At a Crossroads: Stellar Streams in the South Galactic Cap

The Molonglo stream, detected at the $\approx 10\sigma$ level, is some $22^\circ$ long, extending from near the globular cluster NGC 288 across both the bright and faint branches of the Sagittarius stream. While pointing almost directly at NGC 288, the stream is at more than twice the 8.8 kpc distance of the cluster (Harris 1996). Moreover, the measured proper motion of NGC 288 (Dinescu et al. 1997) would predict a stream oriented more or less east-west, along the southern limit of the PS1 survey. Molonglo is therefore unlikely to be related to NGC 288.

In equatorial coordinates, the trajectory of Molonglo can be modeled to $\sigma =0\buildrel{\circ}\over{.} 1$ using

$$\begin{eqnarray}&&\alpha =345.017-0.5843\delta +0.0182{\delta }^{2}\end{eqnarray} \tag{ 2 }$$

over the stream extent of $-24\buildrel{\circ}\over{.} 5\lt \delta \lt -12^\circ$. The FWHM of the stream is about 30 arcmin. At a distance of 20 kpc, the length and breadth of the stream are therefore 7.7 kpc and 170 pc, respectively. The latter suggests a relatively small or low mass progenitor such as a globular cluster. A background-subtracted, color–magnitude Hess diagram is shown in Figure 3. While obviously contaminated by the Sagittarius stream, Molonglo appears to have a discernible subgiant branch and lower main sequence. As with other streams, shifting the filter brightward or faintward by more than half a magnitude causes the structure to effectively vanish. Estimating the distribution of foreground stars using a heavily smoothed version of Figure 1 and then simply counting stars along the stream within the $3\sigma$ color–magnitude envelope yields an estimated total of 311 ± 95 stars to $g=21.8$. Integrating over a globular cluster-like luminosity function then predicts a total of 3700 ± 1100 stars over the $22^\circ$ length of Molonglo. This is certainly comparable to modern day globular clusters and suggests that Molonglo might once have been a globular cluster itself.

Zoom In Zoom Out Reset image size

While stars in streams do not precisely follow the same orbits, or the orbits of their progenitors (Eyre & Binney 2011), the observed trajectories of streams can put some constraints on their orbit shapes. These in turn can be used to identify orbit families or possible progenitors. Starting from a designated fiducial point midway along the stream, we use a simple model of the Galaxy with a spherical halo (Allen & Santillan 1991) to integrate a range of possible orbits. Assuming an uncertainty of $0\buildrel{\circ}\over{.} 2$ in the positions of ten normal points evenly spaced along the stream, we use the ${\chi }^{2}$ in the positional agreement to arrive at 90% confidence limits for the proper motions and heliocentric velocities at the fiducial point necessary to match the trajectory of the stream. We then push each of these quantities to their 90% limits, integrate the orbits, and choose the extreme cases to put approximate limits on the orbital parameters. These limits are given in Table 1. The best-fitting orbits are shown in Galactic cartesian coordinates in Figure 4.

Zoom In Zoom Out Reset image size

Table 1.  Predicted Motions and Orbit Parameters

		Molonglo	Murrumbidgee	Orinoco	Kwando
Fiducial Point	R.A. ($^\circ$, J2000)	0.163	358.614	359.701	25.025
	Decl. ($^\circ$, J2000)	−16.869	+16.274	−25.324	−25.107
	vhel (km s−1)	$+{6.1}_{-329}^{+73}$	−123 ± 50	+64 ± 3.5	+130 ± 35
Prograde Orbit	${\mu }_{\alpha }$ cos(δ) (mas yr−1)	+8.08 ± 0.13	−0.134 ± 0.009	+0.253 ± 0.002	+0.969 ± 0.004
	${\mu }_{\delta }$ (mas yr−1)	−8.15 ± 0.13	+0.388 ± 0.01	−2.28 ± 0.0012	−1.633 ± 0.003
	vhel (km s−1)	$-{118}_{-73}^{+330}$	−182 ± 50	−121 ± 3.5	−94 ± 35
Retrograde Orbit	${\mu }_{\alpha }$ cos(δ) (mas yr−1)	−5.91 ± 0.13	+2.152 ± 0.009	+1.91 ± 0.002	+1.865 ± 0.004
	${\mu }_{\delta }$ (mas yr−1)	+3.96 ± 0.12	−3.074 ± 0.011	−2.06 ± 0.001	−2.009 ± 0.003
Rapo (kpc)	⋯	$\gt 100$	${24}_{-1}^{+4}$	21 ± 0.15	${26.4}_{-1.8}^{+3.2}$
Rperi (kpc)	⋯	20 ± 0.2	${21}_{-2.2}^{+0.1}$	6.9 ± 0.1	4.4 ± 0.7
i ($^\circ$)	⋯	68 ± 1	62 ± 4	79.7 ± 0.1	76.7 ± 4
	⋯	$1\pm 0.3$	${0.08}_{-0.001}^{+0.13}$	0.5 ± 0.003	0.72 ± 0.02
Orbit Pole	l ($^\circ$)	${193}_{-2}^{+9}$	160 ± 4	250 ± 0.3	237 ± 15
	b ($^\circ$)	+22 ± 1	+28 ± 4	+10.3 ± 0.1	+13 ± 4

Download table as:  ASCIITypeset image

With no detectable curvature on the sky, the trajectory of Molonglo evidently requires that the progenitor be essentially unbound from the Galaxy. This is somewhat surprising, and one might wonder how the progenitor could have been so extensively disrupted on so benign an orbit. The progenitor must either have been very loosely bound to begin with, or it may have been disrupted by an encounter with some other resident of the Local Group.

The Murrumbidgee stream is visible (though barely so at its southern end) over the entire field of view in Figure 1, from the SGP to within $5^\circ$ of the Galactic plane. Using the T-statistic of Grillmair (2009), the portion of the stream with $-65^\circ \lt b\,\lt -30^\circ$ is detected at approximately the $6\sigma$ level. The color–magnitude distribution for this portion of the stream is shown in Figure 3 and shows both a weak subgiant branch, along with a somewhat stronger lower main sequence. It also suggests that the stream could be slightly more metal-rich than our adopted search filter. On the other hand, application of a filter based on the color–magnitude distribution of stars in Pal 5 ([Fe/H] = −1.43) yields significantly less signal-to-noise ratio, indicating that the off-locus population at $g\approx 20.6$ in Figure 3 is probably unrelated to the stream.

The trajectory of the stream can be modeled to $\sigma =0\buildrel{\circ}\over{.} 14$ using

$$\begin{eqnarray}\alpha & = & 367.893-0.4647\delta -0.00862{\delta }^{2}+0.000118{\delta }^{3}\\ & & +1.2347\times {10}^{-6}{\delta }^{4}-1.13758\times {10}^{-7}{\delta }^{5}.\end{eqnarray} \tag{ 3 }$$

The stream is some $95^\circ$ long, and some 22 arcmin across (FWHM). At a distance of 20 kpc, these translate to a length of 33 kpc, and a width of 125 pc. The width is again comparable to those of known or presumed globular cluster streams such as Pal 5 (Odenkirchen et al. 2003), NGC 5466 (Belokurov et al. 2006; Grillmair & Johnson 2006), or GD-1 (Grillmair & Dionatos 2006). For the most visible, $45^\circ$-long portion of the stream north of the faint Sagittarius stream, subtracting a smoothed foreground distribution and counting stars in the color–magnitude envelope yields a total of 310 ± 120 stars. Doubling this to account for portions of the stream heavily contaminated by the disk or by Sagittarius stream stars, and integrating over a globular cluster-like luminosity function, we arrive at an estimated total of 7400 ± 2900 stars in the visible stream. This is again consistent with a globular cluster progenitor.

Within the limitations of our adopted Galactic model, an orbit fit to Murrumbidgee puts reasonably tight constraints on the orbital plane. The best-fitting orbit is highly inclined and nearly circular, with perigalacticon occurring just north of the faint Sagittarius stream. The orbit passes within 2 kpc of Pal 5 and 3 kpc of NGC 1261. The orbital path of Pal 5 is fairly well described by its tidal tails, and its orbit pole is inclined by more than $30^\circ$ to that of Murrumbidgee, ruling out any physical association. As we have no proper motion information for NGC 1261, we cannot say at this point whether there might be a physical association. If we allow the uncertainties in the fitted proper motions and radial velocities to extend to their $1\sigma$ limits, then we find that IC 4499 and NGC 5634 also lie with 3 kpc of an orbit integration.

The eastern $19^\circ$ of Orinoco (the portion nearest to the SGP) is detected at the $\approx 15\sigma$ level, and is interesting both for its pronounced curvature and its projected overlap with the ATLAS stream. A portion of the stream appears to be visible in Koposov et al. (2014), though not a length sufficient to have been identified as a separate stream. The western portion of the stream (extending downwards in Figure 1) is much weaker and somewhat conjectural. Fitting an orbit to just the eastern $19^\circ$ yields a best-fit trajectory that lies alongside and separated by less than a degree from the very faint feature in Figure 1 with $-62^\circ \lt b\lt -48^\circ$. Including this faint feature in the fit results in a no less plausible orbit. If indeed the two features are part of the same stream, then it would appear that there is a $\approx 10^\circ$ gap between the two. At this level of significance, it is plausible that this region is simply below our detection threshold. On the other hand, it may also constitute a physical gap in the stream, generated perhaps by a close encounter with a dark matter subhalo or some other Galactic constituent. Finally, we must recognize that this western feature could be an unrelated stream, or not a stream at all.

The easternmost $19^\circ$ of Orinoco can be modeled to $\sigma =0\buildrel{\circ}\over{.} 13$ using

$$\begin{eqnarray}\delta & = & -25.5146+0.1672\alpha -0.003827{\alpha }^{2}\\ & & -0.0002835{\alpha }^{3}-5.3133\times {10}^{-6}{\alpha }^{4}.\end{eqnarray} \tag{ 4 }$$

This section of the stream has a FWHM of about 40 arcmin, indicating a physical breadth of about 240 pc. This is somewhat broader than the streams above, and may be partly a consequence of confusion with the overlapping ATLAS stream. It may also be indicative of a somewhat more massive progenitor, or more significant or extended heating of the stream (Carlberg 2009). Estimating foreground contamination and counting stars in the appropriate color–magnitude envelope, we find a total of 225 ± 95 stars in the eastern $19^\circ$ of the stream. Integrating over a globular cluster-like luminosity function, we estimate a total population of 2700 ± 1100 stars.

The eastern (upper) portion of Orinoco apparently bends over and overlaps with a portion of the ATLAS stream (Koposov et al. 2014; Bernard et al. 2016). However, an orbit fit to Orinoco suggests only a few degrees of overlap before the stream diverges from ATLAS in a more southerly direction. We have attempted to fit an orbit to both Orinoco and the southern portion of ATLAS, but the resulting ${\chi }^{2}$ is much larger than for either stream alone. We conclude that ATLAS and Orinoco are not different parts of the same stream.

The best-fit orbit for Orinoco passes within 1 kpc of NGC 288, and of NGC 6356. It also passes within 2 kpc of the nearby globulars NGC 6121, NGC 6284, and NGC 6397. From inspection of Figure 1 we know that Orinoco cannot be associated with NGC 288. While the stream appears to arc around the cluster, it does not attach to the cluster in the characteristic "S-shape" we expect (e.g., Pal 5). Using proper motion measurements for the remaining four clusters (Dinescu et al. 1999; Casetti-Dinescu et al. 2010, 2013), we find that all of them have orbits confined to less than 5 kpc vertically from the Galactic disk. The estimated distance of Orinoco (20 ± 3 kpc), combined with the best-fit orbit parameters in Table 1, rule out any association between the stream and these particular clusters.

Faint and not particularly long or continuous, Kwando would not normally have been identified as a stream candidate were it not for its similarity and proximity to Orinoco. The same round-the-pole arc, shifted only a few degrees eastward from Orinoco, was sufficiently striking that we opted to consider it. The stream is detected at the $7\sigma$ significance level and, over the range $19^\circ \lt \alpha \lt 31^\circ$, the $13^\circ$ arc of Kwando is matched to $0\buildrel{\circ}\over{.} 1$ by a polynomial of the form:

$$\begin{eqnarray}&&\delta =-7.817-2.354\alpha +0.1202{\alpha }^{2}-0.00215{\alpha }^{3}.\end{eqnarray} \tag{ 5 }$$

With a FWHM of 22 arcmin, Kwando is evidently only 130 pc across, once again in the realm of presumed globular cluster streams. A color–magnitude selected count of stars yields 120 ± 52 stars over the $13^\circ$ arc of the stream, or 1400 ± 600 when integrated over a globular cluster-like luminosity function. In terms of linear density, the 9 ± 4 stars per degree actually exceeds the 8 ± 3 stars per degree measured for Murrumbidgee.

As might be expected based on appearance and proximity, the best fit to the orbit of the stream shows fairly similar parameters to that of Orinoco. To within the uncertainties, the inclination and orbit poles are identical. This suggests that, while these streams obviously had distinct progenitors, those progenitors themselves may have formed in the same cloud, or orbited a common parent body.

For Kwando, the best-fit orbit comes within 1 kpc of 47 Tuc. Based on the proper motion measurements of Watkins & van der Marel (2017), an orbit integration for 47 Tuc reveals that the cluster orbit is confined to within 5 kpc of the Galactic plane. A physical association between Kwando and 47 Tuc can therefore be ruled out. Both E 3 and Pal 8 also lie within 1 kpc of Kwando's putative orbit, but neither have published proper motions with which to estimate their orbits.