THE ASTROPHYSICAL JOURNAL SUPPLEMENT SERIEs, 54:211-227, 1984 February (c) 1984. The American Astronomical Society. All rights reserved. Printed in U.S.A. (OCR@ADS 12/03, proof by S. Trushkin 01/2004)
5 GHz SOURCE VARIABILITY AND THE GAIN OF THE NRAO 300 FOOT TELESCOPE C.L. BENNETT, C.R. LAWRENCE, B.F. BURKE Research Laboratory of Electronics, Massachusetts Institute of Technology Received 1983 June 23; accepted 1983 August 10 ABSTRACT During the course of the MIT-Green Bank 5 GHz survey on the NRAO 300 foot (91.4 m) transit telescope, ~2400 observations were made of strong sources for the purpose of calibration. In this paper we analyze the gain curve of the 300 foot telescope between -10{deg} and +30{deg}, including its time stability, and discuss the variability of the 123 calibrator sources that were observed. Subject headings: instruments - radio sources: variable 1. INTRODUCTION The MIT-Green Bank survey of 2.2 sr of sky between -O.5{deg} and +19.5{deg}, begun in 1979 using the NRAO (1) 300 foot (91.4 m) transit telescope, is now nearly finished. During the course of the survey repeated observations were made of 123 sources, ~1 Jy, for calibration purposes. These observations were used to determine a pointing correction and to calibrate the gain of the 300 foot telescope. The calibrator observations over a period of several years also allow a study of source variability. In this paper we present first an analysis of the gain curves of the 300 foot telescope between -10{deg} and +30{deg}, and then ~2400 measurements of the fluxes of 123 calibrators. (1) The National Radio Astronomy Observatory is operated by Associated Universities, Inc., under contract with the National Science Foundation. 2. OBSERVATIONS Sources from Table 1 of Lawrence et al. (1983) were observed in 1979 July, 1980 April-May, 1980 June-July, 1981 January-February, 1981 August, 1981 September-October, 1981 December-1982 February, and 1983 February. All of the observations were made with the beam-switched "6-25 receiver" in the "Sterling mount." The two feeds are separated by 7.6' on the sky, and the nominal beamwidth (FWHP) of each is 2.8'. The center frequency was 4.775 GHz with a bandwidth of 580 MHz, and the integration time was 0.5s. The telescope was driven in declination from south to north at 6 times the sidereal rate, 90"/s. The feeds were rotated to an angle of -12{deg} (in the sense N through E on the sky), so that a source passing through the midpoint between the two feeds goes through the half-power point of first one feed, then the other. 3. ANALYSIS Calibrator scans were analyzed by cross-correlation with a model beam as described by Lawrence et a!. (1983). If either of the local minima of the cross-correlation did not fall within one integration period of the expected minimum position, based on the feed model, then the scan was thrown out. This occurred only rarely. The antenna temperature was calculated from the values at the two minima. The sources used to calibrate the gain of the 300 foot telescope are most of those in Table 1 of Lawrence et al. (1983) for which fluxes are listed. Antenna temperatures from the cross-correlations were divided by the fluxes from the table to give gain values. Several models were fitted to the gain values as a function of declination. The best fit was obtained with the following polynomial: G(K/Jy) = G0[1+G2/G0(dec-dec0)^2 + G4/G0(dec-dec0)^4 , (1) where dec0 = 38.43deg is the declination of the zenith in Green Bank. Although this model gives a gain curve which is symmetric about the zenith, we have made observations only in the declination range -10{deg} < DEC < +30{deg} and cannot comment on the accuracy of the model outside of this range. The fit was made by reflecting the gain data points about the zenith declination, followed by a change of variables to x_i = (dec- dec0)/48.43, so that x(dec_i = -10{deg}) = -1. Since -1 < x_i < 1, Legendre polynomials could be fitted to the data. Legendre polynomials are approximately orthogonal over the x_i's, and their coefficients are nearly independent. This process automatically forces the gain curve to be symmetric about dec0, since the odd Legendre polynomials will vanish. The Legendre coefficients were then converted into the G`s of equation (1). The results are presented in Table 1, and their time variations are shown in Figure 1. Figure 2 is representative of the high quality of all fits. The values of G0 are significantly different for different times. The variations in G2/G0 =(-5.27+-O.37)x10^-4 and in G4/G0=(7.7+-1.9)x10^-8 are insignificant. Thus, the gain curve used in obtaining the fluxes in this paper is G(K/Jy)=G0(t)[1- 5.27x10^-4(dec-38.43deg)^2+ +7.7x10^-8(dec -38.43deg)^4], (2) where G0(t) is given in Table 1. Table 1. Gain curve coefficients as a function of time Date G_0 G2/G_0 G4/G_0 x10^-4 x10^-8 ----+----1----+----2----+----3----+ 79.504 0.936 -4.86 5.91 80.375 1.06 -5.50 8.52 80.542 0.967 -5.22 7.86 81.125 0.943 -5.08 6.46 81.666 0.867 -5.25 7.64 81.792 0.872 -5.30 8.20 82.000 0.876 -5.03 6.36 82.375 0.855 -6.13 12.1 83.166 0.861 -5.06 6.04 ------------------------------------- NOTE. Differences in G0 are significant. Higher order coefficients are taken as constants with values G2/G0 = -(5.27+-0.37)x10^{-4} and G4/G0 = (7.7+-1.9)x10^{-8}. The important result is that G_0 changes significantly over time, but G2/G_0 and G4/G_0 do not. In other words, the shape of the gain curve is constant, but its amplitude changes. The antenna temperatures of 2400 observations were converted into fluxes using equation (2). An editing algorithm eliminated measurements which were obviously in error, as follows: the mean flux, (S), and the standard deviation of the mean, sigma_m were calculated for each source. The most extreme flux, S_E was thrown out if |S_E-| > 3max({sigma}_m,O.2), (3) where max(x,y) chooses the larger value of the two arguments The quantitiesand {sigma}_m were recomputed, and the procedure iterated until all the measurements passed the test. Only ~25 observations (~1%) were thrown out, so this is a mild editing algorithm. The resulting fluxes and the corresponding dates of observation are shown in Table 2. The uncertainty in a single measurement is declination dependent, but the measured dispersion for - 1O; the standard deviation, sigma; the number of measurements, N; the coefficients A and B as described above; the percentage deviation in the measurements 100sigma_S/