TRAO Survey of Nearby Filamentary Molecular clouds, the Universal Nursery of Stars (TRAO FUNS) I. Dynamics and Chemistry of L1478 in the California Molecular Cloud
Eun Jung Chung, Chang Won Lee, Shinyoung Kim, Gwanjeong Kim, Paola Caselli, Mario Tafalla, Philip C. Myers, Archana Soam, Tie Liu, Maheswar Gopinathan, Miryang Kim, Kyoung Hee Kim, Woojin Kwon, Hyunwoo Kang, Changhoon Lee
DDraft version April 25, 2019
Typeset using L A TEX twocolumn style in AASTeX62
TRAO Survey of Nearby Filamentary Molecular clouds, the Universal Nursery of Stars (TRAO FUNS) I.Dynamics and Chemistry of L1478 in the California Molecular Cloud
Eun Jung Chung, Chang Won Lee, Shinyoung Kim, Gwanjeong Kim, Paola Caselli, Mario Tafalla, Philip C. Myers, Archana Soam, Tie Liu, Maheswar Gopinathan, Miryang Kim, Kyoung Hee Kim, Woojin Kwon, Hyunwoo Kang, and Changhoon Lee Korea Astronomy and Space Science Institute, 776 Daedeokdae-ro, Yuseong-gu, Daejeon 34055, Republic of Korea Nobeyama Radio Observatory, National Astronomical Observatory of Japan, Nagano 384-1305, Japan Max-Planck-Institut f ¨ u r Extraterrestrische Physik, D-85748 Garching, Germany Observatorio Astron ´ o mico Nacional (IGN), Alfonso XII 3, 28014 Madrid, Spain Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA SOFIA Science Centre, USRA, NASA Ames Research Centre, MS N232, Moffett Field, CA 94035, USA Indian Institute of Astrophysics, Koramangala, Bangalore 560034, India Department of Earth Science Education, Kongju National University, 56 Gongjudaehak-ro, Gongju-si, Chungcheongnam-do 32588,Korea (Accepted for publication in the ApJ)
ABSTRACT (cid:48)(cid:48)
TRAO FUNS (cid:48)(cid:48) is a project to survey Gould Belt’s clouds in molecular lines. This paper presents itsfirst results on the central region of the California molecular cloud, L1478. We performed On-The-Flymapping observations using the Taedeok Radio Astronomy Observatory (TRAO) 14m single dishtelescope equipped with a 16 multi-beam array covering ∼ O(1 −
0) mainly tracing low density cloud and about 460 square arcminute area using N H + (1 − −
1) and SO(3 − ) were also used simultaneously to map ∼ O data-cube and 8 dense N H + cores by using FellWalker . Basic physical properties offilaments such as mass, length, width, velocity field, and velocity dispersion are derived. It is foundthat L1478 consists of several filaments with slightly different velocities. Especially the filaments whichare supercritical are found to contain dense cores detected in N H + . Comparison of non-thermalvelocity dispersions derived from C O and N H + for the filaments and dense cores indicates thatsome of dense cores share similar kinematics with those of the surrounding filaments while severaldense cores have different kinematics with those of their filaments. This suggests that the formationmechanism of dense cores and filaments can be different in individual filaments depending on theirmorphologies and environments. Keywords:
ISM: clouds — ISM: kinematics and dynamics — ISM: structure — stars: formation INTRODUCTIONHow stars form in molecular clouds is one of the keyquestions in astronomy. In general stars are known toform by a gravitational contraction in dense cores whichare made in less dense molecular clouds by their hi-erarchical fragmentation. Recent high resolution ob-servations mainly done by
Spitzer
Space Telescope and Herschel
Space Observatory reveal that molecularclouds are filamentary and such a structure is ubiqui-tous over various star-forming environments from activestar-forming molecular clouds like the Orion molecularcomplex to non-star-forming molecular clouds such asthe Polaris flare (e.g., Andr´e et al. 2010; Hacar et al.2018). a r X i v : . [ a s t r o - ph . GA ] A p r Chung et al.
During last decade, several studies have been done andprogresses are made in understanding the physical prop-erties of filaments and dense cores. One of the most in-teresting findings about filaments is that filaments havea characteristic width of 0.1 pc which is comparable tothe typical size of dense cores (e.g., Andr´e et al. 2010;Arzoumanian et al. 2011; Palmeirim et al. 2013; Fed-errath 2016; Arzoumanian et al. 2019). Most prestel-lar cores are found on the dense, supercritical filamentswhere the mass per unit length is larger than the criti-cal value of isothermal cylinders (e.g., Andr´e et al. 2010;K¨onyves et al. 2015; Marsh et al. 2016). Many youngstellar groups in the nearby molecular clouds are foundto be well associated with Hub-filament structure thatconsists of a hub which is a central body with relativelyhigher column density ( > cm − ) and filaments ra-diated from the hub which has lower column density(e.g., Myers 2009, and references therein). Besides, itis observed velocity gradients along filaments and thegas flow along filaments is responsible for the formationof the star cluster (e.g., Kirk et al. 2013; Peretto et al.2014; Imara et al. 2017; Baug et al. 2018; Yuan et al.2018). Hence, it seems clear that filaments can play acrucial role in the formation of cores and stars.These results raise important questions about such as1) how filaments and dense cores form in large molecu-lar clouds? 2) are filaments an intermediate stage of starformation, i.e., from large clouds to dense cores? Thesecan be answered by probing the kinematics and chem-istry of filaments and dense cores with systematic molec-ular line observations toward various molecular clouds.We have been performing such observations for fil-ament clouds using Taedeok Radio Astronomy Obser-vatory (TRAO)
14m antenna with a project (cid:48)(cid:48)
TRAOFUNS (cid:48)(cid:48) which is an acronym for (cid:48)(cid:48) the TRAO survey ofFilaments, the Universal Nursery of Stars (cid:48)(cid:48) . This projectis to make a systematic survey for 10 Gould Belt’s cloudswith several molecular lines in various environments,aiming to obtain 1) the velocity structure of filamentsand dense cores for the study of their formation, 2) radialaccretion or inward motions toward dense cores fromtheir surrounding filaments, and 3) chemical differen-tiation of filaments and their dense cores. For thesegoals, six molecular lines of C O (1 − CO (1 − H + (1 − + (1 − − ), and CS (2 − http://radio.kasi.re.kr/trao/main trao.php In this paper, we present the first results of the TRAOFUNS survey toward L1478 in the California molecu-lar cloud. The California molecular cloud (CMC, here-after) which is also called the Auriga-California molecu-lar cloud has been recently recognized as a massive giantmolecular cloud by Lada et al. (2009). They used in-frared extinction map from the Two Micron All Sky Sur-vey (2MASS; Kleinmann et al. 1994) and CO maps fromthe Galactic plane survey (Dame et al. 2001), and foundcontinuous distribution of the molecular cloud in veloc-ity and space as a single molecular cloud at the same dis-tance. CMC is located at a distance of 450 ±
23 pc, andit is comparable in size ( ∼
80 pc) and mass ( ∼ M (cid:12) )to the Orion giant molecular cloud (OMC). However,the number of young stellar objects (YSOs) in CMC(149) is 15-20 times smaller than that of OMC (3330,Broekhoven-Fiene et al. 2014). Harvey et al. (2013) in-vestigated the young stellar objects and dense gas ofCMC, finding 60 compact sources at 70/160 µm and 11cold, compact sources at 1.1 mm. Recently, Broekhoven-Fiene et al. (2018) identified 59 candidate protostars,and found that 24 among them are associated withYSOs in the catalogs of Spitzer and
Herschel/
PACS.They suggested that CMC is significantly less efficientin star formation than Orion A. There are recent obser-vations of molecular lines for a part of CMC by Imaraet al. (2017). They investigated the relationship of fila-ments and dense cores in the CMC region ( ∼ . ◦ × . ◦ area) with dust continuum data ( Herschel ) and COand CO (2-1) molecular line data obtained from theHeinrich Hertz Submillimeter Telescope, finding that fil-aments in the west region of L1478 are velocity-coherentand gravitationally supercritical.We have investigated filaments and dense cores ofCMC, L1478 ( ∼ ◦ × . ◦ ), which provides a good labo-ratory to investigate the formation of filament and densecores. Various morphologies of filaments such as a longnetwork of filaments and a hub-filaments structure canbe found (see Figure 1). Besides, the star forming prop-erty of L1478 is relatively modest and seems to be alow mass star forming region, in a sense that only threeYSOs are found, and thus can be a good comparisonto the OMC (Broekhoven-Fiene et al. 2018). Our sur-vey provides the first mapping observations in variousmolecule lines toward the central region of the CMC.This paper is organized as follows. In section 2, obser-vation and data reduction are explained. Identificationsof filament and dense core and their basic physical prop-erties are presented in section 3. In section 4, we discussabout velocity structure of filaments, gravitational in-stability, and the formation mechanisms of filaments RAO FUNS I. L1478 in the California MC OBSERVATIONS AND DATA REDUCTION2.1.
Observations
We have carried out On-The-Fly (OTF) mapping ob-servations toward L1478 in CMC with TRAO 14-m tele-scope from January to May in 2017 and from December2017 to February 2018. The equipped frontend is SEc-ond QUabbin Optical Image Array (SEQUOIA-TRAO),which consists of a 4 × (cid:48)(cid:48) . The backend, FFT spectrometer, is4096 × ∼ .
04 km s − at110 GHz) which covers a total bandwidth of 62.5 MHzcorresponding to ∼
170 km s − at 110 GHz. TRAOsystems allow simultaneous observations of 2 molecularlines between the frequency range of 85 and 100 GHzor 100 and 115 GHz. The beam efficiency at 90 GHz is0.48 and at 110 GHz is 0.46.To investigate the physical properties of filaments anddense cores, six molecular lines are observed. C O andN H + (1 −
0) molecular lines are chosen as a tracer of rel-atively less dense material for the filaments and a densegas tracer for dense cores, respectively. CO(1 − O. With N H + observation,HCO + (1 −
0) is concurrently observed. To probe thechemical evolution of dense cores, SO (3 − ) is selectedand CS (2 −
1) observed at the same time. SO(32-21) isknown as one of the most sensitive molecules to the de-pletion and hence can be used as a tracer of very youngdense cores (Tafalla et al. 2006). CS(2-1) line was cho-sen to be useful to study infall motions in the pre-stellarcores (Lee et al. 2001). In this study, four molecularlines of C O, N H + , CS, and SO are mainly used inanalyses.We divided our target area of C O (and CO si-multaneously) into five regions which have a box shapereferred as ‘tiles’ hereafter, shown in Figure 1, and car-ried out OTF mapping observations. Each tile has asize from 12 (cid:48) × (cid:48) to 32 (cid:48) × (cid:48) . The scanning rate was55 (cid:48)(cid:48) per second and the integration time is 0.2 second.Scan step of 0.25 HPBW (44 (cid:48)(cid:48) ) along the scan directionand 0.75 HPBW separation between the rows are ap-plied for C O and CO. We carried out OTF mappingobservations of N H + and HCO + towards the regionsonly where C O is strongly detected. Five tiles wereobserved with a scan step of 0.25 HPBW (44 (cid:48)(cid:48) ) alongthe scan direction and a 0.25 HPBW separation betweenthe rows. Simultaneous observation of SO and CS linesare performed with the same OTF mapping parametersas the observations of C O and CO, but for only four tiles. We made maps alternatively along RA andDec directions. In Figure 1, The observed area of the sixmolecular lines are presented with the
Herschel µ mcontinuum image. Toward two points where C O isstrongly detected but not carried out OTF observationof N H + and HCO + , additional position-switching (PS)observations of N H + (and HCO + ) have been carriedout due to the lack of observing time (denoted with redcrosses in Figure 1. The rms noise level of PS observa-tions is about 0.06 K in the unit of antenna temperaturefor both lines. 2.2. Data Reduction
The raw OTF data for each map of each tile is readand produced into a map with jinc-gaussian functionafter baseline fitting (with 1st order) in
Otftool .We give the resulted cell size of 22 (cid:48)(cid:48) and apply noise-weighting. Further reduction and examinations are donewith the
Class package. Since the baseline is not goodin both ends of the band and the total velocity rangeof the spectra ( ∼
170 km s − ) is much larger than thatof emissions in our object (less than 20 km s − ), base-line fittings are done in two steps. Firstly, both endsof the spectra were cut off so that the velocity rangebecomes 120 km s − and a baseline was subtracted withthe second order polynomial. After that, the spectra areresampled with channel width of 0.06 km s − , both endswere cut off again resulting in a spectral velocity rangeof 60 km s − , and baseline subtraction was done againwith first order polynomial. After this step, all channelmaps are examined and maps having high noise level orshowing some spatial gradient of noise level due to thechange of system temperature were excluded. Finally,the maps are merged into a final fits cube with 44 (cid:48)(cid:48) cellsize and 0.1 km s − velocity channel width for C Oand CO, and 22 (cid:48)(cid:48) cell size and 0.06 km s − velocitychannel width for the other molecular lines. The basicobservational information is given in Table 1. FILAMENTS AND DENSE CORES3.1.
Filament Identification
Figure 2shows the integrated intensity maps of CO(top) and C O (bottom) on
Herschel µ m image.The distributions of molecules seem to well match thatof dust continuum. The long filamentary structure fromthe southeast to the center of the observed area is wellrevealed by C O as well as CO. It spreads out to ∼ Chung et al. D e c ( J ) F l u x ( m J y / s r ) Figure 1.
Herschel µ m image of the Auriga-California region. The area covered in this study, L1478, is indicated by white( CO and C O 1 − H + and HCO + − − and CS 2 −
1) boxes. The red crosses show theregion where position-switching mode observations of N H + and HCO + molecular lines have been done. Table 1.
ObservationsMolecule ν ref a θ FWHMb
Area c θ pixeld δv e rms f (GHz) ( (cid:48)(cid:48) ) (sq. arcmin) ( (cid:48)(cid:48) ) (km s − ) (K[T ∗ A ])C O (1 −
0) 109.782160 47 ∼ CO (1 −
0) 110.201353 47 ∼ H + (1 −
0) 93.173764 56 ∼
460 22 0.06 0.063HCO + (1 −
0) 89.188525 58 ∼
460 22 0.06 0.062SO (3 − ) 99.299870 52 ∼
440 22 0.06 0.094CS (2 −
1) 97.980953 52 ∼
440 22 0.06 0.097 a Rest frequency of each molecular line is taken from The Cologne Database forMolecular Spectroscopy (CDMS: M¨uller et al. 2001, https://cdms.ph1.uni-koeln.de/cdms/portal/). b Full width half maximum of the beam c Total observed area d,e
The pixel size and channel width of the final datacube f Noise level in T ∗ A of the final datacube degree in the sky. The other noticeable feature is thehub-filament structure in the northwest that is referredas Cal-X due to its X shape by Imara et al. (2017). Thehub radiates four filaments to the east, south, west, andnorth. CO is detected at the outer edge of the observedregions, while C O is well matched to the filamentarydust emission. Most of CO spectra show multiple ve-locity components. In the top panel, CO emission isdetected at ∼ − and ∼ − . − , while C Ois only detected at ∼ − . − with a single Gaus-sian shape indicating that the double peak componentat ∼ − . − of CO is a self-absorption feature.Therefore, CO spectra are thought to be self-absorbed toward some dense regions shown in the spectra (a) to(d), because it has relatively larger optical depth thanthat of C O. However, both CO and C O spectraat several positions drawn in the bottom panel of Fig-ure 2shows double peaked features, implying that thefilaments at these places may consist of multiple com-ponents to the line-of-sight.To identify filaments with multiple velocity structure,we used astrodendro Python package and applied it tothe C O data cube. A dendrogram is a tree diagramwhich shows how and where the structures merge and https://dendrograms.readthedocs.io RAO FUNS I. L1478 in the California MC h m m m m m RA (J2000)+36 ◦ +37 ◦ D ec ( J ) − − − T A ∗ ( K ) (a) COC O V lsr (km s − ) − − − (b) − − (c) − − − (d) h m m m m m RA (J2000)+36 ◦ +37 ◦ D ec ( J ) − − − T A ∗ ( K ) (e) × V lsr (km s − ) − − − (f) × − − − (g) × − − (h) × Figure 2.
Integrated intensity maps of CO(1-0) (top) and C O(1-0) (bottom) toward L1478. Background color imagesare
Herschel µ m data and the CO intensity maps are shown in contours. The line intensity maps are integrated over avelocity range of − − for CO, and − − for C O. The contour levels of CO are 7 n × σ rms ( n = 1 , , · · · ,
6) and those of C O are 2 and 3 n × σ rms ( n = 1 , , · · · , CO (gray) and C O (black) aregiven in (a)-(h) inset windows at the selected positions. The spectra in (a)-(d) windows are given to illustrate that C O traceswell the velocity field of filament material while CO more or less self-absorbed due to its larger optical depth compared withC O. The spectra in (e)-(h) are shown to indicate that C O profiles have two components and thus some of filaments canhave the multiple velocity structure to the line-of-sight.
Chung et al. its algorithm identifies the hierarchical structure of 2-and 3-dimensional datasets. Structures start from localmaxima, their volumes get bigger merging with the sur-roundings with lower flux densities, and stop when theymeet neighboring structures (Rosolowsky et al. 2008).Details of filament identification using the dendrogramtechnique are given in the Appendix.Figure 3shows the results of this dendrogram anal-ysis. Resulted branches and leaves are shown in thedendrogram tree (top) and its spatial distribution inthe filament clouds is presented in the bottom panelon the C O moment 0 image. Each identified leave isindicated with color-coded base on their numbers, i.e.,the same color in the tree and map. There are five falseleaves found due to noisy observation near the bound-ary of the map. We excluded them and used the other10 filaments in the analysis. With the given mask indendrogram, we extract datacubes of each filaments anduse them for further analyses, i.e., central velocity andvelocity dispersion. It is noticeable here that F6 (fila-ment 6), F7, and F8 are identified as an independentleaf but they have a single stem, and this is the same forfilaments from F1 to F5. Only F9 and F10 have theirown stems. We will discuss about these filaments later.3.2.
Dense Core Identification
The N H + (1 −
0) molecular line, which is usually op-tically thin, is an appropriate tracer of dense cores innearby star forming regions (e.g., Caselli et al. 1995; San-hueza et al. 2012). To probe the relationship of densecores and filaments, we made five OTF tiles of N H + toward the regions where C O emission is strongly de-tected. Two tiles are in F4, one in F5, one in F6 andone in F8 (see Figures 1and 3). N H + emission is de-tected in all the observed tiles except in F6. We alsocarried out PS observations toward two positions in F7(marked with red crosses in Figure 1), but N H + wasnot detected at the rms level of 0.06 K[T ∗ A ].We applied the FellWalker source extraction al-gorithm (Berry 2015) to the N H + integrated intensityimage to find dense cores. In running this algorithmonly pixels whose intensities are higher than 3 σ wereconsidered. The required minimum number of pixelsthat a core should include to be identified as a real coreis seven, i.e., it should be larger than one beam size of56 (cid:48)(cid:48) . In case there are neighboring peaks, if the differ-ence between the peak value and the minimum value(dip value) is larger than 1 σ rms , it is considered as anindependent core. With these criteria, we obtained eightcores, three in F4, one in F5, and four in F8. The masses of the identified dense core are derivedwith integrated intensity of N H + . We calculated to-tal column density of N H + following Equation (A4)of Caselli et al. (2002) and converted it into the corre-sponding H column density with an average abundanceof N H + of ∼ . ± . × − (Johnstone et al. 2010;Lee & Myers 2011). The uncertainties in the N H + intensities are typically less then 30%, while the un-certainties of excitation temperature and the conversionfactor between the column densities of N H + and H arequite large but less than a factor of 2 (Johnstone et al.2010). Hence, the uncertainties of dense core massesare less than a factor of 2. The virial masses ( M vir ) arederived with the equation of M vir = k R σ /G where R and σ are the radius and total velocity dispersionof the core, respectively (MacLaren et al. 1988). Forsimplicity, we used k = 1 assuming a density profile of ρ ∝ R − where R is the core radius. The dense coresidentified show virial parameters, α = M vir /M , between ∼ α < Physical Properties of the identified filaments
For the ten filaments identified with the dendrogramtechnique, we derived physical quantities such as molec-ular gas mass, length and width, mass per unit length,peak velocity distribution, and nonthermal velocity dis-persion. 3.3.1. H Column Density and Mass
We estimated H column density from the C O col-umn density using the formula (Garden et al. 1991; Pat-tle et al. 2015) : N = 3 k B π Bµ e hBJ ( J +1) /k B T ex J + 1 T ex + hB k B − e − hν/k B T ex (cid:90) τ d v , (1)where B is the rotational constant, µ is the permanentdipole moment of the molecule, J is the lower rotationallevel, and T ex is the excitation temperature. The exci-tation temperature is calculated following Pineda et al.(2008) : T ex = T ln(1 + T ( T R − e − τ + T e T / T bg − ) − ) (2) RAO FUNS I. L1478 in the California MC
280 290 300 310 320 330 340 350
Structure . . . . . . . F l u x F5F5F4F4F3F2F2F1F1F10F10F9F9F8F8F7F7F6F6 h m s m s m s m s m s R.A. (J2000) +36 ◦ +37 ◦ D ec . ( J ) F5 F4 F3 F2F1 F10 F9 F8 F6F7 160240320400480560640720 [ ( K k m / s ) / ] Figure 3. Top :
Branches and leaves for L1478 identified by dendrogram analysis. The x-axis indicates an identification numberof a structure in L1478 and y-axis denotes the intensity of each structure in K[T ∗ A ]. Bottom :
Leaves in the dendrogram overlaidon the C O(1-0) integrated intensity map (the grayscale image). Each identified leaves in the top panel are drawn in this panelas filament, named as F1 to F10. Leaves found in the observing boundary are usually noisy structures falsely identified andthus excluded in further analyses. The red crosses present the locations of dense cores identified with N H + data (Section 3.2). where T = hν/k B and T bg is the cosmic microwavebackground temperature (2.73 K). We used the radia- tion temperature of CO for T R with assumptions thatC O and CO trace materials with the same excitation
Chung et al. h m s m s s R.A. (J2000)+37 ◦ D ec . ( J ) C1C2C3
Filament 4 region
F4-C1F4-C2 − − − )0 . . . T ∗ A ( K ) F4-C34 h m s s R.A. (J2000)+36 ◦ +37 ◦ D ec . ( J ) C1 Filament 5 region − − − )0 . . . . T ∗ A ( K ) F5-C1
Figure 4. Left :
Identified dense cores using
FellWalker is presented with red ellipses on the N H + integrated intensitymaps (moment 0 maps) of the Filament 4 and Filament 5. The 56 (cid:48)(cid:48) FWHM beam at 93.176 GHz is shown by the white circle.The red stars represent positions of YSOs reported by Broekhoven-Fiene et al. (2018).
Right : N H + spectra of dense coresare presented with hyperfine fitting results (green lines). temperature and CO is optically thick. We derivedoptical depths τ of CO and C O with the abundanceratio of [ CO / C O] = 5.5 (Frerking et al. 1982) andthe relation of T max , C O T max , CO = 1 − e − τ C18O − e − τ , (3)where T max , C O and T max , CO are the maximum inten-sities of C O and CO, respectively. In regions farfrom the main filaments, CO is not necessarily op-tically thick. In this case we adopted the excitationtemperature for their nearest position where τ CO > CO data and provided ∼ T ex of these two filaments ranges about 5 to 9.5 K and theaverage T ex is ∼ (cid:90) τ ( v ) d v = 1 J ( T ex ) − J ( T bg ) (cid:90) τ ( v )1 − e − τ ( v ) T mb d v ≈ J ( T ex ) − J ( T bg ) τ ( v )1 − e − τ ( v ) (cid:90) T mb d v , where v is the central velocity, T mb is the observed mainbeam temperature of the line and J ( T ) is the sourcefunction, J ( T ) = T / (e T / T −
1) and T = hν/k B , and T ex and T bg are the excitation temperature and the cos-mic microwave background temperatures described as inthe Equation 1. For calculation of (cid:82) T mb d v , we used thearea under the fitted gaussian function. Since most ofthe spectra have a shape of the Gaussian profile, the dif-ference between the area under the fitted Gaussian func-tion and the integrated T mb within the velocity range of v cen ± (cid:52) v/
2, where v cen and (cid:52) v are the central velocityand linewidth of Gaussian fitting results, respectively, isfound to be less than 5%. N H from N C O is calculated with the conversion fac-tor of N H /N CO = 1 . × (Pineda et al. 2010) andthe abundance ratios of CO / CO = 69 (Wilson 1999)and CO / C O = 5 . column density distribution is shown in thepanel (b) of Figure 6. RAO FUNS I. L1478 in the California MC h m s s R.A. (J2000)+37 ◦ D ec . ( J ) C1C2C3C4
Filament 8 region
F8-C1F8-C2F8-C3 − − − )0 . . . . T ∗ A ( K ) F8-C4
Figure 5.
Same as Figure 4for Filament 8.
Table 2.
Information of the Identified Dense CoresCore ID Position Size PA V peak (N H + ) ∆ V (N H + ) M obs α vir ∗ IR Class † RA Dec Major Minor(hh:mm:ss) (dd:mm:ss) (pc) (pc) (deg.) (km s − ) (km s − ) ( M (cid:12) )Filament 4 regionF4-C1 04:25:03.21 +37:16:14.34 0.39 0.21 126 -1.78 0.35 1.42 1.35 Class 0F4-C2 04:25:13.10 +37:12:13.32 0.23 0.16 15 -2.47 0.51 0.62 2.88F4-C3 04:25:36.96 +37:07:16.54 0.32 0.16 111 -1.85 0.72 1.87 1.86 Class 0/IFilament 5 regionF5-C1 04:26:59.95 +36:56:26.66 0.21 0.13 54 -1.92 0.21 0.33 2.45Filament 8 regionF8-C1 04:21:13.02 +37:37:23.74 0.26 0.17 1 -0.50 0.29 1.55 0.77F8-C2 04:21:16.44 +37:33:39.67 0.23 0.19 14 -0.32 0.49 1.80 1.04F8-C3 04:21:17.10 +37:35:39.73 0.26 0.16 75 -0.41 0.39 0.53 2.74F8-C4 04:21:37.71 +37:34:11.82 0.38 0.34 153 -1.24 0.68 2.98 1.63 Class II ∗ Virial parameter, α vir , is given by the ratio of M vir /M obs . † F4-C1, F4-C3, and F8-C4 are identified by Broekhoven-Fiene et al. (2018) based on the
Spitzer and
Herschel and itsclassification given here is based on its IR spectral slope. F4-C3 and F8-C4 are detected in both of
Spitzer and
Herschel ,but F4-C1 is only detected in
Herschel
PACS. Chung et al.
F1 F2F3F4F5 F6F7F8F9F10 (a) ◦ ◦ RA (J2000) D ec ( J ) N H ( c m − ) F1 F2F3F4F5 F6F7F8F9F10 (b) V p e a k ( k m s − ) F1 F2F3F4F5 F6F7F8F9F10 (c) − − ∆ V ( k m s − ) F1 F2F3F4F5 F6F7F8F9F10 (d)
Figure 6. (a) Locations of ten identified filaments (skeletons) on top of the integrated intensity map of C O (the same asthe contour map in the bottom panel of Figure 2; contours are 2, 3, 6, 9, · · · , × σ in K km s − ). (b) - (d) H columndensity, velocity field, and linewidths maps of each filament. V peak and ∆ V (linewidth) are derived quantities by Gaussianfitting method. Small offset is given to the original position of each filament to avoid spatial overlaps and distinguish from eachother. RAO FUNS I. L1478 in the California MC N HerschelH2 (10 cm − )110100 N C O H ( c m − ) Figure 7.
Comparison of H column densities from C Oand
Herschel data. The solid thin gray line superimposedindicates where N C OH and N Herschel H are identical, and thetwo dashed lines show where the ratio of N C OH /N Herschel H is 10 and 0.1. The thick solid line presents the least squaresfit result. We also estimated the H column density from Herschel data to check whether the value derived withC O is reliable or not. Firstly, 250 µ m, 350 µ m, and500 µ m Herschel data were convolved to the 44 (cid:48)(cid:48) pixelsize and then co-aligned on the TRAO C O and COpixel-grid. Secondly, we derived spectral energy distri-bution (SED) fits from 250 µ m, 350 µ m, and 500 µ mdata for each pixel position with simple dust emissionmodel, F ν = κ ν × B ( ν, T ) × column density. A dustopacity law of κ ν = 0 . ν/ β cm g − is as-sumed with fixed dust emissivity index β of 2 (Draine& Lee 1984; Schnee et al. 2010) and the standard meanmolecular weight, µ , of 2.86 is used for the calculationof H column density (Kauffmann et al. 2008) .The measured H column densities from C O and
Herschel data are compared in Figure 7. It appearsthat N C OH is smaller than N Herschel H by a factor of ∼ N C OH to be smaller than N HerschelH . The spec-tra with multiple velocity components are presented inFigures 2and the Appendix (Figure 14 and 15), and mul-tiple velocity components of filaments are overlappedin the plane of sky. In this area, H column den-sity from C O is derived separately for each filamentcomponents, while N H from Herschel dust emissionis summed for the different filaments. Hence N C OH becomes less than N HerschelH . We should also add theother well known facts the CO depletion and dissocia-tion that can cause differences between H column den-sities derived with the two methods. At high densities of n H (cid:38) cm − and low dust temperatures of < N C OH and N HerschelH can bealso affected by the use of some uncertain parameters.The typical error of H column density measured from Herschel data is a factor of 2, which is mainly caused bythe uncertainty of the dust opacity law. In case of N H from C O, it is much complicated because the conver-sion factor of CO-to-H and the ratios of CO / CO and CO / C O vary by a factor of up to 5 according to themetallicity, column density, and temperature gradients(see Pineda et al. 2010; Bolatto et al. 2013, and refer-ences therein). Hence, the uncertainty of N H measuredfrom C O is likely to be a factor of a few, and we canconclude that the resulted N C OH and N HerschelH is wellmatched with each other within a range of uncertainty.H column density measured with C O ranges2 − × cm − , and the H mass of each iden-tified filament is ∼
10 to 200 M (cid:12) . The estimated massof each filament is tabulated in Table 3.3.3.2. Length and Width
We estimated the filament’s length along the skeletonwithout any correction of the projection effect. Filamentskeletons are determined in the following procedures.First, we made moment 0 images for each filament, thenapplied
FilFinder which uses Medial Axis Transformmethod to the moment 0 images for each filament. Me-dial Axis Transform method gives a skeleton which isthe set of central pixels of inscribed circles having max-imum radius (Koch & Rosolowsky 2015). The skeletonsdetermined with this process are well matched with the skeleton obtained with
DisPerSE (DIScrete PERsis-tent Structures Extractor). The
DisPerSE algorithmfinds the skeleton by connecting the critical points ofmaxima and/or saddle points which has a zero gradientin the map and identifies skeletons of filaments (Sousbie2011). There is a slight deviation among the skeletonsobtained between two algorithms which is less than 2pixel size in most cases. The skeletons of identified fil-aments with their IDs are shown in the panel (a) ofFigure 6. Lengths of filaments in L1478 range from ∼ (cid:48) to 11 (cid:48) which correspond to about 0.4 pc to 1.4 pc at thedistance of 450 pc.The filament width is measured from the H columndensity map (see section 3.3.1and Figure 6) of each fil-2 Chung et al.
Table 3.
Physical Properties of FilamentsFil. ID V peak range ¯ σ tot L W M M line M critline |∇ V peak | cores YSOs(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)1 -1.7 to -1.2 0.18 0.45 0.10 6.9 15.3 14.8 ± ± ± ± ± ± ± · · · ± ± ± Note —Col.(1) Filament ID. Col.(2) The largest and smallest V peak in km s − . Col.(3) Averaged totalvelocity dispersion from the C O linewidths in km s − . Col.(4) Length of filament measured fromthe eastmost (or northmost) point to the westmost (or southmost) point of skeleton in pc. Col.(5)Filament width in pc, i.e., FWHM of radial profile of H column density. Col.(6) H mass of filamentin M (cid:12) . Col.(7) Mass per unit length of filament in M (cid:12) pc − . Col.(8) Effective critical mass perunit length of filament derived with the mean total velocity dispersion in M (cid:12) pc − (see Section 4.2).Col.(9) Mean velocity gradient of filament in km s − pc − . Col.(10) Number of dense cores identifiedwith N H + data. F1, F2, F3, F9, and F10 are not observed with N H + . F6 and F7 are observedwith N H + but no emission is detected at the rms level of 0.06 K[T ∗ A ]. Col.(11) YSOs identified with Spitzer and
Herschel (Broekhoven-Fiene et al. 2018). ament. We made radial profile of H column densityversus distance from the skeleton, applied gaussian fit-ting, and derived the FWHM as filament’s width.The resulted lengths and widths of filaments are listedin Table 3. As can be seen, the width ranges from ∼ Herschel images (Ar-zoumanian et al. 2011). But it should be consideredhere that our spatial resolution of C O is ∼ (cid:48)(cid:48) , corre-sponding 0.1 pc at the distance of 450 pc and thus ourresult on the width may be affected by our observingresolution in space.3.3.3. Mass per unit length
With the mass and length of filament, we calculatedthe mean of the mass per unit length, M line . We simplydivided the mass of filaments with its lengths given inSection 3.3.1 and 3.3.2and thus M line shown here is theaveraged mass per unit length of the filament. The massper unit length of the filaments in L1478 is estimated torange from ∼
20 to 150 M (cid:12) pc − .3.3.4. Global Velocity Field
Intensity weighted mean velocity (moment 1) and ve-locity dispersion (moment 2) maps are very useful forunderstanding of the global velocity properties of molec-ular clouds, but may not be appropriate to study thekinematics of filaments showing multiple velocity com-ponents due to their spatial overlaps. Hence, we carriedout Gaussian fitting for C O spectra to extract velocityinformation of the filaments.The global continuous velocity distribution and gra-dient in the filaments can be seen in the peak velocitydistribution map, panel (c) in Figure 6. In section 3.1,we mentioned that F1 to F5 are from one stem thoughthey are found as an independent leaf, and this is alsotrue for F6 to F8. Those filaments are identified as an in-dependent filament because they have a maximum of 2 σ higher than the saddle point and they are all connectedto each other. The peak velocity map shows that theyare spatially and kinematically continuous. Especially,it is shown that F6 and F7 are continuously connectedwith F8 in the velocity space as well as in the plane ofsky, indicating they are a clear hub-filaments structure,i.e., dense star-forming hubs with multiple filaments.Information of the velocity distribution of each fila-ment is tabulated in Table 3as quantities of V peak rangesand gradients ( | (cid:53) V peak | ) of filaments. Figure 8shows RAO FUNS I. L1478 in the California MC − − V p e a k ( k m / s ) F1 F6 L fil (pc)F7 F9 − − V p e a k ( k m / s ) F2 F3 L fil (pc)F8 F8-C1F8-C2F8-C3F8-C4
F100 . . . . fil (pc) − − V p e a k ( k m / s ) F4 F4-C1F4-C2F4-C3 . . . . fil (pc)F5 F5-C1
Figure 8.
Velocity structures along the filaments. L fil is calculated along the skeleton points ( § V peak from C O is presented with solid black dots. V peak of each core denoted here is derived from the averagedN H + (see Figure 4 and 5). The gray thick lines are drawn to highlight some different velocity components. The YSOs reportedby Broekhoven-Fiene et al. (2018) are presented with gray arrows in their positions along the skeleton. the peak velocity distribution along the skeleton of eachfilament. The slope (velocity gradient) is different fromfilament to filament, and seems to be intrinsic to eachfilament, although no inclination correction has been ap-plied. The variation of V peak ( (cid:52) V peak ) changes between ∼ − and 3 km s − for each filament. F5 and F8have the largest (cid:52) V peak of ∼ − and F7 has thesmallest (cid:52) V peak of 0.3 km s − . The mean velocity gradi-ent ( |(cid:53) V peak | ) ranges from about 0.6 to 2.6 km s − pc − and F3 shows the largest |(cid:53) V peak | of ∼ . − pc − .We can see two clear kinematic properties of filaments.The one is the coherence of velocities of filaments, i.e.,every filament, even F8 which has a hub-filaments shape,has continuous V peak along the skeletons. The other onewe notice is that some different velocity components ap-pear (drawn with thick gray lines for F2, F4, F5, F6 andF9). We will discuss more about the different velocitycomponents in Section 4.3.3.5. Nonthermal velocity dispersion
Nonthermal velocity dispersion ( σ NT ) can be calcu-lated as following: σ NT = ( σ − k B Tµm H ) / (4)where σ is velocity dispersion derived from the FWHM, k B is the Boltzmann constant, T is the gas temperature, µ is the mean molecular weight of C O and N H + , and m H is the mass of the atomic hydrogen.The FWHM for C O and N H + were obtained fromGaussian fit to those spectra. In case of N H + spectraseven hyperfine components were simultaneously fittedwith seven Gaussian forms at once using their line pa-rameters given by Caselli et al. (1995). The gas temper-ature is assumed to be the same as T ex which is derivedin Section 3.3.1.Figure 9 shows the nonthermal velocity dispersionsin our all identified filaments and dense cores. ExceptF8, σ NT /c s of every filament peaks at transonic regime( ∼ σ NT /c s between 0.5and 1.0, the portion of transonic nonthermal velocitydispersion is about the same as that of subsonic compo-4 Chung et al. σ N T / c s F1 F6 F7 F901234 σ N T / c s F2 F3 F8
F8-C1F8-C2F8-C3F8-C4
F100 . . . . fil (pc)01234 σ N T / c s F4 F4-C1F4-C2F4-C3 . . . fil (pc)F5 F5-C1
F1F2F3F4F5F6F7F8F90 1 2 3 σ NT /c s . . Figure 9.
Velocity dispersions in all identified filaments and dense cores.
Left panels:
Nonthermal velocity dispersions normal-ized by the local sound speed as a function of the position along each filament. The σ NT /c s derived from C O is denoted withsolid black dots. σ NT /c s derived from the averaged N H + spectrum of each dense core (see Figures 4 and 5) are presented withdifferent colors of squares. The positions of YSOs reported by Broekhoven-Fiene et al. (2018) are presented with gray arrows. Right panels:
Normalized histograms of σ NT /c s . The σ NT /c s derived from linewidths of C O and of N H + are presented withblue and red, respectively. In here, histograms of σ NT /c s derived from N H + are not from the averaged spectrum but fromevery detected N H + spectra (S/N > nents (47% each). F8 that has YSOs appears to have ∼
51% of spectra of supersonic velocity dispersions.Nonthermal velocity dispersions measured from thedense core tracer N H + (denoted with squares of var-ious colors) are smaller than those from C O, eithersubsonic or transonic. The histogram of σ NT /c s fromN H + (presented with red color) shows that σ NT /c s from N H + of F4 and F8 peaks at transonic regime(1 (cid:46) σ NT /c s (cid:46) O and in dense cores traced byN H + , but F5 and F8 have different distributions of σ C ONT and σ N H + NT .3.3.6. Chemical Evolution of Dense Cores
Integrated intensity maps of N H + , C O, and SOmolecular lines are presented in Figure 10 with the densecores identified in § H + is known to survive much longer at high density when CObecomes depleted in gas phase (e.g., Tafalla et al. 2006).As can be seen in the maps, N H + traces more com-pact regions than C O and SO. Chain-like structure ofcores can be seen in F8 and F4 regions. F8-C1, F8-C2, and F8-C3 and F4-C1, F4-C2, and F4-C3 stand inlines and their N H + emission contours show chain-likestructure which is, however, not clear in C O and SObecause of their possible depletion.In the F8 region (top panels), four dense cores areidentified. F8-C1 and -C3 show relatively weak emis-sions in the three molecular lines of N H + , C O, andSO. Meanwhile, F8-C2 has a SO condensation at thepeak position of N H + . F8-C4, which is the largest andbrightest core, contains a class II object. Hence, F8-C4can be the most evolved core in this filament and F8-C2seems to be relatively younger than F8-C4.In the F4 region (middle panel), three dense cores arefound. C O is extended from the southeast to north-west, but there is no N H + and SO emission toward RAO FUNS I. L1478 in the California MC C1 Filament 5 region
C1C2C3
Filament 4 region
C1C2C3C4
Filament 8 region [N H + ] [C O] [SO]
Figure 10.
Integrated intensity contours maps of N H + (left), C O (middle), and SO (right) on the Herschel 250 µ m backgroundimage. Contours are at every 3 σ interval from 3 σ . The N H + peak positions of dense cores identified are presented with redcrosses. northwest. F4-C3 is the largest core with the brightestand the centrally peaked N H + emission in the F4 re-gion, but C O and SO is not as peaked as N H + . Thisis the same for F4-C1 which has centrally peaked N H + but not C O and SO. On the contrary, F4-C2, whichhas the weakest N H + emission, has a similarly peakedin three lines. Hence, F4-C2 appears to be the youngestcore in the Filament 4 region.In F5 region (bottom panels), only one core is found.Peaks of C O and SO show slight offset from the N H + peak, and thus can be depleted. This indicates that F5-C1 is not a young but evolved core. On the other hand,at the southeastern region of F5-C1, where no N H + emission is detected, C O and SO emissions are clearlydetected indicating that this region is at an earlier stage.The chemical differences between cores which formin a filament indicate that they are at different stage of evolution. Lee & Myers (2011) carried out variousmolecular line observations toward several tens of star-less cores, and found that the column density increasesin a sequence of core evolution that the earliest stageis static cores, and the next is expanding and/or oscil-lating cores, and the most evolved one is contractingcores. To probe the relation of column density and theevolution status, we compared the H column densitybetween the cores. The youngest cores in F4 and F8 areF4-C2 and F8-C2, respectively, and they have lower H column densities than other dense cores, which agreesthe result of Lee & Myers (2011) that cores with higherH column density tend to be more evolved than others.However, the internal motions such as static, expand-ing and/or oscillating, and contracting motions in eachevolutionary stage are not clearly related with the evo-lutionary stage in our dense cores. In Filament 4, the6 Chung et al. youngest core F4-C2 doesn’t show clear infall signature,but the other cores of F4-C1 and -C3 show clear infallmotions indicating the contraction of cores. On thecontrary, in Filament 8, the most evolved core F8-C4doesn’t show any infall signature, but F8-C1 and -C2which are younger than F8-C4 show blue asymmetryshowing contraction. However, because of our smallnumber of dense cores, it is difficult to conclude thatthe internal motions of cores are not correlated with theevolutionary stage of cores. DISCUSSION4.1.
Fibers, building blocks of filaments?
Recent molecular line observations of filaments withhigher spatial and spectral resolutions lead us to con-firm the presence of fibers. Hacar et al. (2018) proposedthat all filaments are bundles of fibers which have a typ-ical length and width of ∼ ∼ V peak distribution along the skeletonappears to be coherent in each filament, but fiber-likevelocity structures can be seen in every filament, espe-cially in F2, F4, F5, and F6. F4, which has YSOs anddense cores, seems to have fibers weaving together atpositions of L fil ∼ L fil ∼ O spectra and the difference ofpeak velocities is ∼ . − . In F2, F6, and F9, thebraided fibers can be found, too. These three filamentsseem to be composed of two parallel fibers which meet each other at the center of the filament. F8 has hub-likemorphology, but its velocity structure also shows thepresence of fibers.What we can see with our data is still limited be-cause of its poor spatial resolution, but one noticeablething is that the filaments having YSOs and dense coresshow clear evidence for presence of fibers than otherfilaments do. F4 and F5 look like to have more fibersthan other filaments in L1478, and furthermore theyare gravitationally unstable and have YSOs and densecores. In other words, the peak velocity distributionsof filaments in L1478 indirectly indicate that filamentsare made of fibers and those filaments gravitationallycollapsing show denser and finer distribution of fibers.These results support the star formation scenario sug-gested by Hacar et al. (2018) that filaments are bundlesof velocity-coherent fibers and fibers are an arbiter ofstar formation in molecular clouds in a sense.4.2. Are the filaments in L1478 gravitationally bound?
The ubiquity of filaments indicates that the filament isan indispensable structure in the star formation process.Andr´e et al. (2010) proposed a core formation scenariobased on the
Herschel
Gould Belt Survey observationthat long and thin filaments are made first in the molec-ular clouds and then prestellar cores form by hierarchi-cal fragmentation of gravitationally unstable filaments.The equilibrium line mass or mass per unit length foran isothermal, unmagnitized filament in pressure equi-librium is M unmagline , eq = 2 c /G where c s is the isothermalsound speed (Ostriker 1964; Inutsuka & Miyama 1992,1997). The critical value at 10 K is ∼ M (cid:12) pc − ,and a filament with larger M line than M line , eq becomessupercritical and undergoes gravitational contraction.Filaments in L1478 have mass per unit length of ∼
20- 150 M (cid:12) pc − that is larger than the equilibrium massper unit lengths. This means that every filaments inL1478 is supercritical and will collapse gravitationally,and dense cores can be formed by gravitational frag-mentation along the filaments. However, in this case,nonthermal components such as turbulence motions aswell as magnetic field are not considered. Hence, wederived the effective critical mass per unit length whichincludes nonthermal motions, M critline = 2¯ σ /G , where¯ σ tot is the average total velocity dispersion of a filament(e.g., ? Peretto et al. 2014). The calculated M critline with¯ σ tot is tabulated in Table 3 and presented in panel (a) ofFigure 11 with thick gray bar. The ratio, M line /M critline ,is given in panel (b) of Figure 11.With M critline , all filaments in L1478 are close to criticaland F1, F3, and F9 are marginally critical. The three RAO FUNS I. L1478 in the California MC M li n e ( M (cid:12) p c − ) M critline (a)1 2 3 4 5 6 7 8 9 10Filament ID0246 M li n e / M c r i t li n e (b) Figure 11.
Criticality of observed filaments. (a) The massper unit length ( M line ) is presented with solid dots and the ef-fective critical mass per unit length derived with the average σ tot for each filament is presented with a gray bar. The graydashed line denotes the equilibrium value ( ∼ M (cid:12) pc − )for isothermal cylinder in pressure equilibrium at 10 K. Theerror bars indicate the factor of two uncertainties of M line .(b) Ratios of line mass to effective critical mass per unitlength. The gray dashed line indicates the line where M line and M critline are identical. filaments, F4, F5, and F8, are highly critical. This isconsistent with the fact that the high density tracer ofN H + is detected and YSOs are found in these filamentsonly (Broekhoven-Fiene et al. 2018).To see whether there are any inward motions from thegravitationally bound filaments to the dense core, wecheck the CS (2-1) and N H + (1-0) line profiles. Blueasymmetry in spectral profiles is generally used to iden-tify infall signatures of dense cores and molecular clouds(e.g., Leung & Brown 1977; Lee et al. 2001). In cen-trally concentrated cores, the foreground gas appears tobe absorbed in optically thick lines. If inward motionsprevail, the absorption of foreground gas becomes red-shifted and the optically thick emission line appears tobe brighter in blue peak. As shown in Figure 12, amongthe eight dense cores found from N H + data, F4-C1,F4-C3, F5-C1, and F8-C2 show the blue asymmetry inthe CS (2-1) line profile implying the existence of gasinfalling motions which is consistent with their statusof criticality.4.3. Do cores form by collisions of turbulent flows?
Turbulence motions in filaments and dense cores areimportant to the formation of filaments and dense cores.Numerical simulations of supersonic turbulence gavea result generating dense stuructures like sheets, fila-ments, and cores (e.g., Padoan et al. 2001a; Xiong et al. 2017; Haugbølle et al. 2018). Padoan et al. (2001a) pro-posed that filaments can form by collision of turbulentflows, and dissipation of turbulence makes the densecores. Following this colliding model, the filaments andthe dense cores may have supersonic and subsonic mo-tions, respectively. To probe this dense core formationscenario, we checked the kinematic properties of fila-ments and dense cores.First of all, as shown in Figure 8, the V peak distribu-tion between dense cores and surrounding material offilaments in F4, F5 and F8 is well matched with eachother. This can be seen more clearly in the left panelof Figure 13. Peak velocity and velocity dispersion ofdense cores and surrounding filament gas were derivedwith N H + and C O spectrum, respectively. In the leftpanel of Fig 8, the peak velocity of dense core is plottedwith that of surrounding filament gas. Every dense corehas identical peak velocity to that of filamentary mate-rial within the total velocity dispersion traced by C O.Hence, we can conclude that there is no systemic ve-locity drift between the dense core and the surroundingmaterial. This result is similar to the previous studies inwhich no systemic velocity difference is found betweenthe dense cores and surrounding gas (e.g., Kirk et al.2007; Hacar & Tafalla 2011; Punanova et al. 2018).The nonthermal velocity dispersions of filaments anddense cores are presented in sound speed ( c s ) scale inthe right panel of Figure 13. As shown in Fig 9, σ C ONT is larger than σ N H + NT for every dense cores. There arefour dense cores which have transonic σ N H + NT (F4-C2,F4-C3, F8-C2, and F8-C4), and two dense cores whichhave subsonic σ N H + NT (F5-C1 and F8-C1). The othertwo dense cores (F4-C1 and F8-C3) have N H + velocitydispersions in the borderline between the transonic andsubsonic.One noticeable thing is that the relationships of σ C ONT and σ N H + NT are different from filament to filament, i.e.,dense cores which are surrounded with materials in su-personic motion in F8 still have subsonic or transonicmotions, while velocity dispersion of dense cores in F4becomes as large as that of their surrounding filament(F4). This different motion between dense cores andsurrounding material in filaments can be seen in thehistogram of Figure 9, too. The nonthermal velocitydistribution of N H + and C O in F4 peaks at tran-sonic, but that of N H + in F8 peaks at transonic whilethat of C O peaks at supersonic. Filament 5 has onlyone dense core, F5-C1, which appears to be subsonic σ N H + NT < . c s while the less dense material is transonic σ C ONT ∼ . c s .8 Chung et al. − − − . . . . T ∗ A ( K ) F4-C1(0 , 0 ) CS(2-1)N H + (102-012) × − − − . . . . , -44 ) × − − − . . . T ∗ A ( K ) F4-C3(0 , 0 ) × − − − . . . , -44 ) × − − − . . . T ∗ A ( K ) F5-C1(0 , 0 ) × − − − . . . , -44 ) × − − )0 . . . T ∗ A ( K ) F8-C1(0 , 0 ) × − − − )0 . . . . , 0 ) × Figure 12.
Infall asymmetric profiles toward dense cores in the supercritical filaments F4, F5, and F8. Averaged CS(2-1) andN H + (102-012) spectra toward dense cores are drawn with black and blue histograms, respectively. Offsets from the core centerare given in the top left corner of each panel. The blue solid curve and gray dashed vertical lines are the hyperfine fitting lineand central velocity of cores, respectively. RAO FUNS I. L1478 in the California MC -2.5 -2.0 -1.5 -1.0 -0.5 V C Opeak (km s − )-2.5-2.0-1.5-1.0-0.5 V N H + p e a k ( k m s − ) F5-C1F4-C1F4-C2 F4-C3 F8-C1F8-C2 F8-C3F8-C4 σ C ONT /c s σ N H + N T / c s F5-C1F4-C1F4-C2F4-C3F8-C1 F8-C2F8-C3F8-C4subsonictransonicsupersonic transonic supersonic
Figure 13.
Left:
Peak velocity of dense cores traced by N H + and surrounding materials of filaments traced by C O. Themean error in V peak is about 0.02 km s − for both N H + and C O, which is comparable or smaller than the point size. Thegray dashed and dotted lines indicate the identical line and its displaced line by the sound speed (about 0.19 km s − at 10 K). Right:
Nonthermal velocity dispersion of dense cores traced by N H + and surrounding materials of filaments traced by C O.The gray dashed line indicates that σ N H + NT and σ C ONT are identical. Cores in F4, F5, and F8 are presented with blue circles,green triangles, and red squares, respectively. Chung et al.
The difference between F4 and F8 implies that thedense cores in F4 and F8 can be formed by differentprocesses. The two filaments are both supercriticaland their velocity fields indicate mass flows along thefilament in case of F4 and from filaments to hub incase of F8. However, their morphologies and dynamicalproperties are totally different. F4 has a single, longfilament shape, but F8 has a hub-like morphology. Thedense cores and surrounding material of F4 are bothsubsonic or transonic while σ C ONT of F8 are transonicor supersonic but σ N H + NT are still subsonic or transonic.Hence, dense cores in F8 may be formed by collisionof turbulent flows (Padoan et al. 2001b), while densecores in F4 may form with the filaments. In case ofF5, its dynamical property is similar to F8, but it hasa shape of a long filament like F4. However, if we fo-cus on the region around F5-C1, small filaments can beseen around the bright hub-like structure by F5-C1 (seethe bottom left panel in Figure 10). The peak velocitydistribution shown in Figure 6 is similar to that of F8,i.e., it has multiple velocity gradients along various di-rections although its velocity difference is not as largeas that of F8. To confirm that F5 and F5-C1 form viathe similar mechanism as F8 and its dense cores, moreobservations with higher spatial resolution are necessary. SUMMARY AND CONCLUSIONWe present dynamical and chemical properties of ∼ O(1 −
0) as a tracerof filaments and N H + (1 −
0) as a dense core tracer.SO(32 −
21) and CS(2 −
1) molecular lines are observed toinvestigate kinematics and chemical evolution of densecores as well as N H + . We mapped ∼ O and CO withthe noise levels of ∼ ∗ A ], and the velocity resolu-tion of 0.1 km s − for both lines. The N H + (1-0) lineobservations were made over an area of ∼
480 square ar-cminute, where C O is bright, with the noise level of (cid:46) .
06 K[T ∗ A ], and the velocity resolution of 0 .
06 km s − .SO and CS molecular lines are also observed over ∼ (cid:46) . ∗ A ] and 0 .
06 km s − , respec-tively.The main results and conclusions are :1. From this data, ten filaments are identified withthe dendrogram technique. The basic propertiesof filaments such as length, width, mass, mass perunit length, and mean velocity gradient are de-rived. We applied the FellWalker algorithm to identify dense cores to the N H + integrated in-tensity image, and found 8 dense cores in threefilaments among the 10 identified filaments.2. Considering observed mass per unit length and theeffective critical mass per unit length for the fil-aments, we found that three filaments (F4, F5,and F8) among the 10 filaments are supercriti-cal. These three filaments are found to have densecores, and young stellar objects are also reportedto be embedded in F4 and F8. In the supercriti-cal filaments F4, F5 and F8, infall signatures areseen toward the dense cores. From the observa-tional results, we conclude that three supercriticalfilaments are gravitationally unstable and contin-uously contracting.3. Every filament shows coherent velocity field. Mul-tiple and grouped velocity components like fibersare frequently found in filaments. Nonthermal ve-locity dispersions derived with C O and N H + indicate that dense cores are subsonic or transonicwhile the surrounding gases are transonic or su-personic. However, the distributions of velocitydispersion are found to be different from filamentto filament, i.e., the nonthermal velocity disper-sions of dense cores in F4 are similar to those ofsurrounding material while dense cores in F8 aresubsonic or transonic in the transonic or super-sonic surrounding filament material. We proposethat the formation process of cores and filamentscan be different for their morphologies and envi-ronments.Our further forthcoming studies on the physical prop-erties of filaments and dense cores in different star-forming environments will be more helpful to under-stand how the filaments and dense cores form.We appreciate to the referee and the editor for thevaluable comments and suggestions. We thank J. Mon-tillaud for the valuable discussion and comments. Thisresearch was supported by Basic Science Research Pro-gram through the National Research Foundation ofKorea (NRF) funded by the Ministry of Education, Sci-ence and Technology (NRF-2016R1A2B4012593). MTacknowledges partial support from grant AYA2016. Software: astrodendro (Rosolowsky 2008), Astropy(http://dx.doi.org/10.1051/0004-6361/201322068), Dis-PerSE (Sousbie 2011), FellWalker (Berry 2013), Fil-Finder (Koch & Rosolowsky 2016; https://github.com/e-
RAO FUNS I. L1478 in the California MC
Andr´e, P., Men’shchikov, A., Bontemps, S., et al. 2010,A&A, 518, L102Arzoumanian, D., Andr´e, P., Didelon, P., et al. 2011, A&A,529, L6Arzoumanian, D., Andr´e, P., K¨onyves, V., et al. 2019,A&A, 621, A42Baug, T., Dewangan, L. K., Ojha, D. K., et al. 2018, ApJ,852, 119Berry, D. S. 2015, Astronomy and Computing, 10, 22Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARA&A,51, 207Broekhoven-Fiene, H., Matthews, B. C., Harvey, P. M.,et al. 2014, ApJ, 786, 37Broekhoven-Fiene, H., Matthews, B. C., Harvey, P., et al.2018, ApJ, 852, 73Caselli, P., Myers, P. C., & Thaddeus, P. 1995,Astrophysical Journal Letters v.455, 455, L77Caselli, P., Walmsley, C. M., Tafalla, M., Dore, L., &Myers, P. C. 1999, ApJ, 523, L165Caselli, P., Walmsley, C. M., Zucconi, A., et al. 2002,arXiv, 344Clarke, S. D., Whitworth, A. P., Spowage, R. L., et al.2018, MNRAS, 479, 1722Dame, T. M., Hartmann, D., & Thaddeus, P. 2001, ApJ,547, 792Dhabal, A., Mundy, L. G., Rizzo, M. J., Storm, S., &Teuben, P. 2018, arXiv, 169Di Francesco, J., Evans, N. J. I., Caselli, P., et al. 2007, inProtostars and Planets V, 17–32Draine, B. T., & Lee, H. M. 1984, ApJ, 285, 89Federrath, C. 2016, MNRAS, 457, 375Frerking, M. A., Langer, W. D., & Wilson, R. W. 1982,ApJ, 262, 590Garden, R. P., Hayashi, M., Hasegawa, T., Gatley, I., &Kaifu, N. 1991, ApJ, 374, 540Hacar, A., Kainulainen, J., Tafalla, M., Beuther, H., &Alves, J. 2016, A&A, 587, A97Hacar, A., & Tafalla, M. 2011, A&A, 533, A34Hacar, A., Tafalla, M., & Alves, J. 2017, A&A, 606, A123Hacar, A., Tafalla, M., Forbrich, J., et al. 2018, A&A, A77Hacar, A., Tafalla, M., Kauffmann, J., & Kov´acs, A. 2013,A&A, 554, A55Harvey, P. M., Fallscheer, C., Ginsburg, A., et al. 2013,ApJ, 764, 133 Haugbølle, T., Padoan, P., & Nordlund, ˚A. 2018, ApJ, 854,35Imara, N., Lada, C., Lewis, J., et al. 2017, ApJ, 840, 119Inutsuka, S.-i., & Miyama, S. M. 1992, ApJ, 388, 392—. 1997, ApJ, 480, 681Johnstone, D., Rosolowsky, E., Tafalla, M., & Kirk, H.2010, ApJ, 711, 655Kauffmann, J., Bertoldi, F., Bourke, T. L., Evans, N. J. I.,& Lee, C. W. 2008, A&A, 487, 993Kirk, H., Johnstone, D., & Tafalla, M. 2007, ApJ, 668, 1042Kirk, H., Myers, P. C., Bourke, T. L., et al. 2013, ApJ, 766,115Kleinmann, S. G., Lysaght, M. G., Pughe, W. L., et al.1994, Experimental Astronomy, 3, 65Koch, E. W., & Rosolowsky, E. W. 2015, MNRAS, 452,3435K¨onyves, V., Andr´e, P., Men’shchikov, A., et al. 2015,A&A, 584, A91Lada, C. J., Lombardi, M., & Alves, J. F. 2009, ApJ, 703,52Lee, C. W., & Myers, P. C. 2011, ApJ, 734, 60Lee, C. W., Myers, P. C., & Tafalla, M. 2001, ApJS, 136,703Leung, C. M., & Brown, R. L. 1977, ApJ, 214, L73Liu, T., Li, P. S., Juvela, M., et al. 2018, ApJ, 859, 151MacLaren, I., Richardson, K. M., & Wolfendale, A. W.1988, ApJ, 333, 821Marsh, K. A., Kirk, J. M., Andr´e, P., et al. 2016, MNRAS,459, 342Maureira, M. J., Arce, H. G., Offner, S. S. R., et al. 2017,ApJ, 849, 89M¨uller, H. S. P., Thorwirth, S., Roth, D. A., &Winnewisser, G. 2001, A&A, 370, L49Myers, P. C. 2009, ApJ, 700, 1609Ostriker, J. 1964, ApJ, 140, 1056Padoan, P., Juvela, M., Goodman, A. A., & Nordlund, ˚A.2001a, ApJ, 553, 227Padoan, P., Nordlund, ˚A., R¨ognvaldsson, ¨O. E., &Goodman, A. 2001b, From Darkness to Light: Origin andEvolution of Young Stellar Clusters, 243, 279Palmeirim, P., Andr´e, P., Kirk, J., et al. 2013, A&A, 550,A38Pattle, K., Ward-Thompson, D., Kirk, J. M., et al. 2015,arXiv, 1094 Chung et al.
Pattle, K., Ward-Thompson, D., Berry, D., et al. 2017,ApJ, 846, 122Peretto, N., Fuller, G. A., Andr´e, P., et al. 2014, A&A, 561,A83Pineda, J. E., Caselli, P., & Goodman, A. A. 2008, ApJ,679, 481Pineda, J. L., Goldsmith, P. F., Chapman, N., et al. 2010,ApJ, 721, 686Punanova, A., Caselli, P., Pineda, J. E., et al. 2018, arXiv,A27Rosolowsky, E. W., Pineda, J. E., Kauffmann, J., &Goodman, A. A. 2008, ApJ, 679, 1338 Sanhueza, P., Jackson, J. M., Foster, J. B., et al. 2012,ApJ, 756, 60Schnee, S., Enoch, M., Noriega-Crespo, A., et al. 2010,ApJ, 708, 127Sousbie, T. 2011, MNRAS, 414, 350Spezzano, S., Bizzocchi, L., Caselli, P., Harju, J., &Br¨unken, S. 2016, A&A, 592, L11Tafalla, M., Myers, P. C., Caselli, P., & Walmsley, C. M.2004, A&A, 416, 191Tafalla, M., Santiago-Garc´ıa, J., Myers, P. C., et al. 2006,A&A, 455, 577Wilson, T. L. 1999, RPPh, 62, 143Xiong, F., Chen, X., Yang, J., et al. 2017, ApJ, 838, 49Yuan, J., Li, J.-Z., Wu, Y., et al. 2018, ApJ, 852, 12
RAO FUNS I. L1478 in the California MC astrodendro
Python package to identify filaments with C O datacube. The first step of astrodendro algorithm is to make grids of the PPV (Position Position Velocity) space and finds local maxima in every grid. Wegave initial parameters of 4 pixels along RA × × σ is used in this step), the structure is identified asan independent structure. But if the difference is less than 2 σ , then the local maximum point is rejected and mergedinto the other structure. There is another parameter to assign the structure as an independent structure, the numberof associated pixels. For the structures with associated number of pixels less than 5, we discarded the structures. Themerging of structures stop when they meet a neighboring structure or a given minimum intensity (1 σ ).Figure 14 shows the tree diagram of Filaments 6, 7, and 8, and sample spectra with their Gaussian fitting results.Panel (a) shows that F6 consists with three leaves (S1, S2, and S3) and two branches (S4 and S5). The leaf S1 includesthe local maximum pixel of F6, merges with S2 which is another leaf from neighboring local maximum, becomes thebranch S4, merges again with a leaf S3, and be the branch S5. In panel (b) and (c), the spatial and spectral structuresof F6 can be seen. Again, S1 leaf which has the local maximum pixel (colored with pink) is in the middle of S4 (green)and S5 (yellow), S4 (green contour) encloses S1 (pink) and S2 (red), and S5 (yellow) surrounds S3 (orange) and S4(green). Likewise, in panel (c), the spectral components of S1 (colored with pink) is surrounded by S4 (green) and S5(yellow). By checking those components of structures, we confirmed that S5 is a single filament with coherent velocitystructure and named F6.Panel (d) shows the spectra in the blue box of panel (b) which is the junction of F7 and F8. F7 looks to be connectedwith F8 in the Herschel dust map and the integrated intensity images of CO and C O (see Figure 2). However,double peak C O spectra can be seen in the two top left boxes of panel (d), and the lower velocity peak componentsare connected with northern spectra (F8) and the higher velocity peak components are linked with southern spectra(F7). Hence, F7 is supposed to be an independent filament though its northern part seems to merge with F8 gradually.The blue and cyan histograms are the resulted structures from dendrogram and they are identified as an independentfilaments.The results shown above confirm that dendrogram represents well the hierarchical structures of 3-dimensional data.However, the structures (datacube given by dendrogram) do not include whole spectra of the structure as shown withcolored histograms in the panel (c) and (d) of Figure 14. Hence, we can not use the datacubes of structures givenby dendrogram but carried out gaussian fitting on C O spectra to derive the physical properties of filaments such asdistribution of peak velocity, velocity dispersion, and mass of filaments (Section 3.3). Gaussian fitting is performedautomatically with a Python code and the datacube resulted by dendrogram is used for its initial guess. Though thedatacube from dendrogram do not fully cover the spectral components, it always contains the peak channel which canbe used as a initial guess of central velocity and velocity dispersion for gaussian fitting. Hence, we used the velocityof the channel having maximum intensity at each position and the widths from the datacube resulted by dendrogramalgorithm. The fitting results are overlaid on the spectra in Figure 14. C O spectra with the gaussian fitting resultsat the overlapped positions of filaments are presented in Figure 15. The spectra have clear double peak componentsand each peaks are assigned to different structures indicating that the filaments identified by dendrogram are distinctstructures.4
Chung et al. (b)F6F7F8 F l u x F8F8F7F7F6F6
S1S2S3 S4S5 (a)
Structure (c) − − . . . − . . (d) Figure 14.
Hierarchical structure of F6, F7, and F8. (a) : Tree diagram of F6, F7, and F8. Leaves (that have no sub-structure, e.g., S1, S2, and S3) and branches (have sub-structures, e.g., S4 and S5) of F6 are presented with different colors. (b) : Contours of leaves and branches derived by dendrogram technique are overlaied on the C O moment 0 image to showthe spatial distribution of the structures. The same color code with panel (a) is used and the grayscale of integrated intensitymap is identical to that of Figure 3. (c) and (d) : C O spectra of the red and blue box regions of panel (b), respectively. Thex- and y- axis are the LSR velocity in km s − and the antenna temperature in kelvin, respectively. The observed spectra aregiven with black histogram, and those of structures (F7, F8, and sub-structures of F6) are presented with the same color codeof panel (a). The gaussian fitting results of filaments are overlaid. The dashed line is 3 σ level. RAO FUNS I. L1478 in the California MC F5 F4 F3 F2F1 F10 F9 F8 F6F7 − − − V LSR (km s − )0 . . T ∗ A ( K ) F1F1 F5F5 − − − . . . − − − . . . − − . . . Figure 15. C O spectra of filaments and the gaussian fitting results at the areas where filaments are overlapped. (center) :Filaments overlaid on the C O integrated intensity map. The squares colored with orange, pink, red, and cyan represent thelocations of the spectra shown. (top left) : spectra at the red square of the central image where F2 and F9 meet together. Thex- and y- axis of the spectra are the LSR velocity in km s −1