Discussion

With the GMRT field of view of $\sim1{^\circ}.4$ at 327 MHz, low frequency mapping in the Galactic plane reveal a variety of sources of compact as well extended emission. The GMRT observations of many the objects presented here, are the first high resolution observations at these low frequencies. With the Central Square providing reliable measurements up to $\sim 100\lambda$, these measurements are also sensitive to angular scales of up to $\sim30$ arcmin. Emission at such large scales with a sub-arcsecond resolution and sensitivity of $5-20$ mJy/beam detects large and small scale structures, not detected in other earlier observations in the Galactic plane. High resolution imaging at other GMRT frequencies of the objects discussed here will provide additional information, not available from any other observation, which will help in determining the nature of these sources. In particular, thermal and non-thermal emission can be separated using the difference in the continuum spectra at these frequencies. Detailed spectral index changes within extended objects, which provides unique information about the emission mechanisms, physical parameters of the objects (e.g. H II and UC H II regions, SNRs) as well as information about the parameters of the intervening ISM, can be studied using such observations.

From the available data for G003.6$-$0.1, components A and the associate extended emission indicates that the emission is non-thermal. At higher frequencies, C and D are seen as distinct compact sources. The spectra of C is consistent with it being a radio loud symbiotic star. HI absorption spectra towards these three components show that D is farther away and probably an extragalactic background source. A and C are at similar kinematic distance and the ridge of emission seen clearly in the 5 GHz image, and marginally in the 1420 MHz image is indicative of an interaction between C and the extended emission. Continuum observations at 8 and possibly at 15 GHz with the VLA are needed to conclusively determine the nature of these objects. Spectral index of the emission from the shell can be measured using continuum images from the GMRT at 233 and 610 MHz. High resolution RRL observations towards G003.6$-$0.1 are also required to determine the source of RRL detected at 5 GHz from this direction.

The linear extended object see in the field of G356.3$-$1.5 is similar to that detected earlier by Gray (1996). Higher resolution observations, and if possible, measurements of polarization properties of this object are required to further ascertain the nature of this object.

327-MHz images of the UC H II regions, namely G004.4$+$0.1, G003.349$-$0.076 and G003.351$-$0.077, are the first images which resolve the low frequency extended emission from these fields. The morphology of G004$+$0.1 seen in this image is typical of UC H II regions. However, the liner size of $\sim 5$ pc is too large compared to the size of $<0.1$ pc of typical UC H II regions. Extended emission at 327 MHz in the immediate vicinity of G003.349$-$0.076 and G003.351$-$0.077 is also detected.

Extended emission around UC H II regions at angular scales of several arcmins (corresponding to linear sizes in the range of $2-20$ pc) have been recently detected at 5 and 1.4 GHz D-array VLA observations (Kim & Koo2001; Koo et al.1996). These extended components have been found to be kinematically and certainly morphologically associated with the compact components and it appears that the ionizing source of for the extended and compact components is same.

About 1000 UC H II regions have been identified so far. This gave rise to what is referred to as the ``age problem'' (de Pree et al.1995, and reference therein): the number of UC H II regions estimate from the IRAS $60-12 \mu$m and $25-12 \mu$m colours, is about an order of magnitude greater than expected from other indicators of massive star formation rate based on their dynamical age (Wood & Churchwell1989b; Wood & Churchwell1989a). From this, it is inferred that the life of the ultra-compact phase of H II regions is $\ge \sim 10^5$ yr, larger by an order of magnitude than their sound crossing time ( $\le \sim 10^4$ yr). Most of these UC H II regions were identified from their small size ($\le 0.1$ pc) and high interfered electron density ( $n_e \ge 10^4$ cm$^{-3}$). Recently, VLA D-array imaging of an UC H II region (G005.58$-$0.24) revealed emission at scales ranging from 0.04 to 40 pc (Koo et al.1996) which appeared morphologically associated with the UC H II region. It has an ultra-compact core, a compact core, an extended halo and a large diffused plateau. This prompted observations of a randomly selected sample of UC H II regions using the VLA at 3.6 cm in the D-array configuration (Kurtz et al.1999). These observations revealed extended emission around compact cores in 12 out of 15 sources. More recently, in a survey by Kim & Koo (2001) of 16 UC H II regions using the VLA (DnC array) at 1.4 GHz, extended emission at scales of $2-12$ arcmin ($4-19$ pc) is detected in each one of the fields. It, therefore appears, that the previous classification of UC H II region was essentially based on high resolution observations with the VLA, which suffered from the selection effect due to the insensitivity to large scale emission and therefore revealed only the compact/ultracompact core. It now appears that the UC H II regions may be just the compact cores of larger HII regions. Evidence of this association comes from the tight correlation between the velocities of the UC H II regions, compact components and the extended envelopes (Kim & Koo2001). The fact that the extended envelopes are detected in such a large fraction of UC H II regions in observations which are sensitive to large scale emission further indicates that the extended emission is associated.

The implications of a physical association of this extended emission with compact cores are far reaching. Kim & Koo (2001) estimate that most sources known as UC H II regions are likely to be associated with extended emission. Existence of extended emission at scales $\ge \sim$ few pc, ionized by the same sources, implies that the actual age of the so-called UC H II regions is $\ge \sim 10^4$ yr which could mitigate the ``age problem'' (de Pree et al.1995). The IRAS colour criteria may select compact or extended H II regions, as well as UC H II regions - an idea consistent with the results of Codella et al. (1994) who found that more than half of the 445 diffuse H II regions are related to IRAS points sources which satisfy this colour criteria. No variation in the IRAS colours was also found for UC H II regions with evidence of extended emission, implying that the colour criteria is insensitive to the presence of extended emission, and a significant fraction of IRAS colour selected UC H II regions may have associated extended emission.

It therefore appears that the sequence of observations based, first on the selection of sources based on IRAS colours and then high resolution radio observations, which led to the ``identification'' and classification of UC H II regions is fraught with severe selection and observational biases. If most of the so-called UC H II regions have extended emission associated with them, these UC H II regions may be just compact cores, possibly composed of several compact components, of large HII regions. Observations of a larger sample of such sources with the GMRT, which provides simultaneous high resolution and sensitivity to large scale emission, will greatly help in settling the issue of the existence of associated large scale emission. None of the models put forth to explain the longevity of UC H II regions predicted the presence of extended emission around them. Kim & Koo (2001) have proposed a model, which is a combination of champagne flow model with the hierarchical structure of massive star-forming regions (Tenorio-Tagle1982, and references therein). A massive star, which forms off-centre within a hot core, which in turn is embedded in a lower density molecular clump, can produce the compact component seen associated with the UC H II regions due to the hot core. The morphology of this compact component can be explained by the champagne flow, which would develop when the ionizing front breaks out of the core. The HII region inside the hot core continues of be ultracompact, while it grows to $\sim1$ pc outside the core. Another champagne flow would develop when the ionization front crosses the edge of the molecular clump, which forms a more extended emission. The morphology in the radio continuum image, as well as the gradient of H76$\alpha$ line emission lends support to this model. However, if most of the so-called UC H II regions show associated emission, it will be difficult to explaining why so many of the UC H II regions would correspond to a situation where the star, the hot core and the molecular clump are all ``carefully'' arranged to give the desired observed morphology. High resolution observations, sensitive to large angular scales, of a larger sample (preferably a complete sample) of UC H II will be most desirable to make progress on this front.

The integrated spectral index for G004.417$+$0.126 between 5 and 1.4 GHz exhibits a negative spectral index ($<-0.1$). This source is well resolved at 327 and 1.4 GHz and from these images, it appears that the compact core has a flat spectrum while the extended emission has a significantly negative spectral index between these frequencies. The images at 327 MHz and 1.4 GHz for G003.349$-$0.076 are not reliable for the measurement of the spectral index, but there are indications that the extended emission there too has a negative spectral index. None of models for UC H II regions predict a non-thermal component. The model proposed by Kim & Koo (2001) for the extended emission, predicts a thermal spectrum for the extended emission. A larger sample of UC H II regions needs to be mapped at 327 MHz and 1.4 GHz to access if a non-thermal component is characteristic of the extended emission around UC H II or these sources are just odd cases, emission from which needs to be explained separately.