2 Azimuth dependence of the ground contamination level

Stray radiation due to ground emission in the GEM experiment was initially recognized to attain hazardous levels during test operations at the Brazilian site (W $44\ifmmode^\circ\else\hbox{$^\circ$ }\fi59\hbox{$^\prime$ }55\hbox{$^{\prime\prime}$ }- {\rm S} 22\ifmmode^\circ\else\hbox{$^\circ$ }\fi41\hbox{$^\prime$ }$) when only the rim-halo protection had been installed. Figure 2 shows a sky map from a sample batch containing 123.92 hours of data taken during this testing period at 1465MHz, where the horizontal striping shows clear evidence of a variable component of sidelobe contamination due to ground emission for the zenith-centered circularly scanning motion of the antenna. The map was prepared according to the same data reduction process that will be outlined in Sect. 5.1 and included custom cuts of $60\ifmmode^\circ\else\hbox{$^\circ$ }\fi$ from axis for the Sun, of $6\ifmmode^\circ\else\hbox{$^\circ$ }\fi$ for the Moon and eventual excision of RFI signals. The relative calibration of the map was, however, not subjected to an adopted baseline subtraction technique which filters out low frequency noise. Instead, we assumed that any continuous set of data between successive firings of the calibrating noise source diode (comprising about 70% of a full scan or, equivalently, 35% of the angular extent of a great circle in the sky) would contain, at least, one pointing direction towards which the sky would appear uniformly cold across the entire declination band being mapped. This assumption is realistically incorrect, but as Fig. 3 shows, it is nevertheless useful to portray a reasonable outline of Galactic features albeit an unnaturally flattened temperature distribution and some residual stray radiation of Solar and artificial origin. The latter was absent during the test runs only to emerge later with a 100% duty cycle and in the direction of a near urban area.

Figure 2: A sky map at 1465MHz of a $60\ifmmode^\circ\else\hbox{$^\circ$ }\fi$ declination band obtained with the GEM experiment in the Southern Hemisphere using only a rim-halo protection and assuming an uniform baseline level across the sky (Epoch $2\,000.0$ coordinates). The pixelization is $1.6\ifmmode^\circ\else\hbox{$^\circ$ }\fi$and the antenna temperature range extends 1.5þK above the lowest temperature in 12 equally-spaced contours in order to enhance the distribution of Galactic radiation away from the Galactic Plane

Figure 3: A GEM map at 1465MHz of the same sky region as that in Fig. 2 (same baseline assumption and pixel size), but using data obtained after the wire mesh screen had been added to the shielding configuration of the antenna. The antenna temperature ranges also over 1.5þK above the lowest temperature, but the contour levels are spaced more tightly (60þmK). Marked locations denote the sky directions of the 6 test measurements discussed in Sect. 5

Figure 3 is a map of the same declination band as that of Fig. 2 after a fence of wire mesh had been built around the rotating antenna. A total of 222.57 hours of data from an optimally-stable receiver were used in the preparation of this map. The azimuth-dependent contamination from the ground has been largely removed and we can estimate its level by subtracting representative azimuth scans from the two maps as described below. No absolute calibration of the baseline was attempted for either of these maps, as it is not relevant for determining differential measurements. This approach will enable us to refine the model used in Sect. 5 for predicting a best estimate of the level of the azimuth-independent component of ground contamination in the survey. The locations marked in the map of Fig. 3 correspond to the chosen set of sky directions, grouped pair-wise, for obtaining the differential sky measurements. They avoid the proximity of the Galactic Plane in order to diminish the chance of scale error corrections.

Figure 4: Antenna temperature profiles obtained before and after the construction of the fence and sampled along the scanning direction in regions of relatively low sky contrast

The variable or azimuth-dependent component of ground contamination can be estimated by adequate comparison of the azimuth antenna temperature profiles before and after the introduction of the fence. Figure 4 shows two such sets of profiles. They were obtained from single time-ordered series of scans covering the same regions of the sky and they sample the sky in 122 alt-azimuth circular bins spanning approximately half-a-beamwidth across. This binning criteria is a basic precept for the relative calibration of the survey and it will not be discussed further in the present context. A full treatment of this calibration technique can be found in Tello (1997) and will be included in the publication of the survey. At this point we just mention that the series of scans were chosen for complying with highly stable receiver performance and relatively high Galactic latitude. This combination favours sky profiles of low emission contrast for easier identification of the ground contribution to the antenna temperature. The circular arrangement of the sampled bins has been schematically superimposed against the observed sky in Figs. 5a,b. Thus the difference between the antenna temperature profiles in Fig. 4 is a good approximation (see Fig. 6) of the ground contamination in the absence of the fence. It can be seen to be made up of a variable component with a mean amplitude of $0.52\pm0.29$þK above the level of a uniform azimuth-independent component. The two components result from the convolution of the antenna beam pattern over the ground temperature distribution, whose spatial extent in the vertical direction is limited by the line of the horizon also depicted in Fig. 6.

Figure 5: Maps of the sky regions revealed by the circularly scanning technique of the GEM experiment at 1465MHz, a) before and b) after the construction of the fence, and the distribution of the alt-azimuth sky bins used in the sampling of the signal displayed in the antenna temperature profiles of Figs. 4a,b. Map a) consumed 2.29þhours of observational time and Map b) 2.18þhours. Both maps are given at a $1.6\ifmmode^\circ\else\hbox{$^\circ$ }\fi$ pixel-resolution and their antenna temperatures range in 12 contour steps of 60 mK above their respective minimum values

Figure 6: The level of ground contamination in the absence of the fence deduced from the difference in antenna temperature between the profiles of Fig. 4. The dotted line depicts the horizon profile at the observational site

Figure 7: Diagram representations of the backfire power patterns at a) 408MHz and b) 1465MHz in a coordinate reference frame centered on the transmitter. Three concentric circles have been superimposed on the diagrams to illustrate the opening angles of the ground shields: the boundary of the dish itself at $\theta =78.9^\circ$, the rim-halo at $\theta=99.2\ifmmode^\circ\else\hbox{$^\circ$ }\fi$ and the farthest lying location on the upper edge of the fence at $\theta=142.9\ifmmode^\circ\else\hbox{$^\circ$ }\fi$when $Z=45^\circ$. The 10-dB level at 1465 MHz is mostly contained inside the elliptical contour. Arrows indicate the $\phi^\ast(\equiv 180\ifmmode^\circ\else\hbox{$^\circ$ }\fi-\phi)$ orientations of the feed for generating the upper (upp) and lower (low) envelopes in Figs. 10-13. The subindex number denotes the number of shields accounted for (1: only the rim-halo, 2: both halo and fence)

In the presence of the fence, we can estimate the azimuth-independent component of ground contamination by convolving the antenna beam pattern with an uniform field of radiation confined to the solid angle that the fence fills in at the prime focus of the parabolic dish. In this case, the beam pattern is the modified feed response which due to the presence of the shields gives rise to spillover and diffraction sidelobes. This is the subject we deal with in the next two sections before we assess the reality of the observations.