The JCMT Legacy Survey of the Gould Belt: a first look at Serpens with HARP

The James Clerk Maxwell Telescope’s (JCMT) Gould Belt Legacy Survey (Ward-Thompson et al., 2007) is a large scale programme to observe nearby ($<$500 pc) regions of active star formation in both submillimetre continuum emission and three CO species $J=\!3\!\rightarrow\!2$ emission¹¹1http://www.jach.hawaii.edu/JCMT/surveys/gb/. The survey will observe molecular clouds with SCUBA-2 (Holland et al., 2006) to detect the continuum emission, and has been carrying out heterodyne spectrometer observations of filamentary regions in a subset of these clouds. These observations are currently being carried out with HARP/ACSIS (Buckle et al., 2009) in the $\mathrm{{}^{12}CO}$, $\mathrm{{}^{13}CO}$ and $\mathrm{C^{18}O}$ isotopologues, in the $J=\!3\!\rightarrow\!2$ transition. POL-2 (Bastien et al., 2005), the SCUBA-2 polarimeter, will also be used to study some of the SCUBA-2 observed regions. A first look at the data from the Orion-B region has already been presented, by Buckle et al. (2010), and from the Taurus region by Davis et al. (2010).

The survey seeks to perform some of the first large scale analyses of very large, high resolution submillimetre maps and data cubes of the closest molecular clouds and examine the star formation occurring within them. In particular, the spectral part of this survey will examine the properties of higher density gas than most large scale surveys, due to the higher critical density of the $J=\!3\!\rightarrow\!2$ transitions ($10^{4}-10^{5}$ cm^-3). This is better matched to the densities probed in submillimetre continuum emission than the lower $J$ transitions. The aims of the spectral survey are to search for and categorise high velocity outflows in order to identify protostars; to examine the column density in cores; to study the support mechanisms of protostars and their evolution; and to examine the turbulence and velocity structure of the parent molecular clouds (Ward-Thompson et al., 2007). SCUBA-2 continuum observations will also be made, and will cover an extremely large area compared to the current data sets available from e.g. SCUBA. As well as looking at the individual regions, the GBS will carry out statistical studies and comparisons on all the data sets and objects found.

The Serpens molecular cloud is a much-studied region of low mass star formation, located $\sim$230 pc from the Sun (Eiroa et al., 2008, and references therein, in particular Straižys et al. 2003). It is forming a compact and high density cluster of new stars, and its close proximity allows us to map its gas properties at high spatial resolution (a FWHM beam size of 3300 AU at the distance of Serpens for $\mathrm{{}^{12}CO}$). The cloud consists of two clusters of embedded YSOs of which many possible Class 0/I/II sources have been detected by the spitzer c2d survey using MIPS (Harvey et al., 2007b); and using both MIPS and IRAC (Harvey et al., 2007a). The region observed in this study is focused on the Serpens main cloud core, which contains the well known submillimetre Class 0-I objects SMM1-11 (Casali et al., 1993; Davis et al., 1999). Hogerheijde et al. (1999) presented high resolution interferometric observations of some of these cores in a variety of molecular tracers. Evidence for infall towards SMM 2, 3 and 4 has been found (see e.g.  Gregersen et al., 1997), further supporting the view that these are very young sources currently undergoing star formation. Duarte-Cabral et al. (2010) find evidence that the global structure of the cloud implies it consists of two colliding sub clouds.

Much evidence of molecular outflows has been detected in the region in previous JCMT observations (White et al., 1995; Davis et al., 1999). Hodapp (1999) measured the proper motions of $\mathrm{H_{2}}$ jets in the north-west part of this region (in the vicinity of SMM 5, 9 and 10), further connecting these jets to outflow sources. The observations were centred on the region known as the Serpens cloud core, containing many protostars, dust sources, HH objects and potential outflows. Eiroa et al. (2008) presented a recent review of the entire Serpens region. This paper presents new observations and results of outflow kinematics along with the general cloud properties.

The $J=\!3\!\rightarrow\!2$ transition of the CO isotopologues observed here have critical densities of the order of 10^4-5 cm^-3. Tracing these high densities allow us to look at the dense gas more intimately involved with the ongoing star formation than lower transitions, tracing lower densities, allow. It also allows us to examine the higher temperature emission from molecular outflows, as these transitions occur at a correspondingly higher energy level – 31-33 K – than the lower transitions.

Fig. 2 (a) shows the integrated $\mathrm{{}^{12}CO}$ emission; Fig. 3 overlays IR sources, submillimetre continuum sources and Herbig Haro (HH) objects. As is evident from the integrated emission alone, the region is a complex, clustered star-forming region. The ‘finger-like’ extensions so common to CO observations in many regions are clearly discernible; however, it should be noted that these do not always correspond to obvious spatially compact bipolar outflows.

Our $\mathrm{{}^{12}CO}$ $J=\!3\!\rightarrow\!2$ map covers a region of 260 square arcminutes centred on RA:$18^{h}29^{m}45^{s}$ Dec:$1\degr 13\arcmin 57\arcsec$ (J2000). The region was extended westwards to follow the strong bow-shock that culminates in the Herbig Haro object HH 106. This extended section has higher noise and was not followed up in $\mathrm{{}^{13}CO}$ and $\mathrm{C^{18}O}$. The integrated intensity map in Fig. 3 is not dominated by emission at the systemic velocity of the cloud, presumably at least in part due to the strong self absorption present in the $\mathrm{{}^{12}CO}$ spectra. Fig. 4 presents the spectra for all three isotopologues plotted on the same axes for each of the SMM cores, and Fig. 5 shows spectra from some typical positions throughout the cloud. Many of the $\mathrm{{}^{12}CO}$ spectra exhibit either a double-peaked profile or a distorted Gaussian shape. By comparing the peak intensity and shape of the $\mathrm{C^{18}O}$ spectra from the same position as the $\mathrm{{}^{12}CO}$, it is clear that the shapes of the $\mathrm{{}^{12}CO}$ line profiles appear to be caused by self absorption rather than multiple velocity components.

The $\mathrm{{}^{12}CO}$ emission is, therefore, dominated by the less dense blue and red-shifted high velocity gas of the cloud. The absorption maximum appears slightly red shifted with respect to the local rest velocity–perhaps indicative of global infall in the cloud, as one would expect in a star forming region. There are also prominent velocity wings present in the $\mathrm{{}^{12}CO}$ spectra as can been seen in Fig. 4, also indicating the presence of molecular outflows in the region, some with velocities up to 20 km s^-1 with respect to the local rest velocity. These outflows have been observed before with molecular tracers (Olmi & Testi, 2002; McMullin et al., 2000; Davis et al., 1999; White et al., 1995) and with masers (Moscadelli et al., 2005); however the data presented here improve on the resolution of the previous single dish CO studies. Recent Spitzer observations (Harvey et al., 2007b, a) have identified several deeply embedded sources within the Serpens molecular cloud–we would expect these sources to be of a Class 0 or early Class I nature and thus drive powerful molecular outflows. Fig. 3 shows the integrated $\mathrm{{}^{12}CO}$ emission, with the positions of YSO candidates, SCUBA cores and HH objects also displayed. We can see that the Spitzer deep sources align with the sources in the 850 µm SCUBA image, which are known in the literature as the 10 SMM sources, SMM 1-11, although note that there is no SMM 7. SMM 7 was first identified by White et al. (1995) at RA: $18^{h}29^{m}55.6^{2}$ Dec:$+01\degr 14\arcmin 52\arcsec$, but was not detected in the SCUBA map of Davis et al. (1999). There is not a clear $\mathrm{C^{18}O}$ emission peak present at its position in Fig. 2(d).

Fig. 6 shows the SCUBA 850 µm emission from Davis et al. (1999) with contours of red and blue shifted gas in $\mathrm{{}^{12}CO}$  demonstrating that there are a large number of molecular outflows associated with the Serpens sources. These maps are at a higher resolution than previous $\mathrm{{}^{12}CO}$ studies, allowing more accurate flow and source identification, compared to the previous results of Davis et al. (1999), where some of the flows are unresolved. Fig.7 shows another view of the $\mathrm{{}^{12}CO}$ velocity structure of the cloud. In this three-colour image showing the red-shifted, blue-shifted and ambient emission (green) the strong red and blue lobes can be clearly seen.

The isotopologues $\mathrm{{}^{13}CO}$ and $\mathrm{C^{18}O}$  are much less abundant than $\mathrm{{}^{12}CO}$, with the abundance ratio for the ‘local’²²2Measured in the Orion GMC ISM being calculated as $X^{{\mathrm{{}^{12}CO}}}/X^{{\mathrm{{}^{13}CO}}}=77$ (Wilson & Rood, 1994). Being less abundant, these species are only seen on the denser parts of the sub-clusters, allowing us to probe the motions of the gas more closely associated with the protostars rather than the outflowing gas. Our $\mathrm{{}^{13}CO}$ and $\mathrm{C^{18}O}$ maps (Fig. 2) cover a region of 17$\times$11 arcminutes centred on RA $18^{h}29^{m}58^{s}$, Dec  $1^{\degr}13^{\arcmin}02^{\arcsec}$(J2000). Fig. 8 shows the SCUBA 850 µm emission and overlays contours of the $\mathrm{C^{18}O}$ integrated intensity emission. We can see the $\mathrm{C^{18}O}$ emission traces the general shape of the dust and the filaments to the east and south identified by Davis et al. (1999) very clearly, although specific peaks do not align exactly with the position of the SCUBA cores. The $\mathrm{C^{18}O}$ spectra (in Fig. 4) show some evidence of multiple components along the line of sight. This suggestion is strengthened by the analysis of Duarte-Cabral et al. (2010), who suggest that there are multiple velocity components and that the whole cloud can be modelled as two colliding flows or sub-clouds. These multiple velocity components from the $\mathrm{C^{18}O}$ emission should not be confused with the double-peaked structure of much of the $\mathrm{{}^{12}CO}$ emission. This occurs at very different velocity scales, hence our belief that it is due to self absorption of the line by nearer, cooler gas.

The correlation between the $\mathrm{C^{18}O}$ and the continuum emission suggests that the $\mathrm{C^{18}O}$ emission is not optically thick in the bulk of the emission. This is also consistent with the CO spectra shown in Fig. 4 where the $\mathrm{{}^{13}CO}$ still shows some self absorption from colder red shifted gas as seen in the $\mathrm{{}^{12}CO}$ spectra, whereas the $\mathrm{C^{18}O}$ spectra do not show such a clear pattern – although the presence of multiple components does make this less clear. In Sec. 3.1 we investigate the opacities of the isotopologues across the cloud.

It can be seen from these observations (see Fig. 2) that the $\mathrm{{}^{12}CO}$ and $\mathrm{C^{18}O}$ emission are tracing different gas structures – the $\mathrm{C^{18}O}$ emission looks very similar to the SCUBA emission, tracing a clumpy medium containing dense cores. The $\mathrm{{}^{12}CO}$ emission on the other hand is bright throughout the cloud, but is dominated by lobe-like filaments extending out from the cloud rather than clumpy emission in the centre of the cloud. The $\mathrm{{}^{12}CO}$ does not appear to be tracing the SMM cores. This suggests that the $\mathrm{{}^{12}CO}$ emission probably becomes optically thick in the gas surrounding the cores, and does not see all the way to the central infrared dense cores.

In order to look further at the different structures traced by the three isotopologues, Fig. 9 shows position-velocity (PV) diagrams taken along a line of constant declination through the position of SMM 1 in all three isotopologues. In this Figure the PV diagrams of both $\mathrm{{}^{13}CO}$ and $\mathrm{{}^{12}CO}$ show evidence of self-absorption dips in the centre of the line, which are not seen in the $\mathrm{C^{18}O}$. The $\mathrm{{}^{12}CO}$ emission also clearly shows strong evidence for red- and blue-shifted emission in the line wings, which is not clearly seen in the $\mathrm{{}^{13}CO}$ and entirely absent in the $\mathrm{C^{18}O}$ maps. Again, this emphasises that the isotopologues are truly tracing different environments, and specifically that the $\mathrm{C^{18}O}$ emission is not influenced by the molecular outflows, at least at the sensitivities of these data.

The variation of optical depth across the cloud can be examined by looking at the ratio of intensities of the different isotopologues, $I_{\mathrm{{}^{13}CO}}/I_{\mathrm{C^{18}O}}$ and $I_{\mathrm{{}^{12}CO}}/I_{\mathrm{{}^{13}CO}}$, across the entire region being studied. The relation between the line intensity ratio and the optical depth (e.g.  Rohlfs & Wilson, 2000) at a specific velocity is:

Where $T$ is the is the brightness temperature, and $\tau$ is the optical depth.

Ratio maps were created, displaying the ratio of the peak intensities of the isotopologues, thresholded at a five sigma peak value For optically thin emission, the ratio of the intensities of two lines is expected to be to tend towards the relative abundances of these species as we look towards the edge of the map.

For the $\mathrm{{}^{12}CO}$/$\mathrm{{}^{13}CO}$ ratio, we see maximum values of only 10 or so at most, compared with a measured abundance ratio of 77 (Wilson & Rood, 1994), although this was from a different cloud. This implies that the $\mathrm{{}^{12}CO}$ is optically thick through the whole of the cloud. Due to the extremely strong self absorption in the $\mathrm{{}^{12}CO}$  we do not attempt to calculate a valid optical depth from this ratio.

The optical depth in the cloud for $\mathrm{C^{18}O}$ and $\mathrm{{}^{13}CO}$ can then be calculated by utilising Eqn. 1 above and the approximations that $\tau_{13}=X_{1318}\tau_{18}$ and $\tau_{12}=X_{1213}\tau_{13}$, where $X$ is the appropriate abundance ratio. Often this is calculated at the velocity of the peak channel. However, as we can see in the spectra from these cubes (see Fig. 4), there is visible self absorption in the $\mathrm{{}^{13}CO}$ and the $\mathrm{{}^{12}CO}$  which would tend to limit the validity of this approach by underestimating the ratio by a varying amount across the map. Therefore, we follow the method in Ladd et al. (1998) and examine the ratio of the integrated intensities instead. Although the self-absorption will still cause the integrated $\mathrm{{}^{13}CO}$ intensity to be underestimated, this should be a lower fractional error than would occur if the intensity ratio at the velocity of the peak was used. This remaining systematic under-estimation will however boost the $\mathrm{C^{18}O}$ optical depth, particularly in the central regions where the $\mathrm{{}^{13}CO}$ self absorption is strongest.

Fig. 10 shows the ratio map for the $\mathrm{{}^{13}CO}$/$\mathrm{C^{18}O}$ integrated ratio.

We then use:

where $\tau_{13}(\nu)=\tau_{13}\exp[-\nu^{2}/2\sigma^{2}]$, and $\tau_{18}(\nu)=\tau_{13}(\nu)/f$, to calculate the optical depth by numerically minimising the following function:

This method is very stable to the value of $\sigma$ that is chosen, therefore we have simply used an average line-width instead of estimating the Gaussian line-width for each spectra We used $\sigma$ value corresponding to a line-width of 1.5 km s^-1, the FWHM of an averaged $\mathrm{C^{18}O}$ spectra. The abundance ratio can be calculated for optically thin lines by looking at the intensity ratio towards the edge of the cloud. We see values approaching the canonical value of 8 (Frerking et al., 1982) therefore we feel confident both in using this as the abundance ratio for $\mathrm{{}^{13}CO}$/$\mathrm{C^{18}O}$  and that the $\mathrm{C^{18}O}$ emission really is optically thin across the cloud.

Across the map the $\mathrm{{}^{13}CO}$/$\mathrm{C^{18}O}$ ratio varies from around 3-8 across the central region to values from 2.6-7.0 towards the cores, with most clustered around a value of 3. This gives $\tau_{18}$ values of 0.07 to 0.9 towards the cores, and $\tau_{13}$ values of 0.61 to 7.3. This suggests that the $\mathrm{C^{18}O}$ 3–2 emission is optically thin across the map, and is even (marginally) optically thin towards the position of the submillimetre cores, and that the $\mathrm{{}^{13}CO}$ 3–2 emission is transitional between optically thick and thin across much of the cloud, and optically thick towards the dense cores. Fig. 11 shows the $\mathrm{C^{18}O}$ optical depth calculated across the map. It is very noticeable that the regions of high optical depth do not correspond to the positions of the submillimetre SMM cores. This is not unexpected, as the peaks of the $\mathrm{C^{18}O}$ emission do not align exactly with that of the continuum cores. However, the effect of the $\mathrm{C^{18}O}$ ceasing to be optically thin towards extremely dense regions, as well as the possibility of depletion towards the centre of dense cores and the slight self absorption present in the $\mathrm{{}^{13}CO}$ emission prevent us from investigating this further.

The excitation temperature of the molecular gas can be calculated from the peak temperature of emission from optically thick gas (via the assumption of local thermodynamic equilibrium). In our observations however, the most optically thick tracer – $\mathrm{{}^{12}CO}$ – is complicated by the strong self absorption features seen in the $\mathrm{{}^{12}CO}$ line profiles across much of the cloud. This would underestimate the excitation temperature in these dense regions. However, as we have calculated that the $\mathrm{{}^{13}CO}$ emission is just optically thick across much of the cloud, we can also calculate the excitation temperature from this emission, using:

where $T_{B}$, the main beam temperature, is assumed to be the same as the peak temperature of the transition, $T_{0}$ is $h\nu/k_{B}=16.6$ K for $\mathrm{{}^{12}CO}$ and $15.9$ K for $\mathrm{{}^{13}CO}$, and $T_{bg}=2.73$ K is the temperature of the cosmic background. $\tau$ is the optical depth for the transition of interest and is assumed to be infinity here. This equation assumes that the emission completely fills the beam. This gives:

The difference in temperatures derived from the $\mathrm{{}^{13}CO}$ and $\mathrm{{}^{12}CO}$ maps is interesting, particularly given that we expect the $\mathrm{{}^{12}CO}$ temperature to be significantly underestimated due to the stronger self absorption. Fig. 13 suggests that the excitation temperature estimated from the $\mathrm{{}^{12}CO}$ appears to trace some of the outflow structure in this cloud. This can be most clearly seen in the hot lobe-like structure visible in the NW of Fig. 13 This structure corresponds to the position of HH 460, and there is known bright $\mathrm{H_{2}}$ emission in the region. This suggests that shocks are primarily responsible for producing the higher $\mathrm{{}^{12}CO}$ temperatures

Also noticeable in this analysis is an arc of high temperatures towards the east of the $\mathrm{{}^{13}CO}$ map in Fig. 12. Near to this position there is both a reflection nebula and the B type star HIP 90707. It is possible that these are influencing the gas temperatures in this region. The $\mathrm{{}^{13}CO}$ map shows narrow but bright spectral line profiles in this region

The mass of the cloud has been estimated from the integrated $\mathrm{C^{18}O}$ intensity. This calculation assumes that the $\mathrm{C^{18}O}$ emission is optically thin. This is generally expected to be true everywhere in molecular clouds apart from the centre of the very densest cores, and for Serpens this was confirmed above in Sec. 3.1. These extremely dense regions contain only an insignificant proportion of the total mass in this region. The ratio of $\mathrm{C^{18}O}$ to $\mathrm{H_{2}}$ is taken to be $10^{-7}$, and the distance to the cloud was assumed to be 230 pc (Eiroa et al., 2008). $T_{ex}$, the excitation temperature, was assumed to be 12 K. This seems reasonable both as a fairly standard value for cold gas in molecular clouds, and from the excitation calculations given above from the $\mathrm{{}^{12}CO}$ and $\mathrm{{}^{13}CO}$ emission.

This gives a mass for the cloud of 203 $M_{\odot}$. Quantifying the error on this number is not trivial given the large number of uncertainties in the input parameters and assumptions; however, the largest sources of systematic errors are probably contained in the estimate of the abundance and in the assumption of LTE. The relative abundance of CO to $\mathrm{H_{2}}$ has not been measured in Serpens, therefore we have adopted the standard literature value.

This virial mass is approximately 80 percent of the measured $\mathrm{C^{18}O}$ mass, suggesting that the cloud is in a bound state. However, this result is of course subject to the same sort of systematic uncertainties as the mass estimate,. These large uncertainties prevent us from drawing a more firm or quantitative conclusion about the state of the cloud other than to estimate that the measured $\mathrm{C^{18}O}$ mass is close the virial mass, even if we attempted to use a more complex model of the density.

We can compare the different energetics present in the cloud by making some simple approximations. We can estimate the gravitational binding energy and the turbulent kinetic energy using the following equations:

where $G$ is the gravitational constant, $M$ is the mass of the gas in the cloud, $R$ is the estimated size of the cloud, and $\sigma$ is the estimated 1-D velocity dispersion in the cloud, equal to $v_{fwhm}/2.35$. This assumes a uniform density cloud. A more centrally condensed cloud would be more tightly bound (i.e. would have a more negative gravitational binding energy). Similarly, due to the $M^{2}$ factor involved, the uncertainties in the mass of the cloud will produce a large uncertainty in the gravitational binding energy. The error in the assumption of a single velocity dispersion for the entire cloud will also produce large uncertainties in these values, due to the $\sigma^{2}$ factor.

The virial mass calculated above corresponds to a gravitational binding energy of 245 M_⊙ km² s^-2 The turbulent kinetic energy can be estimated by measuring the CO line width in regions where the emission is not affected by outflow activity, therefore we use the optically thin tracers. The $\mathrm{{}^{13}CO}$ and $\mathrm{C^{18}O}$ maps give roughly similar results of $\sim 1.5-2$ km s^-1for the line width, 2km s^-1 is used here, giving a turbulent kinetic energy of 220 M_⊙ km² s^-2.

The energetics of the cloud are summarised in Table 1, along with the global outflow energy from Sec. 5.1.

This analysis suggests that the cloud is gravitationally bound. However, it is important to reiterate that there are large uncertainties in the estimates of the values presented here.
We can also examine the importance of the outflows in this region. The outflow momenta and energy have been derived by assuming that the measured speeds are along the line of sight; this will underestimate the values. To estimate the correction to this, we assume the direction of the outflows is distributed isotropically. As we have many flows contributing to the global outflow momentum and energetics, this is a good approximation. Table 1 shows the corrected values. These suggest that the outflow energy is approximately 70 percent of the turbulent energy. This result would suggest that in this dense, clustered region the outflows are a very important effect and could be the dominant factor driving the local turbulence. However, it is important to note that there are significant uncertainties in the estimation of these variables (as mentioned in the discussion of the mass, and in further discussion in Sec. 5.1). Further work on disentangling the individual outflows and on analysing the detailed structure of the entire region will be required to quantify this more accurately, possibly using observations of other tracers.

The velocity structure of the dense gas of the Serpens cloud can be seen on Fig. 14 as a velocity coded 3 colour plot. A first glance at this figure shows that the blue shifted emission is east of the Serpens filament, whereas the red emission is mainly west of it. This gradient has been previously detected by other authors (Olmi & Testi, 2002), who interpreted it as a global rotation of the cloud. However, this conclusion was limited by the resolution of their data, which could not resolve the velocity structure in as much detail as we can.

Figure 14 demonstrates that this velocity structure is more complex than simple rotation, by showing a 3 colour plot of the $\mathrm{C^{18}O}$ red-shifted, blue-shifted and central-velocity emission. For instance, we can see, in the SE end of the Serpens filament (the SE sub-cluster) a sharp merger/overlap of the blue and the red emission, not expected in a solid body rotation scenario. On the other hand, the NW end of the filament (the NW sub-cluster) has very little blue-shifted emission. This velocity structure was also suggested to be possibly a shear motion (Olmi & Testi, 2002) or a collision of two clouds/flows where the interface coincides with the SE end of the filament (Duarte-Cabral et al., 2010).

Another noticeable feature visible in this view of the $\mathrm{C^{18}O}$ velocity structure, are the gas streams east of the filament. Note that these are not seen at the west, nor do they correlate with any of the outflows seen in ¹²CO. These small filaments roughly perpendicular to the main filament are commonly seen in filamentary molecular clouds, but their origin is not yet clear (Myers, 2009).

In order to look at the velocity structure at specific positions we have created PV diagrams. By studying several PV diagrams from our $\mathrm{C^{18}O}$ maps, we can get a better sense of how the velocity changes and evolves across the cloud. Our approach to such a study involves several cuts both with constant Declination and with constant RA. Fig. 15 shows the positions of the PV cuts used here.

Figs. 16 and 17 shows PV diagrams (horizontal slices of the data cube) at five fixed Declinations, varying the Right Ascension. Fig. 16 cuts through the NW sub-cluster, where we can note that this sub-cluster is represented by a velocity mainly between 8 and 8.5 kms^-1. However, there is some lower velocity gas weakly emitting east of the sub-cluster, with velocities around 7.5 kms^-1. Whether this gas is physically connected with the higher velocity gas is not clear, as it could either represent a separate cloud with a slightly different velocity, or the same cloud undergoing rotation. Therefore, we must examine the the evolution of these components as we look southwards through the cloud in order to understand what these gas components are tracing.

Fig. 17 shows the velocity structure of the SE sub-cluster. Here the two velocity components, which in the north are very well separated, begin to merge. As we move south the components become more offset from each other in velocity, and we see regions that are clearly reproduced by two velocities (one around 7 kms^-1 and the other around 8.5 kms^-1). Note that the regions which also show dust emission are coincident with regions where the two velocities coexist along the line of sight. Finally, the lower velocity gas ceases to exist, and at the very southern end of the filament solely consists of the higher velocity gas, at around 8.3 kms^-1.

We also study the velocity structure using vertical PV diagrams (constant RA) in Fig. 18. The information we get from here adds to the horizontal diagrams, in the sense that we can confirm the complex velocity structure, including the double velocity component towards the south sub-cluster (best seen on Fig. 18, top right panel). We can also see velocities below 8 kms^-1 towards the northern region, in the first two panels. Because this corresponds to higher RA positions, these lower velocities in the north are only seen in regions offset to east from the sub-cluster seen in 850 $\mu$m emission. The two last panels of Fig. 18 (bottom row, centre and right panels) show the bulk of emission towards the NW sub-cluster, which traces well the 850 $\mu$m continuum emission, showing us the velocities all concentrated between 8 and 8.5 kms^-1. the outflow structure of the cloud is complex. Although there appear to be many outflows in this region, many are overlapping. Instead of attempting to assign spectra arbitrarily to one or another of an overlapping outflow in order to calculate their properties, we have decided to look first at the global outflow properties (i.e. energetics for the all the high velocity gas), and then to look individually at those outflows that can be spatially located via their very high velocity emission.

To calculate the global outflow properties, we wish to look at all the gas that is moving at a significantly red- or blue-shifted velocity from the line centre. However, it is not possible to completely exclude the ambient emission, and some proportion of this will also be included in these calculations.

We can calculate some basic kinematic properties for the global properties of the outflowing gas. Examination of the spectra in the cloud shows that ambient $\mathrm{{}^{12}CO}$ spectra away from the outflows appear to cover a velocity range of 6 to 10 km s^-1. This is assumed to include ambient emission for the entire cloud and is excluded from the integration. Therefore the integration ranges used are -15 to 6 km s^-1for the blue-shifted emission and 10 to 30km s^-1 for the red-shifted emission.

The masses were calculated under the assumption of LTE, a distance of 230 pc and assuming optically thin emission in the line wings with an excitation temperature of 50 K (Davis et al., 1999; White et al., 1995). Although this is higher than the peak excitation value we calculated earlier, it is expected that the gas in the outflows will be considerably warmer than that in the bulk of the gas.

The momentum and energy are estimated similarly, by replacing $\int T_{A}^{*}dv$ by $\int T_{A}^{*}|v-v_{0}|dv$ and $0.5\times\int T_{A}^{*}(v-v_{0})^{2}dv$, where $v_{0}$ is the central velocity. Here, an average central velocity value is used for the whole cloud.

The global results for the Serpens cloud are summarised in Table LABEL:tabglobal. In Table 1 we give the total energy from both global outflow values: 50  M_⊙ km² s^-2. It is of course again important to consider the sources of systematic uncertainty affecting these figures. Our assumption of the excitation temperature of 50 K, our chosen value for the CO abundance, and our assumption of LTE are sources of uncertainty in these values. We also expect some proportion of the shocked gas to be disassociated from its molecular form, and this will be a larger effect at the high velocities. We have also assumed that the entire line wing is optically thin; in fact this may well not be true, especially at the velocities close to the ambient cloud emission. This effect may mean that our values for the outflow masses are significantly under estimated. There may also be an effect on the momentum and energy estimated, but the optical depth is likely to be a less important factor and the disassociation a larger factor as these values are dominated by the high-velocity emission. These effects would tend to increase the values shown here. It is also important to reiterate that there will be some contamination of the values shown here by the ambient gas.

Eqn. 11 only provides us with the energy and momentum along the line-of-sight, and we must correct for the inclination of the outflows if we wish to calculate 3-D values. The outflow velocity should be corrected by $1/cos(i)$ from the line-of-sight velocity, where $i$ is the inclination angle. If we assume that the outflows are in an isotropic random distribution, this will increase the momentum by a factor of $2$ and the kinetic energy by $3$.

We have also made a rough estimate of the dynamical age of the outflows in the region. An estimate of the average flow lobe length was found from Fig. 6 of 210 arcseconds (0.23 pc at $D=230$ pc) in the plane of the sky. The average maximum velocity in the line of sight is $\sim$14 km s^-1, thus obtaining a value of 1.6$\times 10^{4}$ yr. This very young age agrees with the conclusions in the literature that the Serpens sources are Class 0/I YSOs (Casali et al., 1993; Testi & Sargent, 1998). The values for these energetic properties correspond well to those found in Davis et al. (1999).

The results in Tables 1 and LABEL:tabglobal allow us to discuss the validity of the theory of supersonic turbulence sustained by molecular outflows.

If we apply a correction for random inclination, we estimate the total kinetic energy contained in the outflows to be of order 70 per cent of the total turbulent kinetic energy (see sec. 3.4). This suggests that the outflows are a strong influence on the structure and energetics of the Serpens cloud core, and studies of this region must take care to account for them (particularly with respect to small scale structure).
Our finding that the turbulent kinetic energy and the global outflow energy are of comparable size fits well with theories and simulations that involve the driving of supersonic turbulence by outflows in regions of active star formation (see e.g. Li & Nakamura (2006); Matzner (2007)). However, it is still necessary to identify a mechanism that can convert the outflow momentum (directed outwards from the cloud) into the turbulent energy. An analytical model for this process was developed by Matzner (2007), who proposed that molecular cloud turbulence can be driven by collimated protostellar outflows, resulting in a cascade of momentum rather than energy. Numerical simulations by Li & Nakamura (2006) and Carroll et al. (2009) appear to be consistent with this picture.

A further interesting comparison can be made between the turbulent kinetic energy and the gravitational binding energy. Since the binding energy is very similar to the turbulent kinetic energy this suggests that the entire cloud is only weakly bound and that some small, local, regions may well be in a state of collapse where as others are in a state of expansion. This is one of the expected observations of the theory of star formation control by supersonic turbulence (see  McKee & Ostriker, 2007, and references therein, e.g. Mac Low & Klessen (2004)). Further insight can be gained by considering the ratio of kinetic outflow energy to gravitational binding energy. A value of $\sim 0.7$ is indicative that outflows are extremely important to understanding the cloud support mechanism. However, we expect the uncertainties in the estimation of the various parameters are large enough to prevent a more firm conclusion

As can be seen from the $\mathrm{{}^{12}CO}$ channel maps presented in Fig. 19, the $\mathrm{{}^{12}CO}$ high velocity structure is extremely complex – identifying outflows from the $\mathrm{{}^{12}CO}$ emission alone is extremely difficult. We see considerable ‘finger-like’ structures in this region, that correspond to large scale outflow structure across the cloud. Although areas containing high velocity red- and blue-shifted gas can be seen clearly in the channel maps, categorising these into neat bipolar lobes is not so simple. However, there are reasonable amounts of complementary data available. The presence of known HH objects are indicated on Fig. 6, and a comparison with the positions of known cores allows a few more clear-cut outflows to be identified. Figs. 22 and  23 show PV diagrams along some of these outflows.