FIGURE:2]. According to the completeness test results in reference [29], when the flux is 37 K·km·s⁻¹, the clump recall rate is 90%. To ensure the completeness of the 13CO clump sample, this paper selects clumps with flux greater than 37 K·km·s⁻¹ as the research objects, i.e., Flux > 37 K·km·s⁻¹. When both selection conditions are satisfied simultaneously, we筛选出了7097个13CO分子云团块作为分析对象.

2.2.2 C18O Molecular Cores

Following the 13CO molecular clump extraction method described in reference [29], we employ the same FacetClumps algorithm to detect candidate C18O molecular cores in this region. The FacetClumps algorithm combines facet models with multivariate function extremum theorems to automatically locate source centers within preprocessed signal regions and uses a connectivity-based minimum-distance clustering method to merge local regions segmented by local gradients, thereby identifying regions corresponding to each source. This approach demonstrates high accuracy for detecting dense structures. [TABLE:2] presents the algorithm parameters for detecting C18O spectral lines.

The algorithm outputs both a source catalog and mask data. The source catalog provides relevant parameters for each source, including spatial coordinate indices in the mask data; the centroid positions in galactic longitude, latitude, and radial velocity; source sizes; peak intensities; and fluxes. In the mask, pixels belonging to the same source are labeled with the same integer, while unassigned pixels are labeled as 0. Based on the centroid position and size recorded in the source catalog, the original three-dimensional data cube for each source can be extracted from the input data.

Due to noise, the detection results from the FacetClumps algorithm still contain a certain number of false C18O targets. After obtaining the detection catalog, manual verification is required to ensure the reliability of C18O cores. For each C18O core candidate in the algorithm output catalog, we produced verification images showing integrated intensity maps and spectral line plots in the longitude, latitude, and radial velocity dimensions. Candidates satisfying both of the following criteria were identified as genuine C18O cores: (1) The integrated intensity maps in the first row of the verification image show a concentrated morphology; (2) The average spectrum (second row) and maximum spectrum (third row) exhibit clear emission peaks.

As shown in FIGURE:3, this C18O core verification image satisfies both the concentrated integrated intensity map and clear spectral line peaks, thus it is identified as a genuine C18O core. In FIGURE:3, the integrated intensity map is not concentrated and no clear spectral peaks are visible; in FIGURE:3, the integrated map is concentrated but no clear spectral peaks are observed; in FIGURE:3, the integrated map is not concentrated but spectral peaks are present. Therefore, categories (b), (c), and (d) are identified as false C18O cores.

During this verification process, we adopted a three-person voting strategy. Each candidate was independently verified by three individuals, and when two or more considered it a genuine C18O core, the candidate was marked as a real C18O core. A total of 544 genuine C18O cores were obtained. [TABLE:3] shows a portion of the core catalog.

2.3 Matching of 13CO Molecular Clumps and C18O Molecular Cores

Based on the positional relationships in three-dimensional space between the samples from the two spectral lines, we matched 7,097 13CO molecular clumps with 544 C18O molecular cores. If the centroid of a C18O core falls within the boundary of a 13CO clump, the match is considered successful, meaning that the clump contains a C18O core. The 13CO clump boundaries are obtained from the masks provided by FacetClumps (see Mask in Section 2.2.2). Following this criterion, all 13CO clumps were matched, with results shown in [TABLE:4]. The total number of 13CO clumps matched with C18O cores is presented in the first row of the table. The matching ratio is defined as Matching Number divided by Sample Number.

Among all 544 C18O molecular cores, only two were not matched to corresponding 13CO clumps. We analyzed the 13CO clumps at these positions and found that 13CO clumps do exist at these locations, but their confidence levels are below 0.9 and therefore not included in our selected 13CO clump sample (Section 2.2.1).

Within the specified velocity range, a total of 424 13CO clumps contain C18O cores, with some 13CO clumps matching multiple C18O cores. [FIGURE:4] and [FIGURE:5] illustrate the positional relationships and matching numbers. In [FIGURE:4], subpanel (a) shows blue and magenta triangles, representing a 13CO clump containing two C18O cores. In [FIGURE:5], subpanel (a) shows only a blue triangle, representing a 13CO clump containing one C18O core. Statistics show that 78% of 13CO clumps have one C18O core, 17% have two C18O cores, and the remaining 5% have more than two C18O cores ([FIGURE:6]).

The matching results demonstrate that all C18O cores are contained within 13CO clumps, confirming that the optically thin C18O molecular probe can trace denser substructures within molecular clumps.

3 Statistical Analysis of 13CO Clump Parameters

In Section 2, we completed the matching of 13CO clumps and C18O cores. Based on these results, we performed binary classification of the 13CO clumps, distinguishing between those with and without C18O cores. The 424 13CO clumps in [TABLE:4] are designated as Match C18O clumps (abbreviated as M-type), while the remaining 13CO clumps are designated as No Match C18O clumps (abbreviated as NM-type).

To investigate differences in parameters between these two categories of 13CO clumps, we conducted statistical analysis and comparison of their distribution in the Galactic plane, peak intensities, fluxes, and velocity spans.

3.1 Position Distribution

We examined differences in the Galactic position distribution of the two 13CO clump categories through probability density functions in galactic longitude, latitude, and radial velocity.

As shown in FIGURE:7, colored lines represent the mean longitudes of different massive star-forming regions. In terms of galactic longitude distribution, both categories show concentration trends in star-forming regions. However, since the number of M-type clumps is far smaller than NM-type clumps, the consistency between their peak positions and star-forming regions is less pronounced than for the more numerous NM-type clumps.

In FIGURE:7, between galactic latitudes -5°.25 and 5°.25, both M-type and NM-type clumps are concentrated in the range (-3°, 3°), but M-type clumps are more concentrated at latitude 0°, i.e., the central region of the Galactic plane. This indicates that 13CO clumps containing C18O cores have a higher probability of appearing at galactic latitude 0°.

In FIGURE:7, the radial velocity center distribution of NM-type clumps is similar to that of M-type clumps. Based on the 0.25–0.75 quartiles in the velocity center statistics, both are concentrated in the range (20, 45) km·s⁻¹. However, M-type clumps rarely appear at velocities less than -5 km·s⁻¹ or greater than 75 km·s⁻¹. NM-type clumps have a small fraction at velocities greater than 75 km·s⁻¹ because these clumps are located in more distant gas spiral arms where C18O detection cannot cover these regions.

3.2 Velocity Span

FIGURE:7 shows the probability density distributions of velocity spans for M-type and NM-type clumps. Velocity span is defined as the difference between the maximum and minimum velocities of a clump. For M-type clumps, velocity spans range from 2.66 to 10.13 km·s⁻¹, with most sources (0.25–0.75 quartiles) having values between 4.82 and 6.14 km·s⁻¹. For NM-type clumps, velocity spans range from 1.36 to 18.58 km·s⁻¹, concentrated (0.25–0.75 quartiles) between 3.65 and 4.82 km·s⁻¹. This indicates that M-type clumps have larger velocity spans.

3.3 Peak Intensity

FIGURE:8 shows the differences in peak intensity between M-type and NM-type clumps. For M-type clumps, peak intensity ranges from 3.53 to 30.71 K, while for NM-type clumps, it ranges from 1.36 to 18.58 K. Compared to NM-type clumps, the peak intensity distribution of M-type clumps is shifted overall to higher values. The median peak intensity is 7.62 K for M-type and 3.40 K for NM-type, a difference of nearly 55%. Additionally, the mean peak intensity of M-type 13CO clumps is 4.61 K higher than that of NM-type clumps, indicating that M-type clumps have greater peak intensities.

3.4 Flux

FIGURE:8 shows the flux distribution. The median flux of M-type clumps is 1363.32 K·km·s⁻¹, which is 3.83 times that of NM-type clumps (median 355.95 K·km·s⁻¹). The mean flux of M-type clumps is 1755.00 K·km·s⁻¹, significantly greater than the NM-type mean of 469.21 K·km·s⁻¹. Furthermore, we calculated the total flux of all molecular clumps and both categories. The total flux of all 7,097 13CO molecular clumps is . M-type clumps, accounting for only 5.97% of the total number, contribute 20% of the total flux ( ).

4 Morphological Parameter Analysis

4.1 13CO Clump Morphology Analysis

4.1.1 Eccentricity

Real 13CO clumps have irregular shapes, while the ideal model for clumps is a three-dimensional Gaussian, whose projection on a two-dimensional plane is an ellipse. Therefore, we chose to perform elliptical fitting on the two-dimensional projections of 13CO clumps, using eccentricity as a reference indicator for shape. Elliptical eccentricity is an important parameter describing ellipse shape, reflecting its degree of flattening, with values ranging from 0 to 1. The procedure consists of the following steps: (1) Integrate each 13CO clump's mask data along the velocity direction to use the projected two-dimensional boundary contour as input for elliptical fitting. (2) Use the Ellipse function from the matplotlib.patches library to complete contour fitting. (3) Calculate eccentricity from the output major axis a and minor axis b:

FIGURE:9 shows an example of successful 13CO clump contour fitting. We examined all fitting results and found that among 7,097 13CO clumps, 11 failed fitting (0.2%), with 9 from NM-type clumps. Fitting failures occur because these clump contours are too irregular and deviate significantly from the elliptical model, as shown in FIGURE:9. Due to the small proportion, these 11 clumps were removed without affecting the statistical conclusions.

We compared the eccentricities of the two clump categories, with statistical distributions shown in FIGURE:10. Both have similar eccentricity ranges: 0.24–0.89 for M-type and 0.11–0.95 for NM-type, with M-type clumps showing a more concentrated distribution. The median eccentricity is 0.64 for M-type and 0.67 for NM-type, with similar relationships for the means.

Overall, from the perspective of elliptical fitting, the vast majority of 13CO clumps can be successfully fitted, indicating that using ellipses to describe 13CO clumps in two-dimensional (l–b) integrated maps is fundamentally reasonable. Considering both categories together, there is no significant difference in eccentricity distributions between the two types, though M-type clumps have slightly smaller mean and median values.

4.1.2 Angular Area

Angular area can describe clump size (actual clump size is strongly affected by distance; here we discuss only the raw detection results). Angular area is defined as the pixel resolution of 0.25 square arcminutes (arcmin²) multiplied by the number of pixels in the velocity-integrated projection of the masked clump on the two-dimensional plane:

In FIGURE:10, M-type clump angular areas range from 30.75 to 227.5 arcmin², concentrated between 65.25 and 107.56 arcmin². NM-type clump angular areas range from 14.0 to 255.25 arcmin², mainly between 45.75 and 77.0 arcmin². The former is about 30% larger in median value. Based on the angular area distribution and the median and mean values, M-type clumps have larger angular areas.

4.1.3 Form Factor

Form Factor (FF) is a parameter that quantitatively describes object shape. For a perfect circle, FF = 1, and deviation from circularity results in FF < 1. The form factor indicates that values closer to 1 correspond to more circular gas structures. It is calculated as:

where A represents area and C represents perimeter. We used the polygonarea and polygonlength functions from the shapely library to obtain clump area and perimeter for calculating 13CO clump form factors. The distribution of form factors is shown in FIGURE:10. Both categories have a median of 0.57. In terms of concentrated distribution intervals, M-type clumps range from 0.51 to 0.63, while NM-type clumps range from 0.50 to 0.60. The two categories are very similar, indicating that 13CO clump form factors do not differ significantly based on the presence of C18O cores.

4.2 Morphology of 13CO Clumps and C18O Cores

13CO is generally optically thin, making this line suitable for tracing dense molecular clumps, while completely optically thin C18O can trace even smaller or denser regions. We also compared the morphologies of sources traced by different density probes by calculating form factors for M-type 13CO clumps and C18O cores. The distribution is shown in FIGURE:10. We fitted the distribution histograms with Gaussian curves, where μ and σ represent the fitted mean and standard deviation. The results show clear differences in form factors between sources detected by the two different density probes.

The fitted parameter μ is 0.56 for M-type 13CO clumps and 0.82 for C18O cores, indicating that C18O cores are wrapped inside 13CO clumps with more circular morphologies. In summary, molecular cloud cores are overall smaller and more circular than M-type molecular clumps, representing more regular high-density substructures within 13CO clumps.

5 Conclusions

In the 10°≤l≤20° region of the Milky Way Imaging Scroll Painting survey, we completed the classification of 13CO molecular clumps based on whether they contain C18O cores. The main conclusions are:

Among 7,097 13CO molecular clumps, 424 contain C18O cores (M-type clumps), while the remaining 13CO clumps are NM-type. Specifically, 78% of 13CO clumps have one C18O core, 17% have two C18O cores, and the remaining 5% have more than two C18O cores. 13CO clumps containing C18O cores tend to have larger velocity spans, peak intensities, and fluxes. In the entire 13CO molecular clump sample, M-type clumps account for only 5.97% of the total number but contribute 20% of the total flux. M-type and NM-type clumps show slight differences in morphological parameters. In terms of eccentricity, both categories have means greater than 0.5, indicating that ellipses are reasonable for describing 13CO clumps in two-dimensional morphology. From angular area analysis, the former is about 30% larger in median value than the latter, suggesting that larger 13CO clumps are more likely to form denser C18O cores within them. Regarding form factor, C18O cores contained within 13CO clumps have more regular morphologies: 13CO clumps are more elliptical, while C18O cores are more circular. This pattern indicates that higher-density molecular probes detect more regular structures, possibly suggesting that gravity plays an increasingly important role in molecular core formation as molecular density increases.

Acknowledgments

The data used in this research comes from the Milky Way Imaging Scroll Painting survey. MWISP is a multi-line CO, 13CO & C18O survey of the northern Galactic plane conducted with the Purple Mountain Observatory 13.7 m telescope. We are deeply grateful to all members of the MWISP survey team, especially the staff at the Qinghai Observatory Station, for their long-term efforts.

References

Blitz L, Stark A A. ApJ, 1986, 300: L89
Rosolowsky E W, Pineda J E, Foster J B, et al. ApJS, 2008, 175: 509
Ballesteros-Paredes J, André P, Hennebelle P, et al. SSRv, 2020, 216: 5
Gomez G C, Vazquez-Semadeni E. ApJ, 2014, 791: 124
Shu F H, Adams F C, Lizano S. ARA&A, 1987, 25: 23
Bontemps S, Motte F, Csengeri T, et al. A&A, 2010, 524: A18
Wilson R W, Jefferts K B, Penzias A A. ApJ, 1970, 161: L43
Bolatto A D, Wolfire M, Leroy A K. ARA&A, 2013, 51: 207
孙锦, 李守中. 分子天体物理学基础. 北京: 北京师范大学出版社, 2004: 12-15
Bo R, Jewitt D, Keil K. Protostars & Planets V// Francesco J D, Evans-Ⅱ N J, Caselli P, et al. An Observational Perspective of Low-Mass Dense Cores I: Internal Physical and Chemical Properties. Tucson: University of Arizona Press, 2006: 17
Lada C J, Muench A A, Rathborne J, et al. ApJ, 2008, 672: 410
Mark H H, Thomas M D. ARA&A, 2015, 53: 583
Dame T M, Hartmann D, Thaddeus P. ApJ, 2001, 547: 792
Jackson J M, Rathborne J M, Shah R Y, et al. ApJS, 2006, 163: 145
Rigby A J, Moore T J T, Plume R, et al. MNRAS, 2016, 456: 2885
Schuller F, Csengeri T, Urquhart J S, et al. A&A, 2017, 601: 25
Su Y, Yang J, Zhang S B, et al. ApJS, 2019, 240: 9
Yuan L X, Yang J, Du F J, et al. ApJS, 2021, 257: 51
Neralwar K R, Colombo D, Duarte-Cabral A, et al. A&A, 2022, 663: A56
Jaffa S E, Whitworth A P, Clarke S D, et al. MNRAS, 2018, 477: 1927
Clarke S D, Jaffa S E, Whitworth A P. MNRAS, 2022, 516: 2782
Stutzki J, Guesten R. ApJ, 1990, 356: 513
Williams J P, de Geus E J, Blitz L. ApJ, 1994, 428: 693
Berry D S. A&C, 2015, 10: 22
Rosolowsky E W, Pineda J E, Kauffmann J, et al. ApJ, 2008, 679: 1338
Maity A K, Dewangan L K, Sano H, et al. ApJ, 2022, 934: 2
Luo X Y, Zheng S, Huang Y, et al. RAA, 2022, 22: 12
Messineo M, Clark J S, Figer D F, et al. ApJ, 2015, 805: 110
Jiang Y, Zheng S, Jiang Z B, et al. A&C, 2022, 40: 100518
Oswalt T D, Gilmore G. Planets, Stars and Stellar Systems. Volume 5: Galactic Structure and Stellar Populations//Kroupa P, Weidner C, Pflamm-Altenburg J, et al. The Stellar and Sub-Stellar Initial Mass Function of Simple and Composite Populations. Heidelberg: Springer, 2013: 115
Jiang Y, Chen Z W, Zheng S, et al. ApJS, 2023, 267: 32
Luo X Y, Zheng S, Jiang Z B, et al. RAA, 2024, 24: 015012
Shan W L, Yang J, Shi S C, et al. ITTST, 2012, 2: 593
孙继先, 逯登荣, 杨戟, 等. 天文学报, 2018, 59: 3
李振强, 张旭国, 徐烨, 等. 天文学报, 2021, 62: 24
Hill T, Motte F, Didelon P, et al. A&A, 2012, 542: A79
Chen Z W, Sun W, Chini R, et al. ApJ, 2021, 922: 90
辛科霆, 安东升, 谢基伟, 等. 天文学报, 2023, 64: 46
Shin K T, An D S, Xie J W, et al. ChA&A, 2024, 48: 1
申洪, 朴英杰. 中国体视学与图像分析, 1999, 4: 10