EDP Sciences
Free Access
Issue
A&A
Volume 547, November 2012
Article Number A12
Number of page(s) 11
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201118677
Published online 19 October 2012

© ESO, 2012

1. Introduction

Because of its proximity to the Milky Way, the Magellanic Cloud system (MCS) forms an ideal laboratory for studing the structure formation within the local universe in great detail. The more massive Milky Way galaxy accrete the MCS and tidal forces and perhaps drag separate gas from the stellar bodies. High-sensitivity optical studies disclose even the faintest stellar populations of evolved stars and allow to search for evidence of the stellar-gas-feedback processes triggered by the interaction with the Milky Way halo and/or its gravitational field (Indu & Subramaniam 2011). Both stellar distributions are highly concentrated, so only a fraction also populate the Magellanic Bridge (Putman et al. 2003; Irwin et al. 1990). So far the prominent Hstructures denoted as the Leading Arm (LA, Putman et al. 1998) and the Magellanic Stream (MS) have not detectable stellar counterpart (Wannier & Wrixon 1972; Putman et al. 2003; Brüns et al. 2005).

thumbnail Fig. 1

Left: the map shows the column density distribution of the H21-cm line emission across the velocity interval −80   km   s-1 ≤ vGSR ≤ 250   km   s-1. The complex Hstructure is produced by the superposition of the Milky Way galaxy emission, the LMC, and the LA. Right: the column density distribution of LA clouds across −80   km   s-1 ≤ vGSR ≤ 250   km   s-1 starting at a threshold of ΔTB = 200   mK (4-σ level of the GASS data). All the objects shown are cataloged and parametrized. Coherent structures like the Milky Way disk and LMC emission are according to the applied criteria “border” objects and not part of the catalog. The boxed and labeled areas enclose coherent structures in the data cube. Objects outside these areas also belong most likely to the LA. Note: white colors inside regions with disk emission denote column densities in excess of NHI = 9 × 1022   cm-2.

Open with DEXTER

Several investigations of the MS make use of specialized surveys (Brüns et al. 2005) or the Leiden-Argentine-Bonn all-sky survey (Nidever et al. 2008). Even though a wealth of information has been compiled today (see Nidever et al. 2008, and references therein for a recent review), it is still a matter of debate to what extent tidal or drag forces determine the structure formation of the MS and LA. Not detecting of stars in the MS argues for drag forces, decelerating only the viscous gaseous component, while the stellar component is unaffected and continues its orbital motion. The existence of the LA, however still favors strongly a tidal origin (Nidever et al. 2008).

In this paper we focus on the H21-cm emission distribution of the LA as observed with the Galactic All-Sky Survey (GASS, McClure-Griffiths et al. 2009; Kalberla et al. 2010). Assuming a linear distance of about 50 kpc from the Sun (Cioni et al. 2000), single–dish radio telescopes like the Parkes 64-m dish offer a linear resolution of about 200 pc at H21-cm wavelength. The high spectral resolution of state-of-the-art radio spectrometers not only allows determination of the bulk motion of individual clouds but also examination of well resolved Hline profiles. GASS for the first time makes fully sampled sensitive wide field images of the Hdistribution available.

For our study we extracted a huge portion of the southern sky from this data base to compile a comprehensive catalog of LA clouds and filaments (Sect. 2). To achieve this aim we developed an approach to differentiating between Milky Way gas and superposed (in space and frequency) MCS Hemission. Based on a newly developed extension of a standard image recognition algorithm, we decompose the Hdata without the necessity of Gaussian fits to the emission line profiles (Sect. 3). This allows us to compile a comprehensive view of the LA down to a limit TB = 200 mK. Using this catalog we investigate the velocity structure of the LA and a newly identified leading arm four feature as a whole (Sect. 4). The catalog comprises the basic physical parameters of all identified LA objects; this gives us the opportunity to analyze the occurrence of head-tail structures and their orientation with respect to the MCS (Sect. 5). We end the paper with a brief summary (Sect. 6).

thumbnail Fig. 2

Top: position (lMS) versus velocity (vGSR) diagram showing the maximum brightness temperature distribution across −42.5° ≤ bLMC ≤ 42.5° across the whole velocity range of interest (starting at TB = 200   mK up to 3 K on a linear scale). On the righthand side the LMC Hemission, while the Milky Way emission is visible towards the bottom. The narrowly confined high-velocity dispersion emission lines are associated with nearby galaxies covered by GASS. Obvious is a continuous velocity gradient of the Hemission above the Galactic plane from the LMC systemic velocity to vGSR = 0   km   s-1. The location of clouds belonging to the Wannier complexes WA, WB, WC, and WD are annotated. Bottom: the colored circles denote the location of the individual LA catalog objects while the gray-scale figure represents the “edge objects”, the Milky Way and the LMC. Each color marks an individual LA cloud complex (the ID of the catalog – available at the CDS – entry is given), aiming to demonstrate the coherent velocity structure of the complexes. The circle diameter scales according to . The regression line marks the mass–weighted velocity gradient with its LMC value of vGSR = 84.2 ± 7   km   s-1 demonstrating that the LA gas most likely is released from the LMC gaseous body.

Open with DEXTER

2. The data

We use GASS data, observed with the L-band multi-feed receiver at the Parkes 64 m telescope. Observations and initial data processing of the survey were described by McClure-Griffiths et al. (2009). We extracted data from the final data release (Kalberla et al. 2010) that were corrected for stray radiation, instrumental effects, and radio-frequency interference (RFI). In calculating FITS cubes we used the web service as provided by http://www.astro.uni-bonn.de/hisurvey/ but with two major extensions: 1) the positions were converted to the Magellanic Stream coordinate system as defined by Nidever et al. (2008); 2) the radial velocities were transformed into the galactic-standard-of-rest system (GSR) (Eq. (2) of de Heij et al. 2002) to quantify the motion of the MCS relative to the Milky Way galaxy. We consider the GSR system as best suited to evaluating the dynamics of the coherent MCS structures relative to the Milky Way galaxy. This is most useful for visualization; however, it does not allow ambiguities to be overcome when LA clouds are blended with Milky Way gas.

The GASS web server allows generating maps with different predefined angular resolutions (Kalberla et al. 2010). For our data analysis we chose two settings with effective beams of FWHM = 14.4 arcmin (4.8 arcmin per data cube pixel) and FWHM = 20.7 arcmin, resulting in average rms uncertainties of 50 mK and 25 mK, respectively, for channel maps at a spectral resolution of The smoothed data were used to initialize our source finding algorithm (Sect. 3.1), all further processing was done on the high resolution data base.

The whole field of interest covers 0° ≤ lLMC ≤ 70° and −42.5° ≤ bLMC ≤ 42.5°. This corresponds to 1.59 sr or 13% of the whole sky (Fig. 1 left). The radial velocity interval of interest covers −80   km   s-1 ≤ vGSR ≤ 250   km   s-1. According to van der Marel et al. (2002), the LMC velocity is vGSR = 84    ± 7   km   s-1. Inspecting Fig. 2 (top panel) shows that the negative velocity limit cannot be determined very well because of the confusion with the Milky Way galaxy emission. The positive velocity limit safely includes all MCS features above our 4-σ threshold of TB = 200 mK (Fig. 1 right).

3. Decomposing the LA emission

Our aim was to compile an object catalog for the LA area, containing all data that are necessary for any interpretation, such as position, velocity, brightness temperature, center of emission, geometrical center, column density, and velocity gradient. For an accurate parameterization we also decided to determine first moments in the position-velocity domain. This allows us to investigate the cloud’s morphology and quantify their shape, kinematics, and column density distribution asymetries in detail. The resulting catalog comprises emission features of all kind of structures, regardless of their physical origin.

A major task is to eliminate spurious features and to distinguish between sources belonging to LA and the Milky Way. Nidever et al. (2008) use a Gaussian decomposition for this purpose, which is a very elaborate enterprise. Intuitively, a parameterization in first moments, as mentioned above, also enable us to distinguish between different object classes according to their cataloged parameters, such as full area features (i.e. the Milky Way), “edge” objects (gas clouds touching the limits of the data cube either in position or velocity), small objects, and known galaxies. It is clear that the available source parameters must be complete enough to distinguish between different object classes.

3.1. ImageJ and source identification

The cloud catalog is automatically compiled by a newly developed source finding and parametrization program within the ImageJ enviroment. ImageJ is a scientific image processing software (Rasband 1997–2009) that is most commonly used in biology and medicine but also partly in astronomy. It is an open source software, and it can be found at http://rsbweb.nih.gov/ij/index.html.

The source-finding algorithm is designed to identify both spatially and kinetically coherent Hobjects above a given brightness temperature threshold. For each object some essential physical parameters are derived, as presented above.

3.1.1. Cleaning the catalog

First we clean our database for insignificant features below the instrumental noise, for RFI, and residual systematical baseline defects. We use the smoothed data with a doubled kernel size of 0.25°. All features below the threshold of TB = 200 mK can be excluded by this approach. Based on this threshold, we generated a mask for the high-resolution data cube.

The threshold mask isolates coherent structures and different individual objects. These structures need to be separated from other objects in phase space (spatially but also in velocity). This procedure attributes each pixel above the threshold to an associated object, permitting the calculation of the objects physical parameters.

The GASS beam FWHM (14.4 arcmin = 0.24°) corresponds to three pixels in the data cube. By this, any detected objects smaller than a single Parkes beam were excluded from further analysis. With this step we mitigate the imprint of RFI. We cross correlated our catalog with the HIPASS bight galaxy catalog (Koribalski et al. 2004). Within the area of interest, 361 galaxies are located. Most of them are automatically excluded from further consideration because their velocities are beyond the velocity range considered here. Only three residual galaxies were needed to be accounted manually.

Cold gas with an FWHM line width of up to three spectral channels (2.4 km s-1) are considered to trace mainly the Milky Way emission because a sufficient high gas pressure is needed to warrant cold gas of  ≲ 250 K in a two-phase medium (Wolfire et al. 1995). HVCs containing extremely cold dense cores will still be cataloged because next to the narrow Hemission the enveloping warm gas will also be detected in adjacent spectral channels. RFI events which are however, usually narrowly confined in frequency, will be rejected.

After such a selection, the LA catalog in its final version has 449 entries. In the following we show that these cataloged objects might safely be considered as belonging to the LA because of their coherent velocity structure. Towards the field of interest, next to the LA, the Wannier complexes are well known (Wannier & Wrixon 1972). These HVC complexes (annotated in Fig. 2 and discussed in Sect. 4) form coherent structures but are separated from most of the LA structures by their lower GSR velocities.

3.1.2. The Milky Way galaxy as an “edge” object

Based on the catalog it is feasible to generate data cubes showing different representations of the LA objects. The spacial distribution of all compiled objects is shown in Fig. 1(left panel). It includes objects that are boundary or edge objects of the data cube. These edge objects were excluded from further consideration (see Fig.1 right panel). The comparison of both panels make it evident that the Milky Way disk emission is successfully removed from the analyzed catalog. The Milky Way disk Hemission above the threshold of 200 mK forms a single broad and coherent structure which touches all boundaries of the whole data cube. By definition the Milky Way galaxy Hemission is classified as an “edge” object. Accordingly, when dropping the edge objects from the final catalog, the Milky Way emission is entirely removed.

It is a big advantage of the applied algorithm to distinguish easily between the LA and the Milky Way disk without the need to fit other parameters. However, Hobjects physically belonging to the MCS but apparently “connected” with the Hemission of the Milky Way galaxy will not be part of the final catalog. Here visual inspection might further improve the completeness of the catalog but could also introduce systematic biases.

3.2. The cataloged parameters

An extensive set of parameters is calculated for each cataloged object. The basic parameters are the position, the spatial and the kinematical centers, the intensity weighted center, the geometrical center, and the position of maximum brightness temperature. Further compiled parameters are the object extent in all three cube dimensions, the maximum brightness temperature, the maximum column density, the mean column density, the integrated brightness temperature, the phase space volume, the solid angle, the orientation angle, and the FWHM velocity width. The kinetic gas temperature is derived from the full-width-half maximum of the Hline of the peak brightness temperature position. Assuming a 50 kpc distance of the LA, the linear extent and the mass are calculated as well (see Table 2).

Figure 3 displays numerous features that appear well separated from the diffuse Galactic emission. Only clouds are considered that are unambiguously separated in phase space from the Milky Way. From a comparison of Fig. 1 (right) with Fig. 3 one might expect that some fraction of the total mass assigned to the leading arm might be missing. Correspondingly we find few objects at Magellanic longitude lMS ~ 33° visible in Fig. 2. We use the area filling factor for the affected region in velocity and position (the projected phase space) as a measure to estimate the mass deficit, as less than 5%.

3.3. Mass comparison

The masses of LA features derived here are about 50% higher as recognized previously. In Table 1 masses of LA substructures are compiled and compared with Brüns et al. (2005). The obvious differences are first because of different definition of the feature extents (in general in Brüns et al. 2005, the features are systematically smaller). In particular LA1.1 and LA1.3 are extended objects that are partly located outside the survey area of Brüns et al. (2005).

Table 1

Mass comparison of the specific LA areas indicated in Fig. 1 with Brüns et al. (2005).

Table 2

Excerpt of the LA catalog of objects available at the CDS.

thumbnail Fig. 3

This color-coded figure illustrates the applied method to differentiate between Milky Way, LMC, and LA Hemission. The cataloged objects show up with significant different GSR velocity distributions (color), allowing them to be separated from the neighboring Hemission within the data cube. Applying the strong criterion to separate between Milky Way and LA emission (Fig.1 right) we might miss coherent structures, i.e. lMS = 35° and bMS = 28°. These objects are continuously connected in Hemission with the Milky Way galaxy.

Open with DEXTER

thumbnail Fig. 4

Hbrightness temperature distribution of two HT clouds associated with the LA4 are displayed in gray-scale on a linear scale. Both clouds show a cometary shape in this velocity channel of 84 km   s-1. The contour lines mark Hgas line temperatures at 100, 200, and 400 mK. The closest bow-like object connected with the head of one HT cloud is at vGSR = 144   km   s-1, and all the more distant structures are at vGSR = 25   km   s-1. The solid line marks the orbital motion of the MCS, the long dashed line the lagging rotating halo, and the short dashed line the projection of the resulting velocity vector.

Open with DEXTER

thumbnail Fig. 5

Velocity gradients of the individual coherent structures (see Fig. 1). Top left: LA1.1 and LA1.2 are located close to the Magellanic Clouds. Their ensemble properties are unique, and they dislose a rising vGSR velocity with increasing lMS. Top right: LA1.3 shows the strongest velocity gradient, consistent with the impact hypothesis proposed by McClure-Griffiths et al. (2008). Middle panel: LA2 (left panel) and LA3 (right panel) are localized at large Magellanic stream longitudes. Bottom left: the newly classified feature LA4 shows a velocity gradient comparable to LA2 and LA3 despite its close location to the Magellanic Clouds. Bottom right: those cataloged objects not accounted to the coherent structures defined in Fig. 1 disclose the same overall velocity gradient implying their association with the LA.

Open with DEXTER

4. The velocity structure of the LA as a whole

We first study the velocity distribution of all LA objects together to investigate all the environmental and orbital parameters. To this aim we analyzed the radial velocity pattern of the cataloged objects in the GSR rest frame. Figure 3 gives an overview, while Fig. 2 shows details about the individual LA objects listed in the Appendix. To distinguish between LA features and Wannier complexes, we included positions of Wannier clouds (Wannier & Wrixon 1972) in the upper panel of Fig. 2 as given in the Wakker & van Woerden (1991) catalog. For lMS ≳ 60° WA and WB features cannot be distinguished easily from scattered LA features. LA2 and LA3, however, as is obvious from Fig. 3, stand out clearly. The complexes WC and WD are found at low GSR velocities, so separated well from LA clouds.

The diameters of the circles in the lower panel are proportional to the third root of the cloud masses and specific LA substructures are colored. All together there is an obvious relation between lMS and vGSR. This finding implies that the cataloged LA as a whole has a coherent and continuous velocity structure. The clouds that are the most distant from the LMC tend to have the lowest velocities. Gas within the immediate vicinity of the LMC apparently had less time to get decelerated by the halo medium.

At the location of the Galactic disk (around lMS = 30°, Fig. 2 bottom panel), there is a V-shaped underabundance of LA objects. We attribute this under-abundance to limitations of our approach to separate Milky Way Galaxy and LA at these positions. HI clouds in this region have been attributed to Galactic emission.

Connecting the location of the cataloged LA objects with their Hmass (assuming D = 50 kpc) discloses a remarkable trend (in particular for LA 2 and LA 3): the higher the mass of the objects, the higher their radial velocities (Fig. 2, bottom panel and Fig. 5 middle panel). This argues for drag which decelerates the low-mass diffuse objects significantly while the high mass objects keep most of their momentum. In the immediate vicinity of the LMC (lMS ≤ 25°, Fig. 5 top and bottom left panels), the higher mass objects do not behave any differently in velocity. Here, the low-mass objects are also observed at high vGSR radial velocities. Starting around lMS ≥ 35°, the higher mass objects reveal that the highest radial velocites and lower mass objects are all at lower velocities.

Performing a mass-weighted, least square approximation of the vGSR(lMS), we find the relation accounting for all LA objects: (1)The linear regression line is shown in Fig. 2 (bottom panel). Despite the high number of low-mass objects, the high-mass objects dominate the linear approximation of the overall velocity distribution of the LA. Considering the global motion of the LA, the velocity at the origin of the LA, with vGSR(0) = 84.2   km   s-1 is consistent with the value vGSR = 84    ±    7   km   s-1 as derived by van der Marel et al. (2002) from an ensemble of carbon stars of the LMC. This overall velocity structure of the Hgas suggests that the LA has its origin in the LMC rather than in the SMC (Putman et al. 1998). This is consistent with the morphological and positional coincidences of the LA1 feature with the LMC, as suggested by Nidever et al. (2008), who have used quite different methods for their analysis.

Quantitatively comparing the velocity structure of the gaseous objects of the LA with the N-body simulation of Connors et al. (2006) (their Fig. 9) – who attribute the LA to the SMC – shows that the simulated velocity distribution is in excess of about 200   km   s-1 from the observed one. Their LMC orbit is, however, consistent with the Hvelocity structure of the high-mass objects.

4.1. Velocity structure of LA1

We distinguished three LA1 subpopulations as shown in Fig. 1 (right panel). Towards a single filament of LA1.3 McClure-Griffiths et al. (2008) propose that this filament interacts with the Milky Way galaxy disk. Hmass–weighted linear fits the individual LA filaments underline the general trend of decreasing vGSR with increasing lMS. As is obvious in Fig. 5 (top right panel) LA1.3 shows the strongest velocity gradient of all LA filaments. Based on our statistical analysis of the LA1.3, the filament as a whole appears to be strongly decelerated consistent with McClure-Griffiths et al. (2008).

Considering the overall velocity gradient of the LA as an ensemble of objects (Fig. 2) and the large angular separation between the LA features (Fig. 1), we propose that the LA is most likely a huge coherent structure. LA1.3 appears to pass through sufficiently dense gas of the Milky Way galaxy. LA1.2 and LA1.1 are apparently located at larger z-distances from the Galactic plane. The same is true for LA2 and LA3, but now on the opposite side of the Galactic disk, as evident from Fig. 3.

These filaments must have already passed the equator of the Galactic coordinate system, but we do not know the galactocentric distance. In the case of an exponential radial density distribution of the Galactic disk (Kalberla & Kerp 2009), the strength of the interaction depends critically on the radial distance from the Galactic center. Our findings (see Fig. 2 and Sect. 4) are consistent with the hypothesis that LA2 and LA3 filaments probably passed through a thinner volume density portion at longer galactocentric distances than LA1.3. Using the Kalberla & Kerp (2009) model, we suggest for LA2 and LA3 a galactocentric distance of  ≥ 40 kpc where a density n ≤ 10-4   cm-3 is expected.

4.2. Age and size of the whole LA structure

Figure 1 (right panel) shows a surprisingly high degree of symmetry of the Hobject distribution. LA4 covers the same angular range in |bMS| but is located at the opposite side of the LMC as LA1. To scatter the Hgas across more than 60° on the sky, symmetrically distributed around the LMC orbit (bMS = 0°), one needs a fast acceleration mechanism to overcome the gravitational attraction of the MCS and to widely distribute the gas across tens of kpc. Assuming a total gravitational mass of MMCS = 2 × 1010   M (van der Marel et al. 2002) the escape velocity is vLMC,esc ≃ 280   km   s-1.

The high degree of symmetry in radial velocity between LA 1 and LA 4 implies that the whole LA is viewed nearly face on. Evaluating the ratio of the radial to tangential velocity vector component with the escape velocity |vLMC,tan| + |vLMC,rad| ≤ |vLMC,esc| (van der Marel et al. 2002, gives vLMC,rad = 84    ±    7   km   s-1 and vLMC,tan = 281    ±    41   km   s-1), and we can support this visual impression quantitatively. Almost all of the total velocity vector is in the tangential component. If we assume that we watch a nearly “face-on LA system” with an inclination against the sky of only 13.6°, we can estimate the minimum time needed for an LA cloud to travel from the LMC body to its current location. For this, we assume a) that the linear separation between LA 4 and LA1 is less than 52 kpc, and b) we neglect drag forces and assume a constant objects orbit velocity equal to the escape velocity. This yields t = 6 × 107 a. According to this picture LA1.3 has a radial separation from the Galactic center of 19.3 kpc, while LA4 lies at a galactocentric distance of 71.9 kpc. The LA1.3 distance estimate we derived is consistent with McClure-Griffiths et al. (2008), who determined R ~ 17 kpc via a kinematic distance estimate.

4.3. Physical conditions of the LA4 clouds

4.3.1. Evaluating the LA4 ambient volume density

An estimated galactocentric distance of 72 kpc for LA4 corresponds to a radial distance of 74 kpc and an altitude tan(20°) × 74 kpc  ≃ 26.5 kpc above the disk. Figure 5 (bottom left panel) suggests that even the faint and distant LA 4 clouds are “decelerated” by drag forces. This implies that even at such large distance above the disk a sufficient high volume density exists to cause drag. This hypothesis is supported by individual HT clouds associated with LA4. Figure 4 shows the H21-cm column density distribution of two clouds of LA4, where apparently both appear to be associated with a cometary shape. They appear to “travel” in parallel and consist of fast-moving heads and trailing tails, implying that gas is stripped off the clouds, heated up, and concentrated within the cloud’s slipstream. This gives us the chance to estimate the physical conditions at those long distances from the galaxy, so we derive from the angular extent of the head of 0.5°, with 74 kpc distance, NHI = 5.2    × 1019   cm-2, a volume density of nHI = 0.03   cm-3. The mass of the HT clouds is about M = 4.5 × 105   M. We estimate the kinetic temperature from the line width (FWHM = 6.6   km   s-1) to  1000 K. This yields a gas pressure of P = 60   K   cm-3 in the certainly partly ionized gas. To constrain the location of the HT features we assume that they move with the same velocity as the MCS. The most probable position along the line of sight would be that closest to the MCS orbit. We obtain a distance of  ~40 kpc with a halo density of nHalo = 3 × 10-3   cm-3 (Kalberla 2003). This density would be just high enough to cause significant drag forces (Quilis & Moore 2001). Figure 4 shows that the lMS = 23°, bMS = 30° cloud is positionally correlated with a bow shaped feature at vGSR = 144   km   s-1 (contour lines), suggestive of an interaction. Accordingly, the objects might interact with local density enhancements. An even larger bow-like emission, features are located ahead at velocities vGSR = 25   km   s-1.

The orientation of the projected HT velocity vector indicates its motion with respect to the ambient medium. The resulting vector is aligned with the tail of the HT feature. For this estimate we use the mean 3D velocity vector determined by Kallivayalil et al. (2006) with |v| = 378   km   s-1.

5. Head-tail clouds towards the LA

Head-tail clouds are asymmetric structures. Such asymmetries are usually determined by visual inspection, but such a description may contain personal biases. As an attempt to determine the asymmetry of these structures in an objective way, we included higher moments in our analysis. The skewness in particular appears to be an appropriate measure for head-tail clouds. A well known problem, however, is that higher moments are increasingly stronger affected by uncertainties. Unfortunately, the signal-to-noise ratio is low for most of our clouds. Skewness is then ill-defined. The column density of the tail is typically lower than that of the head. For a lower signal-to-noise ratio it gets increasingly harder to measure asymmetry since the column density of the tail may eventually be undetectable. We searched for another more robust measure of asymmetry using parameters that are less affected by noise. Using our catalog of LA clouds, it is feasible to systematically search for clouds that disclose a positional difference between the on the sky projected center of mass (CM) and the brightness temperature maximum (MI, max. intensity). Both parameters are not severely affected by noise but a displacement represents a measure for asymmetries.

In total we count 136 clouds with angular sizes in excess of 14.4 arcmin and with |CM−MI| ≥ 14.4 arcmin. This implies that about 30% of our cataloged objects can be considered morphologically as HT objects.

thumbnail Fig. 6

Ratio of as a measure of temperature and density gradients. For comparison of the linear scale we plotted the Parkes beam as a black circle in front of the LA 1.3 head. Cooler gas portions (larger ) and warmer (smaller ) with a surprisingly high degree of small-scale structure. The high column density objects are associated with cool cores (heads) localized in the direction of the orbital motion (velocity gradient).

Open with DEXTER

thumbnail Fig. 7

Ratio of TB and Tkin across the whole LA as shown in detail in Fig. 6. The arrows mark the orientation of the HT structures from the center of mass to the maximum brightness temperature.

Open with DEXTER

This rate is comparable to Putman et al. (2011) who have found that 35% of the HIPASS compact and semi-compact HVCs can be classified as HT clouds from their morphology. However, their 116 HT objects were selected from a total number of about 2000 HVCs on the whole southern sky. Our sample of 136 HT objects was derived from a region covering 1.6 sr, only about one quarter of the area studied by Putman et al. (2011). We conclude that the LA region has a significant excess of HT structures.

The fainter one of the two HT features shown in Fig. 4 has remained undetected so far.

HT HVCs are common structures of HVCs (Brüns et al. 2000; Putman et al. 2011). Their cometary shape implies a physical interaction of the HVC gas with the dynamical different ambient medium. Brüns et al. (2000) show that a linear relation exists between the peak column density of HVCs and the fraction of observed HT HVCs. This implies that the higher the peak column density, the higher the probability of identifing an HT structure. This finding suggests that the HVC gas is compressed at the head, the gas cools more efficiently due to the higher volume density, and this leads to lower kinetic temperatures (narrower line widths) and apparently higher Hbrightness temperatures. Only towards HVC 125+41-207 (Brüns et al. 2001) do we presently have evidence of this process in great detail. Brüns et al. (2001) show the proposed temperature structure using a Gaussian decomposition of the Effelsberg Hdata. On a statistical basis, the cometary shape on its own is no more than a morphological implication, nor does the linear correlation between peak column density and HT fraction directly disclose ram pressure as a physical process.

To search for physical indicators for ram pressure interaction with the ambient medium, we investigated the ratio between the brightness temperature (TB) and the kinetic temperature (Tkin). (2)here TS denotes the spin temperature, the Boltzmann population of the two energy levels, Δv is the FWHM of the Hemission line, and τ the opacity of the Hgas. TB can be considered accordingly as lower, while Tkin is an upper temperature limit of the HT cloud. The ratio of both temperatures is used as a marker to differentiate between high volume density portions of the clouds (TB dominant) and warmer low volume density portions (Tkin dominant). Using the LA catalog, we calculated maps of the temperature ratio across the whole LA region of interest. We focused in particular on LA 1.3, because of the compelling morphological argumentation of McClure-Griffiths et al. (2008). Figure 6 shows the temperature ratio. Obviously the HT cloud (lMS = 27°, bMS = −22°) shows a suggestive cometary shape and a “cool” head associated with a trailing “warm” tail. “Warm” rims are obviously common in LA 1.3 and consistently oriented towards the Milky Way disk, indicating physically the interaction of the LA gas with the Milky Way disk matter.

Extending this investigation to the whole field of interest (Fig. 7) discloses a surprising richness of temperature structure. All HT structures have warm rims. Objects with higher masses have a cold head embedded within warm gas, followed by an even warmer trailing gas component. The low-mass objects do not show cold condensations.

Finally we studied the sky parallel component of the vector connecting CM and MI and found that the high mass objects disclose a coherent velocity pattern (Fig. 7). The HT velocity vectors trace the orbital motion of the LA as a whole. In combination with the derived overall velocity gradient, the velocity segregation of the object masses and the warm rims of the HT objects, we conclude that the whole LA is decelerated by drag forces towards the Milky Way disk. Both LA 2 and LA 3 have already passed the Milky Way disk (see Fig. 3). LA 4 fits perfectly, spatially as well as kinematically, in the whole picture and extends the LA to a huge coherent structure across the southern hemisphere.

6. Summary and conclusions

We used the GASS data (McClure-Griffiths et al. 2009; Kalberla et al. 2010) to investigate the LA structure. The analyzed area covers 70° × 85° degrees. Using a sophisticated image characterization tool, it was feasible to differentiate between the diffuse Milky Way emission and the LA structures. In total we identified 449 objects above the 4-σ threshold of the original data (TB = 200   mK). The applied method not only allows to determine the location of the objects, but also compiles derived information such as radial velocity, phase-space volume, column density, geometric center, and temperature weighted center. This allowed us to perform an automated statistical analysis of the LA as a whole and individual substructures down to the scales of individual clouds (HT structures).

Next to the well studied LA1, LA2, and LA3, we discovered LA4, which is localized at high Magellanic latitudes but at comparable longitudes to LA1. This symmetry but also the associated velocity gradient, suggests that this feature belongs to the LA complex. LA4 on average contains less extended clouds. In Sect. 4 we argued for a distance between 40 and 70 kpc, which implies that these clouds could also have a lower average mass.

Averaging across the whole LA1 to LA4 cataloged clouds, we derived a clear negative velocity gradient (deceleration) with Magellanic stream longitude. Assuming D = 50 kpc, the mass weighted linear correlation yields vGSR = 84.2   km   s-1 as axis intercept. This value is identical within its uncertainties with the bulk velocity of the LMC derived by van der Marel et al. (2002) from the carbon star ensemble. According to this we support the hypothesis of Nidever et al. (2008) that the LA as a whole has its origin in the LMC.

The mass of the LA is about a factor of 50% higher than previously assumed (Brüns et al. 2005, at a distance of 50 kpc; Cioni et al. 2000). The main reason for this increase is that our survey covered a significantly larger region than is available for previous, more specialized investigations. Some objects, blended with Milky Way emission features, may be missing in our analysis. The LA mass of 38 × 106   M derived here is probably a lower limit only.

The velocity structure of individual LA filaments can be interpreted as prototypical for deceleration by drag forces. LA1.3 shows the strongest velocity gradient, consistent with the proposal of McClure-Griffiths et al. (2008). All coherent structures of the LA but LA1.1 and LA1.2 show a systematic deceleration towards the Galactic plane. We find an obvious difference in the mass vs. velocity distribution for LA1 and LA4 in comparison to LA2 and LA3. The last two show that the higher the mass, the higher the vGSR velocity. In the framework of drag models, these high-mass compact objects keep their momentum, while the low-mass, more diffuse clouds are significantly decelerated.

The apparent symmetry of the LA structure is a surprising finding. The whole structure covers nearly 65° across the whole sky. At a distance of 50 kpc, this corresponds to a linear scale of 52.5 kpc. Combining the radial velocity information of the Hdata with the proper motion (van der Marel et al. 2002), we estimate an inclination of only 13.6° against the plane of the sky. With the escape velocity for the LMC we can calculate that the LA structures are at least 6 × 107 a old. This nearly face-on geometry of the LA localized LA1.3 at a galactocentric distance of R ~ 19.3 kpc and LA4 at R ~ 71.9 kpc.

We find pronounced HT structures associated with LA4. Using a distance of 74 kpc, an estimate on the kinetic temperature by the Hline width, we estimate a pressure of P = 60   K   cm-3. According to Kalberla (2003) the ambient volume density with n ≃ 10-3   cm-3 is high enough to cause drag.

The analysis of the ratio between TB and Tkin implies that the gas of the objects is compressed towards the leading edge of the objects. Warm rims are at the head of the HT structures, suggesting gas compression and eventually an enhanced cooling. All massive LA objects show HT structures. The sky projected vector component between the location of their centers of mass relative to the maximum Hline intensities are all oriented parallel to the orbital motion of the MCS. In combination with the velocity segregation, the overall velocity gradient and the cool

leading rims of the HT objects, we conclude that the whole LA is a coherent structure decelerated by drag. The newly identified LA 4 fits in the whole structure and physical interpretation perfectly.

The emerging picture of a four-arm structure, with two symmetric arms approaching the Milky Way disk (LA1 and LA4) and two arms that have already passed the disk (LA2 and LA3) provides new constraints for simulations of the orbital history of the Magellanic Clouds. Diaz & Bekki (2011) present a new tidal model in which, for the first time, structures resemble a clear bifurcation of the LA. Concerning LA4, we would like to note that the existence of this feature was predicted by Besla et al. (2010), but their simulations fail to recover LA1. Hydrodynamic processes such as ram pressure are needed to shape the Magellanic Stream.

Acknowledgments

M.S.V. thanks the Hgroup of the Argelander-Institut für Astronomy for a stipend during the preparation phase of the manuscript. The J.K. and P.M.W.K. acknowledge the financial support from the Deutsche Forschungsgemeinschaft (DFG) under the reseach grants KE757/7-1, KE757/7-2, and KE757/9-1. The authors like to thank the anonymous referee for very helpful comments and suggestions.

References

All Tables

Table 1

Mass comparison of the specific LA areas indicated in Fig. 1 with Brüns et al. (2005).

Table 2

Excerpt of the LA catalog of objects available at the CDS.

All Figures

thumbnail Fig. 1

Left: the map shows the column density distribution of the H21-cm line emission across the velocity interval −80   km   s-1 ≤ vGSR ≤ 250   km   s-1. The complex Hstructure is produced by the superposition of the Milky Way galaxy emission, the LMC, and the LA. Right: the column density distribution of LA clouds across −80   km   s-1 ≤ vGSR ≤ 250   km   s-1 starting at a threshold of ΔTB = 200   mK (4-σ level of the GASS data). All the objects shown are cataloged and parametrized. Coherent structures like the Milky Way disk and LMC emission are according to the applied criteria “border” objects and not part of the catalog. The boxed and labeled areas enclose coherent structures in the data cube. Objects outside these areas also belong most likely to the LA. Note: white colors inside regions with disk emission denote column densities in excess of NHI = 9 × 1022   cm-2.

Open with DEXTER
In the text
thumbnail Fig. 2

Top: position (lMS) versus velocity (vGSR) diagram showing the maximum brightness temperature distribution across −42.5° ≤ bLMC ≤ 42.5° across the whole velocity range of interest (starting at TB = 200   mK up to 3 K on a linear scale). On the righthand side the LMC Hemission, while the Milky Way emission is visible towards the bottom. The narrowly confined high-velocity dispersion emission lines are associated with nearby galaxies covered by GASS. Obvious is a continuous velocity gradient of the Hemission above the Galactic plane from the LMC systemic velocity to vGSR = 0   km   s-1. The location of clouds belonging to the Wannier complexes WA, WB, WC, and WD are annotated. Bottom: the colored circles denote the location of the individual LA catalog objects while the gray-scale figure represents the “edge objects”, the Milky Way and the LMC. Each color marks an individual LA cloud complex (the ID of the catalog – available at the CDS – entry is given), aiming to demonstrate the coherent velocity structure of the complexes. The circle diameter scales according to . The regression line marks the mass–weighted velocity gradient with its LMC value of vGSR = 84.2 ± 7   km   s-1 demonstrating that the LA gas most likely is released from the LMC gaseous body.

Open with DEXTER
In the text
thumbnail Fig. 3

This color-coded figure illustrates the applied method to differentiate between Milky Way, LMC, and LA Hemission. The cataloged objects show up with significant different GSR velocity distributions (color), allowing them to be separated from the neighboring Hemission within the data cube. Applying the strong criterion to separate between Milky Way and LA emission (Fig.1 right) we might miss coherent structures, i.e. lMS = 35° and bMS = 28°. These objects are continuously connected in Hemission with the Milky Way galaxy.

Open with DEXTER
In the text
thumbnail Fig. 4

Hbrightness temperature distribution of two HT clouds associated with the LA4 are displayed in gray-scale on a linear scale. Both clouds show a cometary shape in this velocity channel of 84 km   s-1. The contour lines mark Hgas line temperatures at 100, 200, and 400 mK. The closest bow-like object connected with the head of one HT cloud is at vGSR = 144   km   s-1, and all the more distant structures are at vGSR = 25   km   s-1. The solid line marks the orbital motion of the MCS, the long dashed line the lagging rotating halo, and the short dashed line the projection of the resulting velocity vector.

Open with DEXTER
In the text
thumbnail Fig. 5

Velocity gradients of the individual coherent structures (see Fig. 1). Top left: LA1.1 and LA1.2 are located close to the Magellanic Clouds. Their ensemble properties are unique, and they dislose a rising vGSR velocity with increasing lMS. Top right: LA1.3 shows the strongest velocity gradient, consistent with the impact hypothesis proposed by McClure-Griffiths et al. (2008). Middle panel: LA2 (left panel) and LA3 (right panel) are localized at large Magellanic stream longitudes. Bottom left: the newly classified feature LA4 shows a velocity gradient comparable to LA2 and LA3 despite its close location to the Magellanic Clouds. Bottom right: those cataloged objects not accounted to the coherent structures defined in Fig. 1 disclose the same overall velocity gradient implying their association with the LA.

Open with DEXTER
In the text
thumbnail Fig. 6

Ratio of as a measure of temperature and density gradients. For comparison of the linear scale we plotted the Parkes beam as a black circle in front of the LA 1.3 head. Cooler gas portions (larger ) and warmer (smaller ) with a surprisingly high degree of small-scale structure. The high column density objects are associated with cool cores (heads) localized in the direction of the orbital motion (velocity gradient).

Open with DEXTER
In the text
thumbnail Fig. 7

Ratio of TB and Tkin across the whole LA as shown in detail in Fig. 6. The arrows mark the orientation of the HT structures from the center of mass to the maximum brightness temperature.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.