[ CATS home ] Back to CATS list ] [ ftp ]

1974MmRAS..78....1Readhead & Hewish:

Fine Structure in Radio Sources at 81.5 MHz III.

(OCR by H.Andernach 8/97;)

(Coodinates columns by S.Trushkin 11/98;)

Mem. R. astr. Soc. (1974) 78, 1-49

SUMMARY

A survey of radio sources which exhibit interplanetary scintillations has been carried out with the Cambridge 18000 m^2 array at 81.5 MHz. The methods of observation, and of determination of angular structure on a scale 0.2" to 2.0", are discussed and a catalogue lists the structure of ~1500 4C sources between declinations -12d and +90d, roughly 60 % of which scintillate. An analysis of the results will be presented elsewhere.

I. INTRODUCTION

This paper presents the results of a survey of radio sources carried out at 81.5 MHz in which angular structures in the range 0.2" to 2.0" have been studied using the method of interplanetary scintillation. The observations were made during the period 1969 June to 1971 January with the Cambridge 18000 m^2 array described in Paper I (Hewish & Burnell 1970), and covered the declination range -12d<DEC<+90d. The angular diameters of sources exhibiting scintillation were obtained from the variation of scintillating flux with solar elongation using the theory given in Paper II (Readhead 1971). Some of the initial results for a limited area of sky were described by Burnell (1972), while preliminary data on angular scattering due to the interstellar medium have been discussed by Readhead & Hewish (1972).

The reduction and calibration techniques are described in Section 2, and Section 3 deals with the method of deriving angular diameters and the fraction of the total flux density of a source which scintillates. Section 4 comprises the catalogue of ~1500 4C sources, roughly 60 per cent of which scintillate. It should be emphasized that many of the observed diameters may be due to interstellar scattering rather than to the intrinsic source size.

2. THE OBSERVATIONS AND THEIR REDUCTION

(i) Primary data
During the survey the whole sky in the declination range -12d<DEC<63d was observed once per week, four declination strips being covered simultaneously. I,less frequent observations are needed at higher declinations, since there is a smaller variation of solar elongation, and for DEC>63d observations were repeated at intervals not greater than one month. The range 72d < DEC < 81d was also observed once per week as part of another programs.

The half-power beamwidth of the 18000 m^2 array at declination delta is 0.28d*sec(delta) in right ascension and 4.9d*sec(52d-delta) in declination, whereas the spacing of adjacent declination strips is 2.76deg*sec (52d-delta). Thus there was some redundancy in observations on adjacent declination strips.

The scintillation index, F, is defined by

F = D_S/S

where D_S is the rms variation of flux density due to scintillation and S is the flux density. For each receiver D_S was measured, as described in Paper I, by using a high-pass filter with a low-frequency cut-off at 0.02 Hz to separate the scintillating component from the output. After further amplification the signals were rectified and integrated with a time constant of 15 S, sampled at D_S intervals, and recorded sequentially on a single-punched tape. The sampling interval gave slightly more than 3 sec(delta) readings during the drift-time of a source at declination delta through the reception pattern. The reduction of ~1 per cent in D_S caused by removal of the low-frequency component may be ignored. The non-filtered, filtered and integrated signals were recorded on Evershed three-track chart recorders ( see Fig. I(a), (b), (c) ) to facilitate calibration and also detection of interference.

(ii) Calibration
The survey was calibrated using primary and secondary reference sources. The primary calibrators were the 13 sources for which accurate flux density measurements at 81.5 MHz have been made by Scott & Shakeshaft (1971). The 81.5 MHz survey of Collins (1968) and the 86 MHz survey of Artjukh et al. (1968) provided the secondary calibrators. The flux scale of the survey by Artjukh et al. is in agreement with that of Scott & Shakeshaft after adjustment for the small difference in frequency assuming a spectral index alpha = 0 75. The flux density scale of Collins is systematically lower than that of Scott & Shakeshaft and a correction factor of 22 per cent was applied. For each declination strip there were ~10 secondary calibration sources of sufficient intensity to be relatively unaffected by confusing 4C sources. The receivers employed a system of automatic gain control, thus in order to interpolate between calibration sources the variation of sensitivity with the sky background temperature was calculated using the 178 MHz survey of Tartle & Baldwin (1962) scaled by the factor (81.5/178)^-2.55.

The variations of spectral index with galactic latitude determined by Bridle (1968) indicate that the correction to (2) is negligible over most of the sky and amounts to only ~10 per cent towards the galactic plane. Thus the scaling provided by (2) is sufficiently accurate for our purposes, and has been applied without further corrections, to derive the sensitivity of the telescope (see Fig. 2). Apart from some regions near the galactic plane where the background temperature may change significantly in one beam area, the calibration is believed to be accurate to ~10 per cent.

This calibration was used in conjunction with the output of the integrator circuit (see Fig.I(c)) to determine D_S for each source as follows: Let J be the level at the output of the integrator circuit, and J = <J> if the input consists solely of background noise. In the presence of a scintillating source the increase of the integrator output is then J -<J> which may be converted into D_S using the above calibration and the measured response of the filter, amplifier and integrator circuits.

(iii) The sensitivity of the survey
The data analysis was carried out in two stages. The records were initially edited to remove the effects of occasional interference. Since the four declination strips under observation at any time v.ere chosen to be widely separated, any large signal occurring simultaneously on several receivers was likely to be interference. A computer program was therefore used to eliminate large signals which appeared on at least three receivers at the same time. The chart recordings were also inspected visually as a further check for interference.

The magnitude of <J> in the absence of obvious scintillating sources was next found and used to compute J - <J> for each source. The values of J - <J>. for intense scintillating sources were then transcribed on to magnetic tape and subsequently erased from the integrated record. A typical record after the removal of these is shown in Fig. I(d).

The separate records of J - <J> for each declination strip were then added and their sum convolved with a function corresponding to the response of the array to a source during transit, in order to obtain the maximum signal/noise ratio allowed by information theory. An example of the records at this stage is shown in Fig. 1(e). This analysis gave the maximum integration time for detecting weak scintillating sources and the threshold sensitivity may be estimated as follows:

Let D_S define the scintillating flux of a given source, and D_Sn the system noise for a receiver time constant T1. The random fluctuations at the output of the final integrator, for a time constant T2, will have a value

(J - <J>)_rms = D_Sn*sqrt(T1/T2)

In deriving this result e assume, for convenience, that the output of the final rectifier and integrator is J = D_S, although the detector will be to some extent non-linear. .N scintillating source will be at the limit of detection when

J_(s+n) - J_n ~ (J - <J>)_rms

Since the scintillating flux and the system noise are uncorrelated the mean fluctuating signal is D_S^2 + D_Sn^2. and hence

[ formula ]

giving

[ formula ] The addition of repeated observations, combined with the convolution mentioned above, gives a further increase of integration time and we have [ formula ] as the detection limit, where 7'i s the total integration time. This relation corresponds to a signal noise ratio of unity and we assume that a source is unlikely to escape detection if D_S > 2sigma.

Now at 81.5 MHz D_Sn is largely determined by the galactic background temperature, while D_S depends upon solar elongation and is also a function of the ratio source coordinates. contours indicating D_S = 2*sigma are shown in Fig 3.

(iv) Isolating scintillating sources
After removal of intense scintillating sources, as described in the previous section, all peaks greater than 2*(J - <J>)_rms on the integrated record were noted and added to the list.

The individual values of J - <J> for each repeated observation of every source were converted to scintillating fluxes as described in Section 2(ii) and the variation of D_S with solar elongation for each source was plotted by computer (see Fig. 4).

The scintillating flux curves were then used in conjunction with the chart recordings to distinguish between genuine scintillating sources and random fluctuations on the records. The four highest computed values of DS were checked on the charts to eliminate interference which might have escaped declination in stage one. All suspected scintillators which showed an increase in scintillating flux with decreasing elongation~n were then classified according to the following scheme, which is summarized in Table 1:

[Table 1]

The scintillating sources are assigned to classes A, B or C depending on whether they arc strong, moderate or weak scintillators respectively. Some sources, which showed only a marginal increase in scintillation with decreasing elongation, are classed as probable scintillators. An example of a source in each class is given in Fig. 4. There are - 750 sources in classes A, B and C which arc undoubtedly scintillators and of these - 95 per cent are property associated with 4C sources (sec Section 2(v)). In class P only ~80 per cent are properly associated with 4C sources.

(v) The identification of scintillating sources with sources in the catalogue
The 18000-m^2 telescope is limited by confusion for the weakest 4C sources but this confusion can be reduced by examining adjacent declination strips. Thus a preliminary association1 of scintillators with 4C sources has malls by computer, but the final association by inspection. In the computer program the positions of all scintillating sources were precessed to epoch 1950.0 and checked against those of the 4C sources. 4C source within +- 20s in right ascension and +-3 deg in declination was noted as a possible association. At this stage about 20 per cent of the scintillators coincided, within the stated errors, with more than one 4C source. By comparison of adjacent declination strips most of this confusion could be eliminated though for ~5 per cent of the scintillating sources it was impossible to distinguish between two or three 4C sources within the same beam area. A small number of scintillating sources were not associated with sources in the 4C catalogue. These unidentified scintillating sources have been included in the catalogue (see Section 4).
(vi) The 81.5 MHz fluxes of 4C sources
In order to derive the maximum information from the scintillation observations it is necessary to know the 81.5 MHz flux, S, of the source. The observed value of D_S may then be converted into scintillation index using equation (I), and used to determine the structure of the source as described in Sections 2(vii) and 3. For the brightest sources the value of S was obtained from the unfiltered record (Fig. I(a)), but confusion precluded this for the fainter sources, and therefore an independent estimate of the fluxes was needed. This was derived from the surveys of Collins (1968), Artjukh et al. (1968) and Smith (1968) (see Section 2(ii)). For many of the sources there are no flux measurements at 81.5 MHz. In these cases the 81.5 MHz flux was estimated by interpolating between the 4C flux and the 38 MHz flux ( Williams, Kenderdine & Baldwin, 1966 ) if this was available, or alternatively by extrapolating from the 4C flux on the assumption that alpha = 0.75. In all cases the derived 81.5 MHz flux was adjusted to scale of Scott & Shakeshaft (1971) (see Section 2(ii)).
(vii) The non-scintillating sources
Many of the radio sources observed in the survey exhibited no interplanetary scintillation, indicating that the small-scale components in these sources, if any, contribute only a fraction of the total flux. For these non-scintillating sources we may estimate an upper limit to the flux from a possible small-scale component by comparing the limit on scintillation index (see previous section) its the scintillation index for a point source, as was done by Harris & Hardebeck (1969). The upper limit on the flux in a small-scale component derived by this method is often an underestimate since, due to interstellar scattering, at 81.5 MHz there are no point sources (Readhead & Hewish 1972). In fact the present survey indicates that the typical equivalent Gaussian diameter of a scintillating source is ~ 0.5". Thus a better estimate of the flux which could be in small-scale components, and yet have escaped detection, may be derived by comparing the limits on scintillation index for the non-scintillators with the scintillation index of a 0.5" diameter source. Those sources for which this flux is less than 40 per cent of the total flux have been included in the catalog as non-scintillating sources.

3. DETERMINATION OF ANGULAR STRUCTURE

The angular structure of scintillating sources is obtained by comparing the observed variation, F_o(epsilon), of scintillation index with solar elongation, with theoretical curves, F_th(eps, theta), previously derived in Paper II. If all radio sources consisted of a single scintillating component, the angular diameter, theta, of the source could be obtained simply by fitting the best theoretical cure to the observations. I however, most radio sources consist of more than one component, and thus the theoretical scintillation index curves must be multiplied by some fraction R < 1 before being compared with the observations. In practice the best approach is to find the theoretical curve which has the same slope as the observed curve between two widely separated elongations, as described in Paper II, and hence to derive the equivalent Gaussian diameter, fl, of the source. The fraction R is then given by

[formula]

Calculations by Little & Hewish (1966) and Readhead (1972) of the variation of with e for a number of different source models show that the shape of the F(e) curve is very similar for the different models. In the following discussion we have therefore assumed, for convenience, that all sources consist of a circularly symmetrical Gaussian component, of diameter theta and flux S_1, and a non-scintillating component of diameter > 2" and flux S_2 (see Fig. 5). The effect of this assumption on the interpretation of the observations is discussed later. It should be noted that a Gaussian brightness distribution is a good approximation for sources which are significantly broadened by interstellar scattering.

We now consider separately the strongly scintillating sources, corresponding to classes A and B, and the weakly scintillating sources corresponding to classes C and P:

(i) Sources with D_S >> D_Sn
The diameter, theta, is determined (as described in Paper II) from the ratio [ ] where e_1 is the elongation at which D_S is greatest, and e_2 is some larger elongation, preferable ~90d. The ratio R may then be determined from equation (3) provided that the total flux, S = S_1 + S_2, of the source is known, since

[formula]

In this simple model R is equal to the ratio of the flux in the scintillating component to the total flux of the source, i.e.

[formula]

(ii) Sources with D_S >>DS_n
In cases where DS(e1) is measurable, but the scintillation falls below the detection limit at greater elongation c~ the ratio DS(e1)/DS(e2) no longer gives a reliable estimate of theta, but it does provide an upper limit, theta_u, to the angular diameter, and in this case a corresponding upper limit to R is given by equation (3), i.e. we have

[formula]

In some cases, where theta, is so large that the upper limit to R given by equation (4) is > 1, and therefore useless, we may derive a more stringent upper limit, theta_u', to the angular diameter by setting

[formula]

A lower limit to R may be obtained by comparing the observed scintillation index with that of a point source,

[formula]

Thus for sources with D_S ~= D_Sn it is possible to derive an upper limit to the angular diameter, theta by two independent methods. In addition a useful lower limit to R can be derived in all cases, while only in certain cases can a useful upper limit to R be derived.

Since it is known that many extragalactic radio sources contain two or more components of comparable flux density and angular Size', it is important to estimate how the complexity of a source might affect the values of R and theta derived from our simple model. Consider two possibilities, sketched in Fig.6, such that the flux S1 is contained in (i) N components of diameter theta separated by angle S( > 1") which exceeds the resolution of the scintillation method, and (ii) N components of negligible diameter contained within an angle < I". Elementary considerations then show that for model (i) the correct value of ~ is obtained but R is reduced by a factor 1/sqrt(N), while for model (ii) R is underestimated by some factor and the apparent diameter is that corresponding to the group (Little & Hewish 1968).

These examples show that the values of R derived on the assumption of our two-component model will tend to underestimate the total flux contained in fine structure, while fine structure contained within an angle of about 1" will be regarded as a single extended source.

4. THE CATALOGUE

The sources in the catalogue have been arranged in order of right ascension. The columns are as follows:

(i) The right ascension (in hours and minutes) and declination (in degrees) at epoch 1950.0.

(ii) The 4C number of the source. Sources which are not associated with sources in the 4C catalogue are denoted by U. The 3C number of the source is also given. In cases where a scintillating source is associated with a source which is not in the 3C or 4C catalogue the finding survey is denoted as follows: WKB, refers to the 38 MHz survey of Williams, Kenderdine & Baldwin; MSH, refers to the 85 MHz survey of Mills, Slee & Hill; PKS, refers to the 408 MHz Parkes survey.

(iii) The equivalent Gaussian diameter, theta, in seconds of arc. Non-scintillating sources are denoted by NS.

(iv) The fraction R of the flux density in the scintillating component". In the case of non-scintillating sources this is the upper limit on the fraction with structure < 0.5".

(v) The total flux, S, followed by a letter indicating the source of the 81.5 MHz flux: A, Artjukh et al. ; C, Collins; E, Extrapolated from 178 MHz assuming alpha = 0.75; I, Interpolated between 38 MHz and 8 MHz; P, Parker (1968); S, Smith; SS, Scott & Shakeshaft.

(vi) The maximum observed scintillating flux, in flux units (10^26 W m^-2 Hz^-1). (Jy)

(vii) The elongation, e_1, at which maximum scintillation occurs.

(viii) The source classification, see table 1.

(ix) Other 4C sources in the same beam area an asterisk indicates that the confusing source also scintillates.

REFERENCES

Artjukh, V. S., Vitkevitch, V. V., Dagkesamansky, R. D. & Kozhukov, V. N., 1968. Astr. Zh. 45, 712.

Bridle, A. H., 1968, MNRAS 138, 251.

Burnell, S. J., 1972, A&A 16, 379.

Collins, R. A., 1968. Ph.D. Thesis, University of Cambridge.

Harris, D. L. & Hardebeck, E. G., 1969. ApJ 170, 115.

Hewish, A. & Burnell, S. J., 1970. MNRAS 150, 141.

Little, L. T. & Hewish, A., 1966, MNRAS 134, 221.

Little, L. T. & Hewish, A., 1968, MNRAS 138, 393.

Parker, E. A., 1968, MNRAS 138, 407.

Readhead, A. C. S., 1977, MNRAS 155, 785.

Readhead, A. C. S., 1972, Ph.D. Thesis, University of Cambridge.

Readhead, A. C. S., Hewish, A., 1972, Nature (London) 236, 440.

Scott, P. F. & Shakeshaft, J. R., 1971. MNRAS 155, 19P.

Smith, M. A., 1968, Ph.D. Thesis, University of Cambridge.

Turtle, A. J. S: Baldwin, J. E., 1962. MNRAS 124, 459.

Williams. P. J. S., Kenderdine, S. S., Baldwin, J. E., 1966. MmRAS 70, 53.