3.5.1: In tro d u ctio n —radio cores and h otspots.

In document Optical emission lines in radio sources of intermediate power (Page 71-74)

M ost extrag alactic radio sources exhibit com pact regions of enhanced rad io em ission. Since th e sources in our sam ple lie at m oderately large distances and since our resolution is on the order of arcseconds, these regions are unresolved in o u r d a ta . T hey fall into two categories depending on th eir position w ith in th e rad io source.

R adio cores are regions of radio em ission coincident w ith th e nucleus of th e host galaxy. T hey are characterised by sm all an g u lar sizes an d high surface b rig h tn ess, an d are often presum ed to be associated w ith th e cen tral engine of th e rad io galaxies (see K ellerm an and Pauliny-T oth 1980 for a review of com pact rad io sources). Core sources can be fu rth er divided into two categories depending u p o n th e slope of th eir radio spectrum . F lat sp ectru m radio cores have a rad io sp ec tru m w ith a spectral index a < 0.5 (S v oc v ~ a ). VLBI d a ta shows th a t these flat sp ectru m sources have a sp atial extent on th e order of a few parsecs and b rig h tn ess te m p eratu res of 1011 to 1012 K. Steep sp ectru m cores have a radio sp ec tru m a > 0.5, sim ilar to th a t of th e m ore extended em ission. It is generally accepted th a t th e difference betw een these two types of radio cores is due to th e presence of synchrotron em ission from optically thick m a terial in th e flat sp ec tru m cores. Steep sp ectru m cores, however, have typical sizes on th e o rd er of kiloparsecs resulting in optically th in emission regions. We discuss th is fu rth e r in ch ap te sr 4 and 7 after p resen tatio n of th e results of long-baseline in terfero m etry on a subset of our sam ple.

E x tra-n u clear com pact regions are referred to as “h o tsp o ts.” Class II rad io sources have h o tsp o ts a t the edges of th eir radio lobes giving rise to th e c h a r­ acteristic edge-brightened n a tu re of these sources. T hey are generally assum ed to represent th e working surface of a jet where strong shocks convert th e bulk kinetic energy of flow in the jet into synchrotron em ittin g relativ istic p articles (e.g. B landford an d Rees 1974). H otspots in class I sources are no t u b iq u ito u s as in class II sources an d are generally found closer to th e nucleus. T hese are also supposed to represent stro n g shock regions b u t w ith o u t s u b sta n tia l d isru p ­ tio n of th e jet. Only a few of th e class I sources in our sam ple exhibit h o tsp o ts.

A lthough the h o tsp o ts in our sam ple are unresolved, o th er studies show th em to be typically a few kiloparsecs in size, elongated p erp en d icu lar to th e source axis. T hey exhibit a sp ectral index of aro u n d 0.6 to 0.7, steeper th a n com pact cores, b u t slightly fla tte r th a n th e spectral index of lobe em ission (B egelm an, B landford a n d Rees 1984).

We have o b tain ed flux-densities for th e radio cores an d h o tsp o ts in th e sources in our sam ple. Section 3.5.2 describes the m eth o d used to isolate th e com pact em ission and presents th e results.


Filtered VLA data.

A ccurate flux-densities of com pact com ponents are difficult to o b ta in from o rd in ary radio m aps because of confusion w ith surro u n d in g exten d ed stru c tu re . In class I radio sources th e cores and h o tsp o ts are contiguous w ith jets w hich are often b rig h t n ear th e core. In class II sources, although th e jets are of low surface b rig h tn ess (or u n d etectab le) th e radio cores are often em bedded in em ission from “b rid g es” betw een th e hotspots. Sim ply m easuring th e flux-density w ith in an area a ro u n d th e radio core m ay include em ission from jets or th e bridge in eith er case. Indeed, in some of th e class II sources th e surface brightness of th e rad io core is m uch lower th a n th e surro u n d in g bridge em ission rendering th e core u n d etectab le w ith o u t fu rth e r im age processing. For th e sam e reason, th e surface brig h tn ess levels in th e cores of th e contour m aps presented in ap p en d ix C should n o t be used to determ ine radio core fluxes.

Inspection of th e uv d a ta in th e form of a grap h of flux-density as a function of sp atial frequency m ay give an idea of th e stren g th of any unresolved com po­ n en ts in the d a ta , b u t this has a num ber of draw backs for q u a n tita tiv e analysis. W ith o u t phase inform ation it is im possible to determ ine w h eth er th e unresolved flux-density originates from a com pact core or from unresolved h o tsp o ts or b o th . Even w ith phase inform ation th e d a ta is difficult to in te rp re t. T h e b est m eth o d to stu d y unresolved com ponents in th e d a ta is to apply a high pass filter to th e 5 GHz self-calibrated uv d a ta and construct m aps in th e usual m an n er. High pass filtering has the effect of suppressing th e extended em ission an d enabling m ore accu rate m easurem ents of th e unresolved core flux-density. M aps of u n ­ resolved em ission generated from th e high pass filter technique are p resen ted in ap p en d ix D. It is unlikely th a t th e radio core flux in th e filtered m ap is c o rru p te d by flux from spatially sep erated h o tsp o ts because phase inform ation (equivalent to positional inform ation on th e sky) is retain ed in th e filtering process. No m ore flux is shifted aro u n d th e sky th a n m ight h ap p en in o rd in ary deconvolution techniques routinely used in radio astronom y.

One m eth o d for applying a high pass filter w ithin A IPS is to ignore all uv d a ta p o in ts from baselines sh o rter th a n a given value (typically 40 kilow avelengths) d u rin g th e m ap m aking process. However, applying such an a b ru p t cutoff in th e fourier dom ain tends to introduce ringing in th e real dom ain which take the form of ripples across the image. F u rth erm o re tossing away a large n u m b er of

uv d a ta po in ts can produce some ra th e r aw kw ard (usually very elongated) beam shapes. R inging in th e filtered m ap is reduced som ew hat by applying a sm oothly varying high-pass filter to the uv d ata. We achieved this by w riting an AIPS ta sk w hich m ultiplies th e visibility weights interior to a given w avenum ber by a cosbell function which tends sm oothly to zero tow ards th e origin of th e uv plane. For th e 5 GHz d a ta , and th e 1.4 GHz A /B d a ta , visibilities w ith w avenum ber less th a n 40 kA were cosbelled. For th e lower resolution 1.4 GHz m u lti-arra y d a ta , the cosbell was applied to d a ta w ith w avenum ber lower th a n 15 kA. T his lower cutoff for th e 1.4 GHz d a ta was necessary because in m ost cases th ere is unsufficient d a ta beyond 40 kA to allow m apping. However, m ost of th e extended s tru c tu re in these rad io sources lies below 15 kA. Spectral indicies for th e radio cores do no t a p p ea r to be unduly affected by th e filter m ism atch. We fit the resu ltin g cores w ith two dim ensional gaussian functions to determ ine th e ir flux-density a n d wdiether they are unresolved. A few of th e 1.4 GHz cores do show some extension and th eir core flux-densities should strictly be regarded as u p p e r lim its w hen com paring th em w ith th e 5 GHz d ata. T he flux-density m easu rem en ts from g au ssian fits are com pared w ith sim ple m easures of th e flux-density w ith in a box a ro u n d th e feature. In all cases th e two flux-densities are in good agreem ent. T ypically th e flux-densities of com pact com ponents m easured in this way are a cc u ra te to w ithin 10%. T h e m easured flux-densities are checked for consistency w ith a m p litu d e d a ta in th e uv plane. In this way we are able to m easure flux- densities of com pact com ponents down to a few millijansky.

We have tested th e validity of th e high-pass filter technique using th e U VMOD ro u tin e w ithin A IPS to take a uv d atab ase from a sn ap sh o t observation an d fill th e d a tab a se w ith a m odel rad io source consisting of two circular g au ssian lobes each w ith full-w idth a t half-m axim um of 10 arcsecs an d flux-density 0.5 Jy. T he lobes are spaced 30 arcsecs a p a rt. Between th e two gaussian lobes we place a re c t­ an g u lar bridge of length 30 arcsecs and w idth 10 arcsecs. T his bridge also has a flux-density of 0.5 Jy. At th e centre of this m odel we place a point source nucleus. A co n to u r m ap of th e cleaned m odel radio source is presen ted in figure 3.5a. T he use of a real uv d a tab a se in the m odel (derived from observations) ensures th a t th e sam pling of th e radio source in the fourier dom ain is realistic. N ote also, th a t a rectan g u lar bridge joining the two radio lobes is a w orst-case in term s of b ack g ro u n d co n tam in atio n of th e radio core flux-density. Figure 3.5b p resen ts a


PLot fll# vtraion 1 created 02-FEB-1991 22:16:06 I POL 4885.100 MHZ MODELICLN.3

Figure 3.5a: Cleaned model radio source comprising an unresolved core with a

In document Optical emission lines in radio sources of intermediate power (Page 71-74)